CN103413054A - Internet addiction detection device and method based on user-computer interactive events - Google Patents

Abstract

The invention provides an Internet addiction detection device and method based on user-computer interactive events and belongs to the field of data mining. According to the Internet addiction detection device and method based on the user-computer interactive events, quantifiable human-computer interactive operational data are collected through Internet surfing tools which are used by a user frequently, the data are used for computing and analyzing Internet surfing behaviors of the user, accordingly whether the user has an Internet addiction can be detected, and the Internet surfing tools can be controlled effectively. According to the Internet addiction detection device and method based on the user-computer interactive events, the accuracy of Internet addiction detection can reach more than 85%, errors of existing detection methods can be effectively avoided, and the accuracy of the detection can be improved; the detection costs can be reduced, a user can carry out detection at any time, high application value to primary and secondary school students can be obtained, Internet addiction behaviors can be effectively prevented and controlled, and damage caused by Internet addiction can be reduced.

Description

Internet addiction detection device and method based on user computer interaction events

technical field

The invention belongs to the field of data mining, and in particular relates to an Internet addiction detection device and method based on user computer interaction events.

Background technique

The Internet has become an important part of modern people's lives, especially with the rise of social networks such as Facebook, Twitter, and Sina Weibo, more and more people have joined the ranks of surfing the Internet. There is no doubt that while the network brings efficient and convenient services to human society, it also brings negative and negative impacts to human society. Due to unrestrained and excessive indulging in the Internet, some Internet users have suffered physical, psychological, emotional, and moral damages, interfered with normal work, study, and life, and even induced crimes.

By consulting relevant information on Internet addiction, most of the current research work and achievements are based on psychology, sociology, medicine and other fields. Detection is blank. In fact, the computer operations that users interact with the Internet can be recorded in detail and used for analysis. This is of great value in judging Internet addiction and even discovering potential Internet addiction trends, but it is currently ignored.

Contents of the invention

Aiming at the shortcomings of the prior art, the present invention proposes a device and method for detecting Internet addiction based on user computer interaction events, so as to reduce detection cost, improve detection accuracy, and simplify detection methods.

An Internet addiction detection device based on user computer interaction events, including a simple event collection module, a complex event generation module, a time generalization and normalization processing module, a frequent behavior generation sub-acquisition module, a sample library construction module, data processing and detection module and early warning intervention module, in which,

Simple event collection module: used to collect user-computer interaction variables as simple events, including computer CPU usage rate, memory usage rate, number of clicks of the left mouse button per unit time, number of times of right mouse clicks per unit time, number of keyboard strokes per unit time, unit The number of pixels moved by the time mouse, network traffic and running process, and send the above input variables to the complex event generation module;

Complex event generation module: used to compare each simple event with its respective threshold, if it is greater than or equal to the threshold range, then keep the simple event, and then generate complex events according to the combination of all the retained simple events; otherwise, continue to collect simple events ; The complex events include watching videos offline, watching videos online, real-time strategy games, board games, browsing web pages, online chatting and downloading operations;

Time generalization and normalization processing module: used to sort out the complex events that each user does on a certain day, and generalize the occurrence time of complex events, that is, to divide a day into several time periods; Normalization processing, that is, rounding the duration of complex events; and sending the processed complex events to the frequent behavior generation sub-acquisition module;

Frequent behavior generation sub-acquisition module: when establishing a sample library, it is used to obtain the same complex events between multiple Internet addicted users, collect the generation times of the above-mentioned same complex events, and select out according to the number threshold set by the user The complex events greater than the threshold are regarded as frequent Internet addiction complex events, and then a set composed of the above-mentioned frequent Internet addiction complex events is obtained to describe the generator of the distribution of user behavior characteristics, and sent to the sample library construction module; used to obtain multiple non- For the same complex events between Internet addicted users, collect the occurrence times of the above-mentioned same complex events, and select the complex events greater than the threshold as frequent non-Internet addiction complex events according to the number threshold set by the user, and then obtain the above frequent non-Internet addiction complex events. A generator for describing the distribution of user behavior characteristics, which is a collection of non-Internet addiction complex events, and sent to the sample library construction module;

When detecting the user's Internet addiction status, it is used to obtain the same complex events performed by the user to be tested within a period of time, collect the number of occurrences of the above-mentioned same complex events, and select complex events greater than the threshold according to the number of times threshold set by the user As a frequent complex event, obtain a generator for describing the distribution of behavioral characteristics composed of the above frequent complex events, and send it to the data processing and detection module;

Sample library construction module: used to calculate the score of frequent Internet addiction complex event generators using the generator-based sample construction algorithm. If the score exceeds the threshold, it will be saved as an Internet addiction category, otherwise it will be deleted, and the frequent non-Internet addiction complex event The generators are saved as non-Internet addiction categories; or the generators of frequent and complex events are calculated using the sample construction algorithm based on the exposure patterns, and the generators of the exposure patterns are obtained and saved;

Data processing and detection module: used to compare the frequent behavior generators to be detected with the Internet addiction sample library, if the frequent complex event generators to be detected are included in the sample library, it is determined to be Internet addiction, otherwise it is not Internet addiction, and Output the judgment result; or compare the frequent behavior generators to be detected with the sample library, use the Internet addiction detection algorithm based on the exposure mode to calculate their Internet addiction scores and non-Internet addiction scores, and judge their category according to the score;

Early warning intervention module: when it is detected that the user's computer interaction behavior is an Internet addiction behavior, an early warning prompt is issued, and the time range for the user to continue using the computer is limited by a timer, 1 to 2 hours; if it exceeds the time limit, it will be closed computer.

The method for detecting an Internet addiction detection device based on user computer interaction events comprises the following steps:

Step 1. Collect the interaction variables between multiple Internet addicted users and multiple non-Internet addicted users and the computer as simple events, including computer CPU usage, memory usage, the number of times the left mouse button is clicked per unit time, and the right mouse button clicked per unit time Number of times, number of keyboard strokes per unit time, number of pixels moved by the mouse per unit time, network traffic and running process;

Step 2. Compare each simple event with its respective threshold. If it is greater than or equal to the threshold range, then keep the simple event, and then generate a complex event according to the combination of all the retained simple events; otherwise, continue to collect simple events; Complex events include watching videos offline, watching videos online, real-time strategy games, board games, browsing the web, chatting online, and downloading operations;

The process of generating a complex event is as follows:

If the CPU usage exceeds or equals to the threshold range of 45% to 55%, the memory usage exceeds or equals to the threshold range of 55% to 65%, and the Internet traffic exceeds or equals to the threshold range of 35 to 45MB, a complex event of watching videos online is generated. ;

If the CPU usage exceeds or equals to the threshold range of 45% to 55%, the memory usage exceeds or equals to the threshold range of 55% to 65%, and the running process is a multimedia player, a complex event of watching videos offline is generated;

If the CPU usage exceeds or is equal to the threshold range of 45% to 55%, the memory usage exceeds or equals to the threshold range of 55% to 65%, and the number of clicks of the left mouse button per unit time exceeds or equals to the threshold range 25 to 35 times, click the mouse per unit time The number of right clicks exceeds or equals the threshold range of 10 to 20 times, the number of pixels moved by the mouse per unit time exceeds or equals the threshold range of 1550 to 1650, the number of keyboard strokes per unit time exceeds or equals the threshold range of 80 to 120 times, and the running process is For real-time strategy games, generate a complex event for playing real-time strategy games;

If the CPU usage exceeds or equals to the threshold range of 45% to 55%, the memory usage exceeds or equals to the threshold range of 55% to 65%, and the number of clicks of the left mouse button per unit time exceeds or equals to the threshold range of 25 to 35 times, the mouse moves per unit time. If the number of pixels exceeds or is equal to the threshold range of 1550 to 1650, and the running process is a board game, a complex event of playing a board game is generated;

If the CPU usage exceeds or equals to the threshold range of 45% to 55%, the number of clicks of the left mouse button per unit time exceeds or equals to the threshold range of 25 to 35 times, and the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550 to 1650, And if the running process is a board game, another complex event of playing a board game is generated;

If the number of clicks on the left mouse button per unit time exceeds or equals to the threshold range of 25 to 35 times, the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550 to 1650, and the running process is a browser, a complex page browsing process is generated. event;

If the number of clicks on the left mouse button per unit time exceeds or equals to the threshold range of 25 to 35 times, the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550 to 1650, the Internet traffic exceeds or equals to the threshold range of 35 to 45MB, and the running process For the browser, another complex event for browsing the web page is generated;

If the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550 to 1650, the Internet traffic exceeds or equals to the threshold range of 35 to 45MB, and the running process is a browser, another complex event of web browsing is generated;

If the number of keystrokes per unit time exceeds or is equal to the threshold range of 80 to 120 times, and the running process is instant chat software, a complex event of online chat will be generated;

If the number of pixels moved by the mouse per unit time exceeds or equals the threshold range of 1550 to 1650, the number of keystrokes per unit time exceeds or equals the threshold range of 80 to 120 times, and the running process is instant chat software, another online chat will be generated complex events;

If the Internet traffic exceeds or is equal to the threshold range of 35-45MB, a complex event of downloading data will be generated;

Step 3. Generalize the occurrence time of complex events, that is, divide a day into 6:00～11:00, 11:00～14:00, 14:00～18:00, 18:00～23:00, 23 :00～6:00 five time periods, and the duration is normalized, that is, the duration of complex events is rounded up in units of 10 minutes;

