CN110059189B - Game platform message classification system and method - Google Patents
Game platform message classification system and method Download PDFInfo
- Publication number
- CN110059189B CN110059189B CN201910289668.5A CN201910289668A CN110059189B CN 110059189 B CN110059189 B CN 110059189B CN 201910289668 A CN201910289668 A CN 201910289668A CN 110059189 B CN110059189 B CN 110059189B
- Authority
- CN
- China
- Prior art keywords
- message
- messages
- user
- junk
- classification
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/353—Clustering; Classification into predefined classes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Information Transfer Between Computers (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a message classification system and a method of a game platform. The junk message discovery module processes the junk messages through a PU learning algorithm to obtain sample messages and stores the sample messages into a sample library; the model training module uses the sample information of the sample library and trains a similarity classification model or a character-level CNN text classification model according to the online period of the game; the real-time processing module calls a classification model of the model training module to judge the user message, secondary judgment is carried out through an LP algorithm and keyword analysis, and the result is fed back to the junk message classification module. Therefore, the PU learning algorithm can be trained by realizing a small amount of samples, and the problem that the cold start process lacks samples is solved; the recall rate and the compatibility are improved through a character-level CNN text classification model; real-time defense is realized through an LP algorithm and keyword analysis, a new sample message is obtained at the same time to construct a similarity model, and the problem of cold start is solved.
Description
Technical Field
The invention relates to the field of data processing, in particular to a game platform message classification system and a game platform message classification method.
Background
With the gradual development of the game industry, the situations that the chat interface in the online game is filled with spam messages are more and more, how to classify the messages of the game platform (automatically filter out the spam messages) and prevent the user from being flushed by malicious advertisements, 35881, and abuse and even fraud information fraud instead are the problems faced by each game company. The traditional solutions are mainly as follows according to the differences of technical strength and size of different game companies:
1. and (3) keyword shielding: setting keywords corresponding to the junk messages, identifying and classifying the messages of the users, and sealing the junk messages after triggering;
2. manual number sealing: operators search the junk messages to carry out number sealing operation;
3. and (3) increasing the speaking difficulty: such as adjusting the level of possible speech, prop speech, etc.;
4. constructing a supervision type machine learning model: and collecting samples to train a message classification model (such as a Bayesian model, an SVM model and other supervised learning models).
Most game companies can build a game chat message classification system by combining the above modes, operators are primarily used for carrying out prohibition and garbage message collection (optionally increasing speaking difficulty), then keywords are extracted to fill a blacklist for shielding, and companies with certain technical conditions can select a supervised learning algorithm to build a message classification model, so that classification shielding of garbage messages to a certain degree is achieved.
However, the keyword masking can only mask existing keywords (collected manually by a manager), but malicious players can change the expression mode of spam on a targeted basis after finding, and game companies can only passively add blacklists, so that real-time defense is difficult to perform. The machine learning model has certain learning ability to prevent malicious players from changing the expression mode of the spam message, but the problem of real-time defense cannot be completely solved by doing so, because the effect of the supervised machine learning model depends on a sample set and subsequent feature extraction, sample collection and feature engineering are needed, and how to label the data set is a problem to be faced (the collection of data set labeling consumes manpower extremely). If a game just launches to the market, the new spam message is considered as stranded under the condition of no sample, firstly, the sample is not enough to train a new model, the real-time performance of the blacklist mode is too poor, and the problem of cold start needs to be solved urgently. In addition, the existing classification system is not very scalable, for example, some large-scale game companies have money games under flags, or one game releases multiple language versions, and for the case of multiple languages in the game, multiple sets of suitable classification systems may need to be constructed.
In summary, the current game platform message classification method has the following big defects:
(ii) sample collection problem: supervised learning algorithms rely on a large number of samples;
secondly, the problem of unsatisfactory classification effect: the existing model scheme has not ideal effect;
(iii) compatibility issues: a set of classification schemes suitable for any game and language is required;
fourthly, the real-time defense problem of the new type of spam message: newly appearing spam messages need to be identified;
game cold start problem: how to perform the classification of the game platform messages in time at the beginning stage of the game.