Step 4. Acquire the same complex events between multiple Internet addicted users and the same complex events between multiple non-Internet addicted users, collect the number of occurrences of the above-mentioned same complex events, and set the threshold according to the number of times set by the user , select complex events greater than the threshold as frequent Internet addiction complex events and frequent non-Internet addiction complex events, and then obtain multiple generators for describing the distribution of user behavior characteristics composed of the above-mentioned frequent Internet addiction complex events, and obtain by Multiple generators used to describe the distribution of user behavior characteristics formed by the above-mentioned frequent non-internet addiction complex events are sent to the sample library construction module;

Step 5. Use the generator-based sample construction algorithm to calculate the score for the generators of frequent Internet addiction complex events. If the score exceeds the threshold, save it as an Internet addiction category, otherwise delete it, and save the generators of frequent non-Internet addiction complex events. Save it as a non-Internet addiction category; or use the sample collection algorithm based on the exposure pattern to calculate the generators of frequent and complex events, obtain the generators of the exposure pattern and save them;

Step 6, collect the interaction variables between the user to be tested and the computer within a period of time, and repeat steps 2 to 3;

Step 7. Obtain the same complex events performed by the user to be tested within a period of time, and collect the occurrence times of the above-mentioned same complex events, and select complex events greater than the threshold as frequent complex events according to the number threshold set by the user, and then Obtain a generator for describing the distribution of behavioral characteristics composed of the above-mentioned frequent and complex events, and send it to the data processing and detection module;

Step 8, comparing the frequent behavior generator to be detected with the Internet addiction sample library, if the frequent complex event generator to be detected is included in the sample library, then determine it is Internet addiction, otherwise it is not Internet addiction, and output the judgment result; or Compare the frequent behavior generators to be detected with the sample library, and use the Internet addiction detection algorithm based on the exposure model to calculate their Internet addiction scores and non-Internet addiction scores, and judge the scores of the two. The frequent and complex events to be detected belong to the category with a higher score ;

Step 9. If it is detected that the user's computer interaction behavior is an Internet addiction behavior, the early warning intervention module is used to issue an early warning prompt, and the time range within which the user can continue to use the computer is limited by a timer, 1 to 2 hours; if the time limit is exceeded, then Shut down the computer.

The respective thresholds of the simple events described in step 2: the computer CPU usage ranges from 45% to 55%, the memory usage ranges from 55% to 65%, and the number of clicks on the left mouse button per unit time ranges from 25 to 35 times , The value range of the number of clicks on the right mouse button per unit time is 10-20 times, the value range of the number of keyboard strokes per unit time is 80-120, the value range of the number of pixels moved by the mouse per unit time is 1550-1650, and the value range of network traffic is 35- 45MB.

In Step 5, the generator-based sample construction algorithm is used to calculate the score of frequent Internet addiction complex event generators, specifically:

Step 5-1. Determine the weight coefficient of complex event types according to user needs, and ensure that the weight coefficient of real-time strategy games > the weight coefficient of watching videos online > the weight coefficient of board games > the weight coefficient of browsing webpages > the weight of watching videos offline Coefficient > weight coefficient of online chat > weight coefficient of download, and the total weight coefficient is 1;

Step 5-2. Determine the weight coefficient of the time generalization segment of complex events according to user needs, and ensure that the weight coefficient of dawn > the weight coefficient of evening > the weight coefficient of noon > the weight coefficient of afternoon > the weight coefficient of morning, and the total number of weight coefficients is 1;

Step 5-3: Multiply the weight coefficient of each complex event type in the generator, the weight coefficient of the complex event time generalization section and the duration to obtain the score, and sum the scores of each complex event to obtain the score of the generator ;

Step 5-4. If the score exceeds the threshold, and the threshold ranges from 1 to 6, save it, otherwise delete it.

In step 5, the sample construction algorithm based on the exposure pattern is used to calculate the generators of frequent complex events, and the generators of the exposure patterns are obtained and stored in categories, specifically:

Step 5-a, obtaining the generators that belong to the frequent Internet addiction complex event but not the frequent non-Internet addiction complex event;

Step 5-b. Obtain the generators that do not belong to the frequent Internet addiction complex event but belong to the frequent non-Internet addiction complex event;

Step 5-c, obtain the co-producers that belong to both frequent Internet addiction complex events and frequent non-Internet addiction complex events, and determine their support degree in the frequent Internet addiction complex events, that is, the occurrence rate. The degree of support in complex events of Internet addiction;

Step 5-d, calculate the growth rate of the co-producer in the frequent Internet addiction complex event, that is, divide its support degree in the frequent Internet addiction complex event by the support degree in the frequent non-Internet addiction complex event; The growth rate of the co-producer in the frequent non-Internet addiction complex event is to divide its support degree in the frequent non-Internet addiction complex event by the support degree in the frequent Internet addiction complex event;

Step 5-e, compare the size of the above two growth rates, retain the generator whose growth rate is greater than 1, and delete the generator whose growth rate is less than 1;

Step 5-f: Union the generators of the frequent Internet addiction complex events, and union the generators of the frequent non-Internet addiction complex events, that is, obtain and save the generators of the exposure pattern.

The Internet addiction detection algorithm based on the exposure mode described in step 8 is used to calculate its Internet addiction score and non-Internet addiction score respectively, and the score score (C, D _i ) formula is as follows:

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; RS</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\frac{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Max</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; RS</span>}}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; Sup</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="WS"\; filenames="HDA0000368918860000041.png"\; state="\{\{state\}\}">WS</figure-callout></span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Min</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; WS</span>}}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; Sup</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Among them: D _i|j represents D _i sample library or D _j sample library, D _j|i represents D _j sample library or D _i sample library, i and j are the class label numbers of sample library, X represents the exposure mode of D _i Generator, Y represents the generator of the exposure pattern of D _j ; RS is the positive sample set of the target class, that is, the set that plays an active role; WS is the reverse sample set of the target class, that is, the set that plays a negative role; Gr( X,D _j|i ,D _i|j ) represents the growth rate of X from one class to another class; MaxGr _RS represents the maximum growth rate threshold in the target class set, RS represents the positive sample set of the target class, MinGr _WS Represents the minimum growth rate threshold in the non-target class set, WS represents the reverse sample set of the target class, C is the complex event to be classified, Sup _c (X) represents the support of X in C, Sup _c (Y) represents Y Support in C.

Advantages of the present invention:

A device and method for detecting Internet addiction based on user computer interaction events in the present invention collects quantifiable human-computer interaction operation data through commonly used Internet tools (such as desktop computers and notebook computers), and uses these data to calculate and analyze user Internet behavior, so as to detect whether the user suffers from Internet addiction, and effectively control the Internet tool; the correct rate of detecting Internet addiction in this patent can be as high as 85%, which effectively avoids the mistakes of existing detection methods and improves the accuracy of detection The invention can also reduce the detection cost, and the user can carry out the detection at any time. It has high application value for primary and middle school students, effectively prevents and controls Internet addiction behavior, and reduces Internet addiction damage.

Description of drawings

Fig. 1 is a device structure diagram of an embodiment of the present invention;

Fig. 2 is the flow chart of the method for detection by the Internet addiction detection device of an embodiment of the present invention;

Fig. 3 is a schematic diagram of an equivalence class and a generator of an embodiment of the present invention;

Fig. 4 is an efficiency test result graph of an embodiment of the present invention, wherein, graph (a) is a running time coordinate graph, and graph (b) is a memory space coordinate graph;

Fig. 5 is a diagram of the effectiveness test results of an embodiment of the present invention, wherein, diagram (a) is a coordinate diagram of correct rate, diagram (b) is a coordinate diagram of missed diagnosis rate, and diagram (c) is a coordinate diagram of misdiagnosis rate.

Detailed ways

An Internet addiction detection device based on user computer interaction events, as shown in Figure 1, includes a simple event collection module, a complex event generation module, a time generalization and normalization processing module, a frequent behavior generation sub-acquisition module, and a sample library construction module, data processing and detection module and early warning intervention module, among which,

Simple event collection module: used to collect user-computer interaction variables as simple events, including computer CPU usage rate, memory usage rate, number of clicks of the left mouse button per unit time, number of times of right mouse clicks per unit time, number of keyboard strokes per unit time, unit The number of pixels moved by the time mouse, network traffic and running process, and send the above input variables to the complex event generation module;

in,

The CPU is the core hardware component of the computer. When the user runs different applications, the CPU usage rate changes in real time. Correspondingly, the user's different computer operation behaviors make the CPU usage rate correspond to different value ranges. At the same time, this indicator can also be used to further Infers the type of user complex events. In the embodiment of the present invention, the current CPU usage is monitored through the task manager in the computer.

The monitoring of memory usage is similar to the monitoring of CPU usage, both of which can be used to judge the behavior of users operating computers. Similarly, this indicator should also be used to infer complex event types of user behavior. In the embodiment of the present invention, the current memory usage is monitored through the task manager in the computer.

The number of clicks of the left mouse button per unit time. The data difference of different computer operation behaviors on this indicator is very obvious. For example, when the user performs a large-scale game operation, the number of mouse clicks per unit time is quite large. Hundreds of times per minute. In contrast, when watching movies, videos, etc., there are only a few times per minute, so this indicator can clearly distinguish the user's high-level semantic behavior. In the embodiment of the present invention, the number of mouse clicks per unit time is obtained by monitoring mouse click events in real time.

The record of the number of keyboard strokes per unit time is basically the same as the record of the number of mouse clicks, and it can be significantly expressed for a large number of behaviors that require keyboard operations. This indicator is also one of the important indicators for reasoning about complex events with differences. In the embodiment of the present invention, the number of keyboard strokes per unit time is obtained by monitoring keyboard stroke events in real time.

The network traffic (including uplink traffic and downlink traffic) after the computer is connected to the network can effectively distinguish the online behavior from the local machine behavior. When a user frequently uses the network, the uplink or downlink traffic data is much higher than the local behavior, which is of great significance for judging whether the user is going online. In the embodiment of the present invention, the network flow of the computer is acquired by invoking the statistical information of the Ethernet in the command prompt.

All the programs running in the computer will be displayed in the computer task manager, but many processes have nothing to do with Internet addiction, such as computer operating system processes, etc., which do not need to be monitored. And some typical processes related to Internet addiction, such as browser process (IE browser process, Google browser process, Sogou browser process, etc.), game process (Age of Empires game, Three Kingdoms game, etc.), offline video watching software (Windows multimedia player), instant messaging software (QQ, Skype, MSN, etc.) need to focus on monitoring. At the same time, different Internet addiction users may run different programs. For example, for game Internet addiction users, they may run more game processes, so monitoring this item can more accurately determine the user's advanced and complex behavior events.

Complex event generation module: used to compare each simple event with its respective threshold, if it is greater than or equal to the threshold range, then keep the simple event, and then generate complex events according to the combination of all the retained simple events; otherwise, continue to collect simple events ; The complex events include watching videos offline, watching videos online, real-time strategy games, board games, browsing web pages, online chatting and downloading operations;

Time generalization and normalization processing module: used to sort out the complex events that each user does on a certain day, and generalize the occurrence time of complex events, that is, to divide a day into several time periods; Normalization processing, that is, rounding the duration of complex events; and sending the processed complex events to the frequent behavior generation sub-acquisition module;

Frequent behavior generation sub-acquisition module: when establishing a sample library, it is used to obtain the same complex events between multiple Internet addicted users, collect the generation times of the above-mentioned same complex events, and select out according to the number threshold set by the user The complex events greater than the threshold are regarded as frequent Internet addiction complex events, and then a set composed of the above-mentioned frequent Internet addiction complex events is obtained to describe the generator of the distribution of user behavior characteristics, and sent to the sample library construction module; used to obtain multiple non- For the same complex events between Internet addicted users, collect the occurrence times of the above-mentioned same complex events, and select the complex events greater than the threshold as frequent non-Internet addiction complex events according to the number threshold set by the user, and then obtain the above frequent A generator for describing the distribution of user behavior characteristics, which is a collection of non-Internet addiction complex events, and sent to the sample library construction module;

When detecting the user's Internet addiction status, it is used to obtain the same complex events performed by the user to be tested within a period of time, collect the number of occurrences of the above-mentioned same complex events, and select complex events greater than the threshold according to the number of times threshold set by the user As a frequent complex event, obtain a generator for describing the distribution of behavioral characteristics composed of the above frequent complex events, and send it to the data processing and detection module;

Sample library construction module: it is used to calculate the score of frequent Internet addiction complex event generators using the generator-based sample construction algorithm. If the score exceeds the threshold, it will be saved as Internet addiction, otherwise it will be deleted, or the EPBSBA algorithm (based on the exposure mode The sample construction algorithm) calculates the generators of frequent and complex events, obtains the generators of the revealed patterns and saves them by category;

Data processing and detection module: used to compare the frequent behavior generators to be detected with the Internet addiction sample library, if the frequent complex event generators to be detected are included in the sample library, it is determined to be Internet addiction, otherwise it is not Internet addiction, and Output judgment results; or compare frequent behavior generators to be detected with the sample library, use EPBPDA algorithm (Internet addiction detection algorithm based on exposure mode) to calculate their Internet addiction scores and non-Internet addiction scores, and judge their category according to the score ;

Early warning intervention module: when it is detected that the user's computer interaction behavior is an Internet addiction behavior, the device will issue an early warning prompt to warn the user that he has symptoms of Internet addiction, and limit the time range for the user to continue using the computer through a timer. It is 1-2 hours; once the time limit is exceeded, the software device will be further started to shut down the computer, so as to prevent the user from using the computer, so as to effectively intervene in Internet addiction.

The method for detecting an Internet addiction detection device based on user computer interaction events, as shown in Figure 2, comprises the following steps:

Step 1. Collect the interaction variables between multiple Internet addict users and multiple non-Internet addict users and the computer as simple events, as shown in Table 1, including computer CPU usage, memory usage, left mouse click times per unit time, The number of right mouse clicks per unit time, the number of keyboard strokes per unit time, the number of pixels moved by the mouse per unit time, network traffic and running processes;

Table 1

Step 2. Compare each simple event with its respective threshold. If it is greater than or equal to the threshold range, then keep the simple event, and then generate a complex event according to the combination of all the retained simple events; otherwise, continue to collect simple events; Complex events include watching videos offline, watching videos online, real-time strategy games, board games, browsing the web, chatting online, and downloading operations;

The types of complex events are shown in Table 2,

Table 2

The process of generating a complex event is as follows:

If the CPU usage exceeds or equals to the threshold range of 45% to 55%, the memory usage exceeds or equals to the threshold range of 55% to 65%, and the Internet traffic exceeds or equals to the threshold range of 35 to 45MB, a complex event of watching videos online is generated. ;

If the CPU usage exceeds or equals to the threshold range of 45% to 55%, the memory usage exceeds or equals to the threshold range of 55% to 65%, and the running process is a multimedia player, a complex event of watching videos offline is generated;

If the CPU usage exceeds or is equal to the threshold range of 45% to 55%, the memory usage exceeds or equals to the threshold range of 55% to 65%, and the number of clicks of the left mouse button per unit time exceeds or equals to the threshold range 25 to 35 times, click the mouse per unit time The number of right clicks exceeds or equals the threshold range of 10 to 20 times, the number of pixels moved by the mouse per unit time exceeds or equals the threshold range of 1550 to 1650, the number of keyboard strokes per unit time exceeds or equals the threshold range of 80 to 120 times, and the running process is For real-time strategy games, generate a complex event for playing real-time strategy games;

If the CPU usage exceeds or equals to the threshold range of 45% to 55%, the memory usage exceeds or equals to the threshold range of 55% to 65%, and the number of clicks of the left mouse button per unit time exceeds or equals to the threshold range of 25 to 35 times, the mouse moves per unit time. If the number of pixels exceeds or is equal to the threshold range of 1550 to 1650, and the running process is a board game, a complex event of playing a board game is generated;

If the CPU usage exceeds or equals to the threshold range of 45% to 55%, the number of clicks of the left mouse button per unit time exceeds or equals to the threshold range of 25 to 35 times, and the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550 to 1650, And if the running process is a board game, another complex event of playing a board game is generated;

If the number of clicks on the left mouse button per unit time exceeds or equals to the threshold range of 25 to 35 times, the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550 to 1650, and the running process is a browser, a complex page browsing process is generated. event;

If the number of clicks on the left mouse button per unit time exceeds or equals to the threshold range of 25 to 35 times, the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550 to 1650, the Internet traffic exceeds or equals to the threshold range of 35 to 45MB, and the running process For the browser, another complex event for browsing the web page is generated;

If the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550 to 1650, the Internet traffic exceeds or equals to the threshold range of 35 to 45MB, and the running process is a browser, another complex event of web browsing is generated;

If the number of keystrokes per unit time exceeds or is equal to the threshold range of 80 to 120 times, and the running process is instant chat software, a complex event of online chat will be generated;

If the number of pixels moved by the mouse per unit time exceeds or equals the threshold range of 1550 to 1650, the number of keystrokes per unit time exceeds or equals the threshold range of 80 to 120 times, and the running process is instant chat software, another online chat will be generated complex events;

If the Internet traffic exceeds or is equal to the threshold range of 35-45MB, a complex event of downloading data will be generated;

For example, when the following simple events are detected: the CPU usage is greater than 50%, the memory usage is greater than 60%, and the number of mouse clicks per minute is 100 times/minute (comprehensively calculated based on the number of left and right mouse clicks and the number of pixels moved by the mouse) obtained later), the number of keystrokes per unit time is greater than 90 times/minute, and the running application is detected to be the game sanguosha.exe, and the network traffic is greater than 40MB/minute, a complex event of playing a real-time strategy game is generated.

Step 3. Generalize the occurrence time of complex events, that is, divide a day into 6:00～11:00, 11:00～14:00, 14:00～18:00, 18:00～23:00, 23 :00～6:00 five time periods, and the duration is normalized, that is, the duration of complex events is rounded up in units of 10 minutes;

In the process of complex event generation, due to the diversity of event occurrence times, a large number of different time points will be generated, resulting in the generation of too many complex events, that is, the problem of fragmentation of complex events. In fact, there is no essential difference in the semantics of many events that occur at different times. For example, "47 minutes online at 8 am" and "52 minutes online at 9 am" are both online in the morning, with similar start times and similar durations , so from the perspective of semantic rationality, these two complex events can be regarded as one event, which is the time generalization processing operation of complex events. Table 3 lists the occurrence time range after generalization processing.

As can be seen from Table 3, a day is generalized into five levels, namely: morning period: 6:00 am to 11:00 am; noon period: 11:00 am to 2:00 pm; afternoon period: 2:00 pm to evening 6 o'clock; evening time: 6:00 pm to 11:00 pm; dawn time: 11:00 pm to 6:00 am the next day. Using this generalization strategy can effectively reduce time fragmentation. Based on the above method, as long as the event occurs at a certain generalization level, it will be attributed to a certain time period.

In addition, considering that the duration also has the problem of diversity, that is, the difference between 47 minutes and 52 minutes is not obvious, so when the time of occurrence is the same, they can be regarded as the same event, which requires normalization of the duration The rounding operation is to normalize the duration into an integer form according to the rounding principle. At this time, both 47 minutes and 52 minutes are rounded to 50 minutes, which facilitates the further processing of complex events.