In view of the above, the present inventors have made extensive studies and research efforts to develop and design the present invention in view of the disadvantages and inconveniences caused by the above-mentioned problems.
Disclosure of Invention
The invention aims to provide a message classification system and a message classification method for a game platform, which solve the defects of the existing message classification system, have the characteristics of training with a small number of samples, high recall rate and strong compatibility, realize real-time defense and solve the problem of cold start.
In order to achieve the above purpose, the solution of the invention is:
a message classification system of a game platform comprises a junk message discovery module, a sample library, a model training module and a real-time processing module; the junk message discovery module acquires junk messages of the game platform through the customer service platform, the data warehouse and the user reporting and real-time processing module; the junk message classification module processes the acquired junk messages through a PU learning algorithm to obtain sample messages and stores the sample messages into a sample library; the model training module trains a classification model according to the online period of the game by using the sample information of the sample library, wherein the classification model comprises a similarity classification model and a character-level CNN text classification model; at the initial stage of online game, the model training module trains a similarity classification model; in the middle and later online stages of the game, a model training module trains a character-level CNN text classification model; the real-time processing module calls a classification model of the model training module to judge normal or junk messages of the user messages and feeds the user messages judged to be the junk messages back to the junk message finding module; the real-time processing module is provided with a blacklist, and special words of the blacklist are set by a manager of the game platform; the real-time processing module carries out secondary judgment on the user message judged to be the normal message through LP algorithm and keyword analysis, the user message with abnormal keywords is marked as junk message through the keyword analysis, and the type of the user message with abnormal keywords is marked through LP algorithm.
The customer service platform stores the junk messages collected by the customer service personnel; the data warehouse stores all user messages stored in the game platform; the user provides spam messages for the message classification system in the form of user reports during the game process.
The similarity classification model is used for calculating the distance between the training sample messages for classification.
The character-level CNN text classification model consists of an input layer, 6 convolutional layers, 3 full-connection layers and an output layer; the input layer receives user information, and inputs the character-level CNN text classification model after one-hot coding is carried out on the user information according to the size of an alphabet, the CNN text classification model carries out word embedding conversion on the coded characters into tensors with the shapes of (None, 1014, 128), and then data are input into the convolutional layer; performing one-dimensional convolution on the data by each convolution layer, then pooling to finally obtain a tensor with the shape of (None, 34, 256), and inputting the data into the full-connection layer; the fully-connected layers convert the data into tensors (None, 8704), and a dropout layer is arranged between every two adjacent fully-connected layers for model regularization; and the output layer outputs the probability of the category to which the user message belongs through softmax, and gives a classification result.
A message classification method of a game platform, which uses the message classification system of the game platform, comprises the following steps:
step 2, the real-time processing module calls a classification model of the model training module to classify the user messages, if the user messages are spam messages, the result of the spam messages is output, and the user message processing is finished; if the message is normal, turning to the step 3;
step 3, the real-time processing module judges the user message through the LP algorithm and keyword analysis, judges whether the keyword is abnormal, if not, outputs the result of the normal message, and the user message is processed; if yes, marking the user message as a junk message through keyword analysis, marking the type of the user message through an LP algorithm, and feeding back the result to a junk message discovery module;
step 4, the junk message discovery module processes the user message and the type thereof through a PU learning algorithm to acquire a new sample message and update a sample library; the model training module trains and updates the classification model using the sample messages of the sample library.
After the system and the method are adopted, the PU learning algorithm is utilized, training can be carried out only by a small amount of samples, and the problem that the cold start process lacks samples is solved; based on the word component classification model, the recall rate is greatly improved through the character-level CNN text classification model, and the method can be compatible with a multi-game or multi-language game platform; by means of the LP algorithm and keyword analysis, real-time defense is achieved on junk messages which cannot be identified by the model and newly-appeared junk messages, sample messages are obtained in the real-time defense process, then a similarity model is constructed for defense, and therefore the cold start problem is solved.
Drawings
FIG. 1 is a schematic diagram of a system architecture according to the present invention;
FIG. 2 is a schematic structural diagram of a character-level CNN text classification model;
fig. 3 is a schematic diagram of a process for processing a user message according to the present invention.
Detailed Description
In order to further explain the technical solution of the present invention, the present invention is explained in detail by the following specific examples.