In the embodiment of the present invention, it is detected that the CPU usage rate of the user's computer is greater than 60%, the memory usage rate is greater than 50%, and the running program is a Windows multimedia player program, then the user is watching videos offline. If it is inferred that the user watched the video offline continuously during the 46 minutes from 12:00 to 12:46 noon, then a complex event will be generated by using the complex event generalization and normalization processing method, that is, watching the video offline at noon 50 Minutes, recorded as Bc50.

Step 4. Acquire the same complex events between multiple Internet addicted users and the same complex events between multiple non-Internet addicted users, collect the number of occurrences of the above-mentioned same complex events, and set the threshold according to the number of times set by the user , select complex events greater than the threshold as frequent Internet addiction complex events and frequent non-Internet addiction complex events, and then obtain multiple generators for describing the distribution of user behavior characteristics composed of the above-mentioned frequent Internet addiction complex events, and obtain by Multiple generators used to describe the distribution of user behavior characteristics formed by the above-mentioned frequent non-internet addiction complex events are sent to the sample library construction module;

In the embodiment of the present invention, taking the collection of Internet addicted users as an example, after collecting the simple event data of computer interaction behaviors of 10 Internet addicted users on a certain day, the behavior sequence is obtained through complex event reasoning as follows:

Internet addiction user 1: {Bc60, Ce80, Ae70, Ed80, Be100, Ab100, Eb30};

Internet addiction user 2: {Bc60, Ce80, Be100, Bb40, Ga20, Eb30, Ac20};

Internet addiction user 3: {Bc60, Ce80, Ae70, Ed80, Da30, Bb40};

Internet addiction user 4: {Ae70, Bc60, Ed80, Be100, Ab100};

Internet addiction user 5: {Ae70, Bc60, Ed80, Be100, Ab100, Ga20, Eb30};

Internet addiction user 6: {Ae70, Ed80, Da30, Bb40, Fa20};

Internet addiction user 7: {Be100, Ab100, Bc60, Ce80, Da30};

Internet addiction user 8: {Be100, Ab100, Ga20, Ad60};

Internet addiction user 9: {Ab100, Bc60, Bb40, Fa20, Ad60};

Internet addiction user 10: {Be100, Ed80, Da30, Eb30, Ac20}.

Use the frequent behavior generator acquisition algorithm to find the generator (Generator) corresponding to the computer interaction behavior sequence of the above 10 Internet addicted users on a certain day as follows:

Generate child G1: <Ab100,Be100>

Producer G2: <Ab100,Bc60>

Generate child G3: <Ed80,Ae70>

Produces child G4: <Ce80,Bc60>

Generate child G5: <Ed80, Ab100>

Generate child G6: <Ae70,Be100>

Generate child G7: <Ae70,Bc60>

Produced child G8: <Ae70, Ab100>

Produce child G9: <Ce80,Be100>

Generate child G10: <Be100>

Generate child G11: <Bc60>

Produced child G12: <Ab100>

Generate child G13: <Ed80>

Spawn child G14: <Ae70>

Generate child G15: <Ce80>

Get the sub method as follows:

Set the input as D and MinSup, where D represents the standardized complex event database, MinSup is the support threshold of the frequent behavior generator to be obtained (set by the user according to the needs), and the output result is FG (that is, the frequent behavior generator gather).

The steps to obtain frequent behavior generators are as follows:

1) Scan each item in the complex event database D to generate an item set F, record the number of occurrences of each item in the database (that is, support), and arrange the item set F in descending order of support.

2) Construct an FP-Tree tree for the itemset set F.

3) Start accessing from the item with the minimum support (that is, access according to the reverse order of the itemsets in the itemset set F), scan the complex event database and establish its condition database.

4) Select each item in the conditional database and its power set, and judge whether it is a generator according to the definition of the generator. If yes, further judge whether there is a proper subset with the same support as the generator, if it exists, delete the generator, otherwise keep the generator and store it in the frequent behavior generation subset FG, and finally output FG.

The instructions for generating children are as follows:

All combinations of complex events that appear in the same sequence of complex events with the same number of occurrences are called frequent complex event equivalence classes. A producer is the simplest representation of this equivalence class. If the behavior of creating a child occurs, it means that all accompanying behaviors in the equivalence class to which it belongs will occur, so it is only necessary to use the child to represent its corresponding equivalence class.

Table 4

table 5

Assume that a and c represent 60 minutes of online gaming at 9:00 and 40 minutes of online shopping at 11:00, respectively. In the equivalence class {a, ac}, a is a sub-behavior of ac, so it can be said that as long as a occurs, ac must occur, then ac can be represented by a, so the equivalence class can only be represented by a.

All equivalence classes in Table 5 are shown in Figure 3. In Figure 3, the black box line shows the equivalence class with a frequency of 2. The behavior set contained in this equivalence class is {ab, ae, abc, abe, ace, abce}, and its corresponding generator It is {ab, ae}, which means that as long as the two joint behaviors of ab and ae occur, it can significantly represent all the behaviors in the set. Subsequent processing only needs to generate sub-corresponding behaviors for this, instead of dealing with all behaviors in the equivalence class .

Step 5. Use the generator-based sample construction algorithm to calculate the score of frequent Internet addiction complex event generators. If the score exceeds the threshold, save it by classification, otherwise delete it, and save the frequent non-Internet addiction complex event generators; or use The EPBSBA algorithm calculates the generators of frequent and complex events, and obtains the generators of the revealed patterns and saves them by category;

The generator-based sample construction algorithm is used to calculate the score of frequent Internet addiction complex event generators, specifically:

Step 5-1. Determine the weight coefficient of complex event types according to user needs, and ensure that the weight coefficient of real-time strategy games > the weight coefficient of watching videos online > the weight coefficient of board games > the weight coefficient of browsing webpages > the weight of watching videos offline Coefficient > weight coefficient of online chat > weight coefficient of download, and the total weight coefficient is 1;

In the embodiment of the present invention, the complex event type weight coefficients are shown in Table 6:

Table 6

Step 5-2. Determine the weight coefficient of the time generalization segment of complex events according to user needs, and ensure that the weight coefficient of dawn > the weight coefficient of evening > the weight coefficient of noon > the weight coefficient of afternoon > the weight coefficient of morning, and the total number of weight coefficients is 1;

In the embodiment of the present invention, the weight coefficients of the complex event time generalization section are shown in Table 7:

Table 7

Step 5-3: Multiply the weight coefficient of each complex event type in the generator, the weight coefficient of the complex event time generalization section and the duration to obtain the score, and sum the scores of each complex event to obtain the score of the generator ;

A user's frequent computer interaction behavior can be represented by multiple frequent behavior generators, but not all of these generators have the characteristics of Internet addiction, so it is necessary to further screen out the Internet addiction behavior generators in order to classify Internet addiction behavior.

分别代表第i个行为产生子集合中第k个复杂事件的行为类型、发生时间和持续时间，这与前面的复杂事件含义相同，并由用户给出这些参数对应的权重系数；The algorithm uses scoring method to select Internet addiction behavior generators, and its scoring formula is shown in formula (1), where the parameter

respectively represent the behavior type, occurrence time and duration of the kth complex event in the i-th behavior generation subset, which have the same meaning as the previous complex event, and the weight coefficients corresponding to these parameters are given by the user;

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; G</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; type</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; weight</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; t</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; dur</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

为G_i中第k个复杂事件的事件类型权重系数；

为G_i中第k个复杂事件的事件时间泛化段权重系数；

为G_i中第k个复杂事件的持续时间。Among them, G _i is the i-th generator, n is the number of complex events contained in G _i ,

is the event type weight coefficient of the kth complex event in G _i ;

is the event time generalization segment weight coefficient of the kth complex event in G _i ;

is the duration of the kth complex event in G _i .

Step 5-4, if the score exceeds the threshold 4, save it, otherwise delete it.

EP is the abbreviation of Emerging Pattern, including JEP (Jumping Exposure Pattern) and eEP (Essential Exposure Pattern). Itemsets with significant changes can capture the differential features between target classes and non-target classes, so a classifier with good classification effect can be established based on EP.

The EPBSBA algorithm is used to calculate the generators of frequent and complex events, and the generators of exposure patterns are obtained and stored in categories; the algorithm screens out JEP (Jumping Exposure Pattern) and eEP (Essential Exposure Pattern) of Internet addiction and non-Internet addiction respectively, Among them, the former is obtained through the difference set, and the latter is obtained through the union set. Then, filter out EPs according to a predetermined EP threshold (assuming the threshold is 1) and calculate the corresponding growth rate. Next, check whether there are subsets that contain each other in the screening set. If so, further delete operations are required; if not, end.

According to the minimum support threshold MinSup (defined by the user), the sub-algorithm for generating frequent behaviors is invoked, and the sub-sets FG ₁ and FG ₂ are calculated for the two types of databases (namely D ₁ and D ₂ ) for Internet addiction and non-Internet addiction respectively.

In the embodiment of the present invention, assuming that MinSup=3/10, 9 _Internet addiction behavior generators G ₃ , G ₄ , G ₇ , G ₁₀ , G 11 , G ₁₂ , G ₁₃ , G ₁₄ and G ₁₅ are screened out, and Constitute the initial Internet addiction Generator library, as shown in the Generator under the Internet addiction label in Table 8. Since the non-Internet addiction Generator is generated in the same way as the Internet Addiction Generator, the only difference is that it is obtained by collecting computer interaction data of normal users, so I won’t go into details here. Assume that the non-Internet addiction Generator is shown in Table 8 under the Non-Internet Addiction label, this type of library is used by the EPBPDA algorithm, and the GBPDA algorithm does not use this type of library. The number in parentheses after the Generator represents the support, which is only useful for the EPBPDA algorithm, and the GBPDA algorithm does not use this indicator.

Table 8

Obtaining JEP:

A JEP is an EP with a growth rate of ∞. After obtaining the frequent behavior generators FG ₁ and FG ₂ of Internet addiction and non-Internet addiction, the JEP between the two types of databases can be obtained through the set difference operation, that is, for the Internet addiction data D ₁ , The result set of FG ₁ -FG ₂ represents the JEP of database D ₁ to D ₂ , because the difference set of FG ₁ -FG ₂ represents the generator that exists in FG ₁ but does not exist in FG ₂ , therefore, the difference of FG ₁ -FG ₂ No matter how much support an element in the set has in D ₁ , its support in D ₂ must be zero, because D ₂ does not contain this element. Further, according to the definition of the growth rate, the growth rate of the elements in the difference set FG ₁ -FG ₂ from D ₂ to D ₁ is ∞, so the elements in the difference set FG ₁ -FG ₂ must be the JEP of D ₁ to D ₂ , denoted as JEP _D1/D2 . Similarly, the difference set of FG ₁ -FG ₂ represents the JEP of D ₂ to D ₁ , denoted as JEP _D2/D1 .

Obtaining eEP:

An eEP is an EP greater than the growth rate threshold 1. First, the intersection operation of subsets FG ₁ and FG ₂ is generated for frequent behaviors. Then, for each item in the intersection, it is judged whether its growth rate is greater than or equal to the given growth rate threshold MinGr, and if the condition is met, it is stored in the candidate set. Next, judge whether there are subsets that contain each other in the candidate set. If so, take the smallest subset and delete its superset. If not, do not operate on the candidate set. Thus, the elements contained in the final candidate set are eEPs.

The acquisition of EP. Calculate the union of eEP and JEP of Internet addiction D ₁ and non-internet addiction database D ₂ respectively, that is, the EP of D ₁ is the union of eEP and JEP _D1/D2 , and the EP of D ₂ is the union of eEP and JEP _D2/D1 .

Specific steps are as follows:

Step 5-a, obtaining the generators that belong to frequent Internet addiction complex events but not frequent non-Internet addiction complex events;

The JEP _PIU for internet addiction is:

D1-D2={<Be100>,<Ab100>,<Bc60,Ce80>}

Step 5-b. Obtain the generators that do not belong to frequent Internet addiction complex events but belong to frequent non-Internet addiction complex events;

The JEP _NPIU for non-Internet addiction is:

D ₂ -D ₁ ={<Ga40>,<Fe40>,<Ae70,Fe40>}

Step 5-c, obtain the co-producers that belong to both frequent Internet addiction complex events and frequent non-Internet addiction complex events, and determine their support in the frequent Internet addiction complex events, that is, the occurrence rate, and determine the frequent non-Internet addiction complex events support in complex events;

eEP _PIU (Internet Addiction eEP) = eEP _NPIU (Non-Internet Addiction eEP) is:

D1∩D2={<Ae70>,<Bc60>,<Ce80>,<Ed80>,<Ae70,Ed80>,<Ae70,Bc60>}

The support degree of each frequent complex event in frequent Internet addiction complex events and in frequent non-Internet addiction complex events is shown in Table 8 (percentage in brackets).

Step 5-d, calculate the growth rate of the co-producer in the frequent Internet addiction complex event, that is, divide its support degree in the frequent Internet addiction complex event by the support degree in the frequent non-Internet addiction complex event; The growth rate of the co-producer in the frequent non-Internet addiction complex event is to divide its support degree in the frequent non-Internet addiction complex event by the support degree in the frequent Internet addiction complex event;

Taking <Bc60>[7/6.5] as an example, since <Bc60> has a support rate of 70% in the Internet addiction category and 65% in the non-Internet addiction category, its growth rate is 7/6.5, The growth rates of other EPs can also be obtained separately according to this, and will not be repeated here.

Step 5-e, compare the size of the above two growth rates, retain the generator whose growth rate is greater than 1, and delete the generator whose growth rate is less than 1;

According to the pre-specified EP _PIU threshold (assuming the threshold is 1), the Internet addiction category (eEP _PIU ) is screened out and the corresponding growth rate is calculated. The screening result is {<Bc60>[7/6.5],<Ae70,Ed80>[1.5] >}, where the numbers in square brackets indicate the growth rate.

Then check whether there are subsets that contain each other in the screening set, and if so, delete it; if not, end it. Ends because this set has no mutually inclusive subsets.

Therefore, the final eEP _PIU of Internet addiction = {<Bc60>[7/6.5],<Ae70,Ed80>[1.5]>}, and the minimum and maximum thresholds of the corresponding growth rates are 7/6.5 and 1.5, respectively.

Non-Internet addicts are similarly as follows:

According to the pre-specified EP _NPIU threshold (the threshold is assumed to be 1), the eEP _NPIU of non-Internet addiction category is screened to obtain {<Ae70>,<Ce80>,<Ed80>,<Ae70,Bc60>}.

Since <Ae70> in the screening set is a subset of <Ae70, Bc60>, <Ae70, Bc60> needs to be deleted, so

eEP _NPIU ={<Ae70>[90%/50%=1.6],<Ce80>[70%/40%=1.75],<Ed80>[70%/55%=7/5.5]}

Therefore, the minimum threshold and maximum threshold of growth rate in the non-Internet addiction category are 7/5.5 and 1.75, respectively.

Step 5-f, combining the generators of the frequent Internet addiction complex events and the union of the generators of the frequent non-Internet addiction complex events, that is, obtaining the generators of the exposure pattern and storing them in categories.

Step 6, collect the interaction variables between the user to be tested and the computer within a period of time, and repeat steps 2 to 3;

The collected online behavior sequence of the user to be detected for a week is as follows:

Monday: {Be100, Ab100, Bc60, Ae70, Db90, Fe40, Da50};

Tuesday: {Ed80, Ab100, Ee30, Fd30};

Wednesday: {Be100, Ed80, Bc60, Ae70, Db90, Fe40}

Thursday: {Cc30,Da40};

Friday: {Be100, Ce20};

Saturday: {Fa90,Ad20};

Sunday: {Be100, Ab100, Ed80, Bc60, Ae70, Db90, Fe40, Ge60, Aa70};

Step 7. Obtain the same complex events performed by the user to be tested within a period of time, and collect the occurrence times of the above-mentioned same complex events, and select complex events greater than the threshold as frequent complex events according to the number threshold set by the user, and then Obtain a generator for describing the distribution of behavioral characteristics composed of the above-mentioned frequent and complex events, and send it to the data processing and detection module;

Step 8, comparing the frequent behavior generator to be detected with the Internet addiction sample library, if the frequent complex event generator to be detected is included in the sample library, then determine it is Internet addiction, otherwise it is not Internet addiction, and output the judgment result; or Compare the frequent behavior generators to be detected with the sample library, and use the EPBPDA algorithm to calculate their Internet addiction scores and non-Internet addiction scores respectively, and judge the scores of the two. The frequent and complex events to be detected belong to the category with a higher score;

In the embodiment of this patent, two Internet addiction pattern detection algorithms are proposed, namely, Generator-Based PIU Detecting Algorithm-GBPDA and EP-Based PIU Detecting Algorithm-EPBPDA. .

GBPDA algorithm:

The behavior event to be detected is scanned, and if the behavior event includes any sub-Generator in the database, it is judged that the behavior to be detected is an Internet addiction behavior. Considering that the larger the value of the duration, the more likely the behavior to be detected is an Internet addiction behavior, therefore, for the complex event CE _m (type _m ,t _m ,dur _m ) in the behavior sequence to be detected and Generator For the complex event CE _n (type _n ,t _n ,dur _n ), if type _m =type _n , t _m =t _n , and dur _m >=dur _n , then CE _m and CE _n are still considered to be the same complex event.

In the embodiment of the present invention, scan the frequent itemsets of the online behavior sequence of the user to be detected for a week, where the support threshold is 3/7, then the frequent itemsets are {Be100, Ab100, Bc60, Ae70, Db90, Fe40}. It is found that it contains at least one generator in the Internet addiction class, namely <Be100>, according to the GBPDA algorithm Internet addiction determination principle, that is, as long as it contains a generator in the Internet addiction behavior pattern library, it is judged to have Internet addiction behavior, so the The behavior to be tested is finally determined to be an Internet addiction behavior.

EPBPDA algorithm:

After obtaining JEP and eEP, the scores of Internet addiction and non-Internet addiction are calculated respectively, and then compared, and the higher score is selected as the final judgment result.

The formula of Score(C,D _i|j ) is as follows:

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; C</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; RS</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\frac{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Max</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; RS</span>}}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; Sup</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="WS"\; filenames="HDA0000368918860000041.png"\; state="\{\{state\}\}">WS</figure-callout></span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Min</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; WS</span>}}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; Sup</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; Y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&NotEqual;</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Among them: i and j are the class label numbers of the sample library, D _i and D _j are both sample libraries, which are unified expressions, and are distinguished by the subscript i|j or j|i. Note that there are only 2 types of sample libraries in this patent, i.e. Internet addiction sample library and non-Internet addiction sample library, so in order to better distinguish between the two, D ₁ is used to represent the Internet addiction library, and D ₂ is used to represent the non-Internet addiction sample library library. X represents the generator of the exposure pattern of D ₁ , and Y represents the generator of the exposure pattern of D ₂ ; RS is the positive sample set of the target class, that is, the set that plays an active role, and WS is the reverse sample set of the target class, namely A set that plays a negative role; Gr(X,D _j|i ,D _i|j ) represents the growth rate of X from one class to another class, which is determined by the order of the factors before and after the separator "|". For example, Gr(X,D ₂ ,D ₁ ) indicates the growth rate of X from D ₂ to D ₁ , Gr(X,D ₁ ,D ₂ ) indicates the growth rate of X from D ₁ to D ₂ ; MaxGr _RS indicates the target The maximum growth rate threshold in the class set, MinGr _WS represents the minimum growth rate threshold in the non-target class set, C is the complex event to be classified, Sup _c (X) represents the support of X in C, Sup _c (Y) represents The support of Y in C.