As shown in FIG. 1, the invention is a message classification system of a game platform, comprising a spam discovery module, a sample library, a model training module and a real-time processing module.
The junk message discovery module acquires junk messages of the game platform through a customer service platform, a data warehouse, a user reporting and real-time processing module and the like; and the junk message classification module processes the acquired junk messages through a PU learning algorithm to obtain sample messages and stores the sample messages into a sample library. The customer service platform stores the junk messages collected by the customer service personnel; the data warehouse stores all user messages stored in the game platform and can extract junk messages from the user messages; the user provides spam messages for the message classification system in the form of user reports during the game process.
The model training module trains a classification model according to the online period of the game by using the sample information of the sample library, wherein the classification model comprises a similarity classification model and a character-level CNN text classification model. At the initial stage of online game, the model training module trains a similarity classification model, and the similarity classification model is used for calculating the distance between training sample messages for classification; and in the middle and later online stages of the game, the model training module trains a character-level CNN text classification model.
At the initial stage of online game: the junk messages in the game are few, only a few sample messages exist, and the distances among the sample messages can be calculated by utilizing the similarity to classify. For any sample message, it can be represented as a vector
For any one user message, can be expressed as
Then, according to the cosine similarity, the distance between any one user message and any one sample message can be obtained as follows:
whether the user input is spam or not can be judged according to whether the distance is larger than the threshold value, so that a simple and reliable spam classification model can be generated in a short time to temporarily deal with the problem of initial spam of the game.
And in the later stage of online game: with the increasing of spam messages and the increasing of the number of sample messages, a spam message discovery module is combined, a training sample set is used for training a character-level CNN text classification model, the structural schematic diagram of the character-level CNN text classification model is shown in FIG. 2, the character-level CNN text classification model comprises an input layer, 6 convolutional layers, 3 full-link layers and an output layer, at the moment, an input space is a set of all user messages, an output space is a set of spam messages and normal messages, and each user message is a message example. The input layer receives the user information, carries out one-hot coding on the user information according to the size of the alphabet, then inputs a character-level CNN text classification model, carries out word embedding conversion on the coded characters into tensors with the shapes of (None, 1014, 128) by the CNN text classification model, and then inputs data into the convolutional layer; performing one-dimensional convolution on the data by each convolution layer, then pooling to finally obtain a tensor with the shape of (None, 34, 256), and inputting the data into a full connection layer; the fully-connected layers convert the data into tensors with the shapes of (None, 8704), and a dropout layer is arranged between every two adjacent fully-connected layers for model regularization; and the output layer outputs the probability of the category to which the user message belongs through softmax, and gives a classification result. The details of each step of the model are given in the following table:
for example, the steps for classifying the message "fill 100 with 500vip top up" using the character level CNN text classification model are as follows:
1. the message is one-hot encoded and converted into the following vector:
2. the output obtained by the convolution layer and the full connection layer is as follows: [ [ 0.659829850.34017012 ] ], wherein 0.65982985 represents the probability of being a spam message and 0.34017012 represents the probability of being a normal message.
3. The output message is an advertising alert such as: "message: the 100 charge is sent 500vip charges are judged by the model as an advertisement.
By utilizing the character-level CNN text classification model, the text characteristics can be automatically extracted by the neural network in the training process, so that the system has extremely strong compatibility, no matter which game or which language can have a good effect, and the recall rate is greatly improved compared with the traditional and learned classification models.
The real-time processing module calls the classification model of the model training module to judge normal messages or junk messages of the user messages of the game platform, and feeds the user messages judged as the junk messages back to the junk message finding module, so that the junk messages in most real-time chatting can be resisted in real time. The real-time processing module is provided with a blacklist, and special words of the blacklist are set by a manager of the game platform so as to prevent the possible special words from bypassing the classification model. The real-time processing module carries out secondary judgment on the user message judged to be the normal message through LP algorithm and keyword analysis, the user message with abnormal keywords is marked as junk message through the keyword analysis, and the type of the user message with abnormal keywords is marked through LP algorithm.