表示了Y的贡献值。因此，公式（2）是综合考虑的结果。If the sample C to be detected is regarded as RS, that is, the target class, the first item in formula (2) is used to calculate the score; if C is regarded as WS, that is, the non-target class, then the score is calculated using The second item calculates the score. Among them, the first term in formula (2) $\underset{x\&Element;\mathrm{RS}}{\mathrm{\&Sigma;}}\frac{\mathrm{Gr}(x,{\mathrm{D.}}_{j|i},{\mathrm{D.}}_{i|j})}{\mathrm{Gr}(x,{\mathrm{D.}}_{j|i},{\mathrm{D.}}_{i|j})+\mathrm{Max}{\mathrm{Gr}}_{\mathrm{RS}}}\&times;{\mathrm{Sup}}_c\left(x\right),$ Not only the contribution of eEP and JEP to scoring is given, but the parameter MaxGr also emphasizes the contribution of JEP. At the same time, the EP of the non-target class also contributes to the fact that sample C belongs to the target class. For C, when judging whether C belongs to D ₁ category, the EP in D ₂ category, namely Y, can also make a certain contribution. When C contains Y, it can be judged that the probability that C belongs to D ₁ class is 1/(Gr(X,D ₂ ,D ₁ )+1), when Gr(Y,D ₂ ,D ₁ ) is relatively large, The contribution of Y is very small and negligible, but when Gr(Y,D ₂ ,D ₁ ) meets the minimum threshold and is small, it will show a large contribution to the judgment that C belongs to D ₁ , so it needs to be considered The impact of this factor, while considering the difference between ordinary EP and JEP, the second item in formula (2)

Indicates the contribution value of Y. Therefore, formula (2) is the result of comprehensive consideration.

In the embodiment of the present invention, the Generator and its corresponding support degree are obtained from the online behavior sequence of the user to be detected for a week, that is,

G _C ={<Be100>(50%),<Ab100>(60%),<Bc60>(45%),<Ae70>(30%),<Ae70,Fe40>(50%)}.

The JEP of class C to be tested is obtained as follows:

JEP _PIU-C ={<Be100>(50%),<Ab100>(60%)}

JEP _NPIU-C ={<Ae70,Fe40>(50%)}

Since the EP threshold = 1, the eEP sets of the Internet addiction class and the non-Internet addiction class in the generation of the class C to be detected are respectively:

eEP _PIU-C ={<Bc60>(45%)}

eEP _NPIU-C ={<Ae70>(30%)}

Therefore, the sum of the Internet addiction and non-Internet addiction EP sets of the class to be detected is respectively:

EP _PIU-C ={<Be100>(50%),<Ab100>(60%),<Bc60>(45%)}

EP _NPIU-C ={<Ae70,Fe40>(50%),<Ae70>(30%)}

MaxGr _RS = MaxGr _D1 = 1.5, MinGr _WS = MinGr _D2 = 7/6.5 for Internet addiction EP; MaxGr _RS = MarGr _D2 = 1.75 for non-Internet addiction EP, MinGr _WS = MinGr _D1 = 7/5.5 .

Next, the scores of each generator of the user behavior to be detected in the Internet addiction category D ₁ and the non-Internet addiction category D ₂ are calculated respectively.

1) Because <Be100> (50%) and <Ab100> (60%) are the jumping exposure patterns of Internet addiction class D ₁ , for Internet addiction class D ₁ , it is X, that is, it belongs to the positive set and RS, and for non- Internet addiction class D ₂ is Y, that is, it belongs to the reverse set WS, so there are:

Gr(<Be100>,D ₂ ,D ₁ )=∞，Sup _c (<Be100>)=50%

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; be</span>}\mathrm{<span\; class="notranslate">\; 100</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; lim</span>}\frac{<span\; class="notranslate">\; \&infin;</span>}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Max</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; lim</span>}\frac{<span\; class="notranslate">\; \&infin;</span>}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}$

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; be</span>}\mathrm{<span\; class="notranslate">\; 100</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="lim"\; filenames="HDA0000368918860000041.png"\; state="\{\{state\}\}">lim</figure-callout></span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Min</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="lim"\; filenames="HDA0000368918860000041.png"\; state="\{\{state\}\}">lim</figure-callout></span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}$

Gr(<Ab100>,D ₂ ,D ₁ )=∞，Sup _c (<Ad100>)=60%

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="100"\; label="Ab"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Ab</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 100</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; lim</span>}\frac{<span\; class="notranslate">\; \&infin;</span>}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Max</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.6</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; lim</span>}\frac{<span\; class="notranslate">\; \&infin;</span>}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; 5</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.6</span>}$

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="100"\; label="Ab"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Ab</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 100</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="lim"\; filenames="HDA0000368918860000041.png"\; state="\{\{state\}\}">lim</figure-callout></span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Min</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.6</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="lim"\; filenames="HDA0000368918860000041.png"\; state="\{\{state\}\}">lim</figure-callout></span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.6</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.6</span>}$

2) Because <Bc60> (45%) is the child of Internet addiction class D1, it is X for Internet addiction class D1, that is, it belongs to the forward set and RS, and it is Y for non-Internet addiction class D2, that is, it belongs to The reverse set WS, therefore has:

Gr(<Bc60>,D ₂ ,D ₁ )=70%/65%=7/6.5,Sup _c (<Bc60>)=45%

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="60"\; label="Bc"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Bc</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="60"\; label="Bc"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Bc</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="60"\; label="Bc"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Bc</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Max</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; Sup</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="60"\; label="Bc"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Bc</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}}{\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.45</span>}$

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="60"\; label="Bc"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Bc</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="60"\; label="Bc"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Bc</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Min</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; \&times;</span>{\mathrm{<span\; class="notranslate">\; Sup</span>}}_{\mathrm{<span\; class="notranslate">\; c</span>}}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="60"\; label="Bc"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Bc</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; )</span>$

$<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.45</span>}$

3) Because <Ae70> (30%) is the child of non-Internet addiction class D ₂ , it is X for non-Internet addiction class D ₂ , that is, it belongs to positive set and RS, and it is Y for Internet addiction class D ₁ , which belong to the reverse set WS, so there are:

Gr(<Ae70>,D ₁ ,D ₂ )=90%/50%=1.6, Sup _c (<Ae70>)=30%;

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; Ae</span>}\mathrm{<span\; class="notranslate">\; 70</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; Ae</span>}\mathrm{<span\; class="notranslate">\; 70</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Min</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.3</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 1.6</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 5.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.3</span>}$

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; Ae</span>}\mathrm{<span\; class="notranslate">\; 70</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="60"\; label="Bc"\; filenames="HDA0000368918860000032.png"\; state="\{\{state\}\}">Bc</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; Ae</span>}\mathrm{<span\; class="notranslate">\; 70</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Max</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.3</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1.6</span>}}{\mathrm{<span\; class="notranslate">\; 1.6</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; 75</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.3</span>}$

4) Because <Ae70, Fe40> (50%) is the jumping exposure pattern of non-Internet addiction D ₂ , it is X for non-Internet addiction D ₂ , that is, it belongs to the positive set and RS, and for Internet addiction D ₁ is Said to be Y, which belongs to the reverse set WS, so there are:

Gr(<Ae70,Fe40>,D ₁ ,D ₂ )=∞，Sup _c (<Ae70,Fe40>)=50%;

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; Ae</span>}\mathrm{<span\; class="notranslate">\; 70</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; Fe</span>}\mathrm{<span\; class="notranslate">\; 40</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="lim"\; filenames="HDA0000368918860000041.png"\; state="\{\{state\}\}">lim</figure-callout></span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Min</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="lim"\; filenames="HDA0000368918860000041.png"\; state="\{\{state\}\}">lim</figure-callout></span>}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 5.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}$

$\mathrm{<span\; class="notranslate">\; Score</span>}<span\; class="notranslate">\; (</span><span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; Ae</span>}\mathrm{<span\; class="notranslate">\; 70</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; Fe</span>}\mathrm{<span\; class="notranslate">\; 40</span>}<span\; class="notranslate">\; ></span><span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; D.</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; lim</span>}\frac{<span\; class="notranslate">\; \&infin;</span>}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; Mar.</span>}{\mathrm{<span\; class="notranslate">\; Gr</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; lim</span>}\frac{<span\; class="notranslate">\; \&infin;</span>}{<span\; class="notranslate">\; \&infin;</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; 75</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}$

Putting the above calculation results into formula (2), the scores of Internet addiction and non-Internet addiction of the behavior to be tested are obtained as follows:

${\mathrm{<span\; class="notranslate">\; Score</span>}}_{\mathrm{<span\; class="notranslate">\; PIU</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 50</span>}<span\; class="notranslate">\; \%</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 60</span>}<span\; class="notranslate">\; \%</span><span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}}{\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.45</span>}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 1.6</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 5.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.3</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1.39</span>}$

${\mathrm{<span\; class="notranslate">\; Score</span>}}_{\mathrm{<span\; class="notranslate">\; NPIU</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.6</span>}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 6.5</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.45</span>}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 1.6</span>}}{\mathrm{<span\; class="notranslate">\; 1.6</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; 75</span>}}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.3</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \&times;</span>\mathrm{<span\; class="notranslate">\; 0.5</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0.85</span>}$

Since SCORE _PIU =1.39 is greater than SCORE _NPIU =0.85, it is finally determined that the behavior to be detected belongs to the category of Internet addiction.

Step 9. If it is detected that the user's computer interaction behavior is an Internet addiction behavior, an early warning intervention module is used to issue an early warning prompt, and a timer is used to limit the user to continue using the computer for 1 hour; if the time limit is exceeded, the computer is turned off.

Once the time limit is exceeded, the software device will be further started to shut down the computer. In the embodiment of the present invention, the API interface of the Windows system, namely the Shutdown function, is called to shut down the computer, thereby effectively intervening in Internet addiction.

In the embodiment of the present invention, the application effect of the proposed method for detecting Internet addiction is tested through experiments.

(1) Evaluation of the efficiency of Internet addiction detection algorithm.

Using the proposed two Internet addiction detection algorithms GBPDA algorithm and EPBPDA algorithm, mainly evaluate from the time complexity and space complexity. From Figure (a) in Figure 4, it can be seen that with the increase of the data scale (that is, the computer interactive behavior data generated in one day as the detection unit), the running time of the two algorithms from collecting four days of data to ten days of data The average increase trend, at this time, because the more data, the more processing time the algorithm takes. In addition, the running time of the EPBPDA algorithm is slightly higher than that of the GBPDA algorithm because the EPBPDA algorithm requires more processing than the GBPDA algorithm, thus resulting in more running time. Judging from the memory space in (b) in Figure 4, the upper limit of memory required by the two algorithms is not greater than 16MB of memory space. At the same time, as the size of the data set increases, the memory space occupation will gradually increase. In addition, the EPBPDA algorithm consumes more memory space than the GBPDA algorithm, because EP mining is based on Generator mining, so it needs to occupy relatively more memory resources.

(2) Evaluation of the effect of Internet addiction detection algorithm.

It can be seen from Figure (a) in Figure 5 that, firstly, both algorithms have achieved a high accuracy rate, which shows the effectiveness of the two Internet addiction detection algorithms, and as the data scale increases, the two algorithms The accuracy of the algorithm has been improved to a certain extent, indicating that the more data, the better the detection effect. Secondly, the correct rate of the EPBPDA algorithm is higher than that of the GBPDA algorithm. The reason is that the EPBPDA algorithm is obtained by comparing two types of data sets, that is, Internet addiction and non-Internet addiction. It is more objective and fairer for the evaluation of Internet addiction. The GBPDA algorithm is only reflected by comparing with the Internet addiction generator, that is, it only judges whether the user behavior is close to the Internet addiction category, and the determination of the judgment threshold is also highly subjective.

The two algorithms are evaluated from two angles of missed detection rate and false detection rate respectively. The missed detection rate is to judge Internet addiction as non-Internet addiction, and the false detection rate is to judge non-Internet addiction as Internet addiction. Obviously, these two indicators will affect the detection effect of Internet addiction, so they can be unified as error rate. First of all, it can be seen from the graphs (b) and (c) in Figure 5 that the error rate is at a low level, indicating that the classification effect is good. Secondly, when the data size is small, the error rate is high, which just verifies the law reflected by the accuracy rate from another aspect; and when the data size increases, the error rate decreases to a certain extent, which proves once again that the Internet addiction behavior detection needs to collect as much user behavior data as possible. Next, the error rate of the EPBPDA algorithm is slightly lower than that of the GBPDA algorithm, indicating that the classification effect of the EPBPDA algorithm is better. .

Claims (6)

1. A net addiction detection device based on user computer interaction events is characterized in that: comprises a simple event acquisition module, a complex event generation module, a time generalization and normalization processing module, a frequent behavior generator acquisition module, a sample library construction module, a data processing and detection module and an early warning intervention module, wherein,

simple event collection module: the system comprises a complex event generating module, a user interaction module and a user interaction module, wherein the simple event generating module is used for acquiring interaction variables of the user and a computer, and the interaction variables comprise the CPU utilization rate, the memory utilization rate, the left mouse button clicking times in unit time, the right mouse button clicking times in unit time, the keyboard clicking times in unit time, the pixel moving times of the mouse in unit time, network flow and an operation process of the computer, and transmitting the input variables to the complex event generating module;

a complex event generation module: the system comprises a plurality of simple events, a plurality of complex events and a plurality of storage units, wherein the simple events are used for comparing each simple event with respective threshold, if the simple events are larger than or equal to the threshold range, the simple events are reserved, and then the complex events are generated according to the combination conditions of all the reserved simple events; otherwise, continuing to collect simple events; the complex events comprise offline video watching, online video watching, instant strategy games, table game games, webpage browsing, online chatting and downloading operations;

a time generalization and normalization processing module: the system is used for sorting the complex events made by each user on a certain day and generalizing the occurrence time of the complex events, namely dividing one day into a plurality of time periods; normalizing the duration, namely rounding the duration of the complex event; sending the processed complex event to a frequent behavior generation sub-acquisition module;

the frequent behavior generation sub-acquisition module: when a sample library is established, the method is used for acquiring the same complex events carried out among a plurality of network addiction users, collecting the generation times of the same complex events, selecting the complex events larger than a threshold value as frequent network addiction complex events according to a time threshold value set by the user, acquiring a generator for describing the behavior characteristic distribution of the user by a set formed by the frequent network addiction complex events, and sending the generator to a sample library construction module; the system is used for acquiring the same complex events carried out among a plurality of non-network-addiction users, collecting the generation times of the same complex events, selecting the complex events larger than a threshold value as frequent non-network-addiction complex events according to a time threshold value set by the users, then acquiring a generator which is a set formed by the frequent non-network-addiction complex events and used for describing the behavior characteristic distribution of the users, and sending the generator to a sample library construction module;

when the internet addiction condition of a user is detected, the method is used for acquiring the same complex event performed by the user to be detected within a period of time, acquiring the generation times of the same complex event, selecting the complex event larger than a threshold value as a frequent complex event according to a time threshold value set by the user, then acquiring a generator for describing behavior feature distribution of a set formed by the frequent complex event, and sending the generator to a data processing and detecting module;

a sample library construction module: the method is used for calculating scores of the frequent net addiction complex event generators by adopting a generator-based sample construction algorithm, if the scores exceed a threshold value, the frequent net addiction complex event generators are stored as net addiction classes, otherwise, the frequent net addiction complex event generators are deleted, and the frequent non-net addiction complex event generators are stored as non-net addiction classes; or calculating the frequent complex event producers by adopting a sample construction algorithm based on the exposure mode to obtain and store the producers of the exposure mode;

the data processing and detecting module: the network addiction judging module is used for comparing the to-be-detected frequent behavior generator with the network addiction sample library, judging the to-be-detected frequent complex event generator as network addiction if the to-be-detected frequent complex event generator is contained in the sample library, otherwise judging the to-be-detected frequent complex event generator as non-network addiction, and outputting a judgment result; or comparing the to-be-detected frequent behavior generator with a sample library, respectively calculating the net addiction score and the non-net addiction score by adopting a net addiction detection algorithm based on an exposure mode, and judging the category of the to-be-detected frequent behavior generator according to the score;

an early warning intervention module: the computer interaction behavior detection device is used for sending out an early warning prompt when the fact that the computer interaction behavior of the user is the internet addiction behavior is detected, and limiting the time range of the user for continuously using the computer through the timer for 1-2 hours; if the time limit range is exceeded, the computer is shut down.

2. The method for detecting the internet addiction based on the user computer interaction events, which is characterized by comprising the following steps of: the method comprises the following steps:

step 1, collecting interaction variables between a plurality of network addict users and a plurality of non-network addict users and a computer as simple events, wherein the interaction variables comprise the utilization rate of a computer CPU (Central processing Unit), the utilization rate of a memory, the times of clicking a left mouse button in unit time, the times of clicking a right mouse button in unit time, the times of clicking a keyboard in unit time, the number of pixels moved by the mouse in unit time, network flow and an operation process;

step 2, comparing each simple event with respective threshold value, if the simple event is larger than or equal to the threshold value range, reserving the simple event, and generating a complex event according to the combination condition of all reserved simple events; otherwise, continuing to collect simple events; the complex events comprise offline video watching, online video watching, instant strategy games, table game games, webpage browsing, online chatting and downloading operations;

the process of generating complex events is as follows:

if the CPU utilization rate exceeds or is equal to the threshold range of 45% -55%, the memory occupancy rate exceeds or is equal to the threshold range of 55% -65%, and the internet traffic exceeds or is equal to the threshold range of 35-45 MB, generating a complex event of watching videos online;

if the CPU utilization rate exceeds or is equal to 45% -55% of the threshold range, the memory occupancy rate exceeds or is equal to 55% -65% of the threshold range and the running process is a multimedia player, generating a complex event for watching videos offline;

if the CPU utilization rate exceeds or is equal to 45-55% of the threshold range, the memory occupancy rate exceeds or is equal to 55-65% of the threshold range, the left mouse button clicking frequency exceeds or is equal to 25-35 times of the threshold range in unit time, the right mouse button clicking frequency exceeds or is equal to 10-20 times of the threshold range in unit time, the number of pixels moved by the mouse exceeds or is equal to 1550-1650 times of the threshold range in unit time, the keyboard knocking frequency exceeds or is equal to 80-120 times of the threshold range in unit time, and the running process is a timely strategy game, a complex event for playing the instant strategy game is generated;