When a novel junk message bypasses the classification model and the blacklist (namely is misjudged as a normal message), the real-time processing module can automatically mark the message as the junk message through keyword analysis at the moment, and then automatically mark the type of the message through an LP algorithm, so that the novel junk message is defended.
As shown in fig. 3, based on the above system, a message classification method for a game platform includes the following steps:
step 2, the real-time processing module calls a classification model of the model training module to classify the user messages, if the user messages are spam messages, the result of the spam messages is output, and the user message processing is finished; if the message is normal, turning to the step 3;
step 3, the real-time processing module judges the user message through the LP algorithm and keyword analysis, judges whether the keyword is abnormal, if not, outputs the result of the normal message, and the user message is processed; if yes, marking the user message as a junk message through keyword analysis, marking the type of the user message through an LP algorithm, and feeding back the result to a junk message discovery module;
step 4, the junk message discovery module processes the user message and the type thereof through a PU learning algorithm to acquire a new sample message and update a sample library; the model training module trains and updates the classification model using the sample messages of the sample library.
By the system and the method, the training can be performed by using a PU learning algorithm with only a small amount of samples, and the problem that the cold start process lacks samples is solved; based on the word component classification model, the recall rate is greatly improved through the character-level CNN text classification model, and the method can be compatible with a multi-game or multi-language game platform; by means of the LP algorithm and keyword analysis, real-time defense is achieved on junk messages which cannot be identified by the model and newly-appeared junk messages, sample messages are obtained in the real-time defense process, then a similarity model is constructed for defense, and therefore the cold start problem is solved.
The above embodiments and drawings are not intended to limit the form and style of the present invention, and any suitable changes or modifications thereof by those skilled in the art should be considered as not departing from the scope of the present invention.
Claims (5)
1. A message classification system for a gaming platform, comprising: the system comprises a junk message discovery module, a sample library, a model training module and a real-time processing module; the junk message discovery module acquires junk messages of the game platform through the customer service platform, the data warehouse and the user reporting and real-time processing module; the junk message classification module processes the acquired junk messages through a PU learning algorithm to obtain sample messages and stores the sample messages into a sample library;
the model training module trains a classification model according to the online period of the game by using the sample information of the sample library, wherein the classification model comprises a similarity classification model and a character-level CNN text classification model; at the initial stage of online game, the model training module trains a similarity classification model; in the middle and later online stages of the game, a model training module trains a character-level CNN text classification model;
the real-time processing module calls a classification model of the model training module to judge normal or junk messages of the user messages and feeds the user messages judged to be the junk messages back to the junk message finding module; the real-time processing module is provided with a blacklist, and special words of the blacklist are set by a manager of the game platform; the real-time processing module carries out secondary judgment on the user message judged to be the normal message through LP algorithm and keyword analysis, the user message with abnormal keywords is marked as junk message through the keyword analysis, and the type of the user message with abnormal keywords is marked through LP algorithm.
2. The message classification system of a gaming platform of claim 1, wherein: the customer service platform stores the junk messages collected by the customer service personnel; the data warehouse stores all user messages stored in the game platform; the user provides spam messages for the message classification system in the form of user reports during the game process.
3. The message classification system of a gaming platform of claim 1, wherein: the similarity classification model is used for calculating the distance between the training sample messages for classification.
4. The message classification system of a gaming platform of claim 1, wherein: the character-level CNN text classification model comprises an input layer, 6 convolutional layers, 3 full-connection layers and an output layer, wherein the input layer receives user messages, carries out one-hot coding on the user messages according to the size of an alphabet and then inputs the user messages into the character-level CNN text classification model, the CNN text classification model carries out word embedding conversion on the coded characters into tensors with the shapes of (None, 1014, 128), and then data are input into the convolutional layers; performing one-dimensional convolution on the data by each convolution layer, then pooling to finally obtain a tensor with the shape of (None, 34, 256), and inputting the data into the full-connection layer; the fully-connected layers convert the data into tensors (None, 8704), and a dropout layer is arranged between every two adjacent fully-connected layers for model regularization; and the output layer outputs the probability of the category to which the user message belongs through softmax, and gives a classification result.