if the CPU utilization rate exceeds or is equal to 45-55% of the threshold range, the memory occupancy rate exceeds or is equal to 55-65% of the threshold range, the number of times of clicking the left mouse button per unit time exceeds or is equal to 25-35 times of the threshold range, the number of pixels moved by the mouse per unit time exceeds or is equal to 1550-1650 pixels, and the running process is a table game, a complex event for playing the table game is generated;

if the CPU utilization rate exceeds or equals to 45% -55% of the threshold range, the number of times of clicking the left button of the mouse per unit time exceeds or equals to 25-35 times of the threshold range, the number of pixels moved by the mouse per unit time exceeds or equals to 1550-1650 pixels of the threshold range, and the running process is a table game, generating another complex event for playing the table game;

if the number of times of clicking the left mouse button per unit time exceeds or equals to the threshold range of 25-35 times, the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range of 1550-1650 pixels, and the running process is a browser, generating a complex event for browsing a webpage;

if the number of times of clicking the left mouse button per unit time exceeds or equals to the threshold range by 25-35 times, the number of pixels moved by the mouse per unit time exceeds or equals to the threshold range by 1550-1650, the internet traffic exceeds or equals to the threshold range by 35-45 MB, and the running process is a browser, generating another complex event for browsing the webpage;

if the number of pixels moved by the mouse in unit time exceeds or is equal to 1550-1650 pixels in the threshold range, the internet traffic exceeds or is equal to 35-45 MB in the threshold range, and the running process is a browser, generating another complex event for browsing the webpage;

if the number of times of keyboard knocking exceeds or is equal to the threshold range of 80-120 times in unit time and the running process is instant chat software, generating a complex event of online chat;

if the number of pixels moved by the mouse in unit time exceeds or is equal to 1550-1650 pixels in the threshold range, the number of times of keyboard knocking in unit time exceeds or is equal to 80-120 times in the threshold range, and the running process is instant chat software, generating another complex event of online chatting;

if the internet traffic exceeds or equals to the threshold range of 35-45 MB, generating a complex event for downloading data;

step 3, generalizing the occurrence time of the complex event, namely dividing one day into five time periods of 6: 00-11: 00, 11: 00-14: 00, 14: 00-18: 00, 18: 00-23: 00 and 23: 00-6: 00, and normalizing the duration, namely rounding the duration of the complex event by taking 10 minutes as a unit;

step 4, acquiring the same complex events carried out among a plurality of network addiction users and the same complex events carried out among a plurality of non-network addiction users, collecting the generation times of the same complex events, selecting the complex events larger than a threshold value as frequent network addiction complex events and frequent non-network addiction complex events according to a time threshold value set by the user, then acquiring a plurality of generators for describing user behavior characteristic distribution in a set formed by the frequent network addiction complex events, acquiring a plurality of generators for describing user behavior characteristic distribution in a set formed by the frequent non-network addiction complex events, and sending the generators to a sample library construction module;

step 5, calculating scores of the generators of the frequent Internet addiction complex events by adopting a generator-based sample construction algorithm, if the scores exceed a threshold value, saving the generators as Internet addicts, otherwise, deleting the generators, and saving the generators of the frequent non-Internet addiction complex events as non-Internet addicts; or calculating the frequent complex event producers by adopting a sample acquisition algorithm based on the exposure mode to obtain and store the producers of the exposure mode;

step 6, collecting interaction variables between the user to be tested and the computer within a period of time, and repeating the step 2 to the step 3;

step 7, acquiring the same complex events performed by a user to be detected within a period of time, collecting the generation times of the same complex events, selecting the complex events larger than a threshold value as frequent complex events according to a time threshold value set by the user, acquiring a generator for describing behavior feature distribution of a set formed by the frequent complex events, and sending the generator to a data processing and detecting module;

step 8, comparing the frequent behavior generator to be detected with a network addiction sample library, if the frequent complex event generator to be detected is contained in the sample library, judging that the frequent behavior generator is network addiction, otherwise, judging that the frequent behavior generator is non-network addiction, and outputting a judgment result; or comparing the to-be-detected frequent behavior generator with a sample library, respectively calculating the net addiction score and the non-net addiction score by adopting a net addiction detection algorithm based on an exposure mode, and judging the scores of the net addiction score and the non-net addiction score, wherein the to-be-detected frequent complex events belong to the category with one higher score;

step 9, if the computer interaction behavior of the user is detected to be the internet addiction behavior, an early warning intervention module is adopted to send out an early warning prompt, and the time range within which the user can continue to use the computer is limited by a timer for 1-2 hours; if the time limit range is exceeded, the computer is shut down.

3. The method for detecting internet addiction detection device based on user computer interaction events as claimed in claim 2, wherein: respective thresholds for the simple events described in step 2: the utilization rate of a computer CPU ranges from 45% to 55%, the utilization rate of a memory ranges from 55% to 65%, the left button clicking frequency of a mouse in unit time ranges from 25 to 35 times, the right button clicking frequency of the mouse in unit time ranges from 10 to 20 times, the keyboard clicking frequency in unit time ranges from 80 to 120, the pixel moving frequency of the mouse in unit time ranges from 1550 to 1650, and the network flow ranges from 35 to 45 MB.

4. The method for detecting internet addiction detection device based on user computer interaction events as claimed in claim 2, wherein: step 5, calculating scores of the frequent net addiction complex event generators by adopting a generator-based sample construction algorithm, which specifically comprises the following steps:

step 5-1, determining a weight coefficient of a complex event type according to user requirements, ensuring that the weight coefficient of the instant strategy game is more than the weight coefficient of online watching video, more than the weight coefficient of a table game, more than the weight coefficient of browsing a webpage, more than the weight coefficient of offline watching video, more than the weight coefficient of online chatting, more than the downloaded weight coefficient, and the total weight of the weight coefficients is 1;

step 5-2, determining a weight coefficient of a time generalization section of the complex event according to user requirements, ensuring that the dawn weight coefficient is greater than the evening weight coefficient is greater than the noon weight coefficient is greater than the afternoon weight coefficient is greater than the morning weight coefficient, and the total number of the weight coefficients is 1;

step 5-3, multiplying each complex event type weight coefficient and the complex event time generalization section weight coefficient in the generator by the duration to obtain a score, and summing the scores of each complex event to obtain the score of the generator;

and 5-4, if the score exceeds a threshold value, keeping the threshold value range from 1 to 6, and otherwise, deleting the score.

5. The method for detecting internet addiction detection device based on user computer interaction events as claimed in claim 2, wherein: step 5, calculating the frequent complex event generators by adopting a sample construction algorithm based on the exposure mode, obtaining the generators of the exposure mode and storing the generators in a classified manner, specifically:

step 5-a, obtaining producers belonging to frequent net addiction complex events but not to frequent non-net addiction complex events;

step 5-b, acquiring producers which do not belong to the frequent Internet addiction complex events but belong to the frequent non-Internet addiction complex events;

step 5-c, obtaining a common producer belonging to both frequent and frequent non-network addiction complex events, determining the support degree, namely the occurrence rate, of the producer in the frequent and frequent network addiction complex events, and determining the support degree in the frequent and non-network addiction complex events;

step 5-d, calculating the growth rate of the joint producers in the frequent net addiction complex events, namely dividing the support degree of the joint producers in the frequent net addiction complex events by the support degree of the joint producers in the frequent non-net addiction complex events; calculating the growth rate of the joint product in the frequent non-network addiction complex events, namely dividing the support degree of the joint product in the frequent non-network addiction complex events by the support degree in the frequent network addiction complex events;

step 5-e, comparing the two growth rates, reserving the producers with the growth rates larger than 1, and deleting the producers with the growth rates smaller than 1;

and 5-f, merging the generation sub-sets in the frequent network addiction complex events, and merging the generation sub-sets in the frequent non-network addiction complex events, namely acquiring and storing the generation sub-sets in the exposure mode.

6. The method for detecting internet addiction detection device based on user computer interaction events as claimed in claim 2, wherein: extraction as described in step 8Respectively calculating net addiction score and non-net addiction score with net addiction detection algorithm based on exposure mode, and scoring score (C, D)_i) The formula is as follows:

$\mathrm{Score}(C,D_{i|j})=\underset{X\&Element;\mathrm{RS}}{\mathrm{\&Sigma;}}\frac{\mathrm{Gr}(X,D_{j|i},D_{i|j})}{\mathrm{Gr}(X,D_{j|i},D_{i|j})+\mathrm{Max}{\mathrm{Gr}}_{\mathrm{RS}}}\&times;{\mathrm{Sup}}_c\left(X\right)$

$+\underset{Y\&Element;\mathrm{WS}}{\mathrm{\&Sigma;}}\frac{1}{\mathrm{Gr}(Y,D_{i|j},D_{j|i})+\mathrm{Min}{\mathrm{Gr}}_{\mathrm{WS}}}\&times;{\mathrm{Sup}}_c\left(Y\right),(i\&NotEqual;j)---\left(2\right)$

wherein: d_i|jRepresents D_iSample library or D_jSample library, D_j|iRepresents D_jSample library or D_iSample library, i and j are sample library class tag numbers, X denotes D_iY represents D_jThe generator of the exposure mode of (1); RS is the forward sample set, i.e. the actively acting set, of the target class, WS is the reverse sample set, i.e. the negatively acting set, of the target class; gr (X, D)_j|i,D_i|j) Represents the growth rate of X from one class to another; MaxGr_RSRepresenting a maximum growth rate threshold in the set of target classes, RS representing a target classSet of forward samples, MinGr_WSRepresenting the minimum growth rate threshold value in the non-target class set, WS representing the reverse sample set of the target class, C representing the complex event to be classified, Sup_c(X) represents the support of X in C, Sup_c(Y) represents the degree of support of Y in C.

Images