5. A message classification method for a game platform using the message classification system for a game platform according to any one of claims 1 to 4, comprising the steps of:
step 1, a system receives real-time user information, judges whether a classification model exists or not, and if yes, turns to step 2; if not, turning to the step 3;
step 2, the real-time processing module calls a classification model of the model training module to classify the user messages, if the user messages are spam messages, the result of the spam messages is output, and the user message processing is finished; if the message is normal, turning to the step 3;
step 3, the real-time processing module judges the user message through the LP algorithm and keyword analysis, judges whether the keyword is abnormal, if not, outputs the result of the normal message, and the user message is processed; if yes, marking the user message as a junk message through keyword analysis, marking the type of the user message through an LP algorithm, and feeding back the result to a junk message discovery module;
step 4, the junk message discovery module processes the user message and the type thereof through a PU learning algorithm to acquire a new sample message and update a sample library; the model training module trains and updates the classification model using the sample messages of the sample library.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910289668.5A CN110059189B (en) | 2019-04-11 | 2019-04-11 | Game platform message classification system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910289668.5A CN110059189B (en) | 2019-04-11 | 2019-04-11 | Game platform message classification system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110059189A CN110059189A (en) | 2019-07-26 |
| CN110059189B true CN110059189B (en) | 2021-03-05 |
Family
ID=67317608
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910289668.5A Active CN110059189B (en) | 2019-04-11 | 2019-04-11 | Game platform message classification system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110059189B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112446476A (en) * | 2019-09-04 | 2021-03-05 | 华为技术有限公司 | Neural network model compression method, device, storage medium and chip |
| CN111198947B (en) * | 2020-01-06 | 2024-02-13 | 南京中新赛克科技有限责任公司 | Convolutional neural network fraud short message classification method and system based on naive Bayes optimization |
Family Cites Families (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8396927B2 (en) * | 2004-12-21 | 2013-03-12 | Alcatel Lucent | Detection of unwanted messages (spam) |
| CN101106748A (en) * | 2006-07-11 | 2008-01-16 | 华为技术有限公司 | A content filtering system, device and method for mobile network |
| CN101317376B (en) * | 2006-07-11 | 2011-04-20 | 华为技术有限公司 | Method, device and system for contents filtering |
| US7836061B1 (en) * | 2007-12-29 | 2010-11-16 | Kaspersky Lab, Zao | Method and system for classifying electronic text messages and spam messages |
| CN101516071B (en) * | 2008-02-18 | 2013-01-23 | 中国移动通信集团重庆有限公司 | Method for classifying junk short messages |
| CN101415161A (en) * | 2008-11-26 | 2009-04-22 | 深圳华为通信技术有限公司 | Method and apparatus for filtrating rubbish short message |
| CN103312585B (en) * | 2012-03-08 | 2016-12-28 | 中兴通讯股份有限公司 | A kind of rubbish message processing method and system |
| US10621276B2 (en) * | 2013-10-24 | 2020-04-14 | Wmware, Inc. | User interface virtualization for web applications |
| CN103634473B (en) * | 2013-12-05 | 2016-03-23 | 南京理工大学连云港研究院 | Based on mobile phone method for filtering spam short messages and the system of Naive Bayes Classification |
| US20160232231A1 (en) * | 2015-02-11 | 2016-08-11 | Hung Dang Viet | System and method for document and/or message document and/or message content suggestion, user rating and user reward |
| CN105389379B (en) * | 2015-11-20 | 2018-12-04 | 重庆邮电大学 | A kind of rubbish contribution classification method indicated based on text distributed nature |
| CN105488029A (en) * | 2015-11-30 | 2016-04-13 | 西安闻泰电子科技有限公司 | KNN based evidence taking method for instant communication tool of intelligent mobile phone |
| US10171403B2 (en) * | 2015-11-30 | 2019-01-01 | International Business Machines Corporation | Determining intended electronic message recipients via linguistic profiles |
| CN106937262B (en) * | 2015-12-29 | 2020-03-13 | 中国移动通信集团公司 | Message processing method, device and system |
| US10765956B2 (en) * | 2016-01-07 | 2020-09-08 | Machine Zone Inc. | Named entity recognition on chat data |
| CN106202177B (en) * | 2016-06-27 | 2017-12-15 | 腾讯科技（深圳）有限公司 | A kind of file classification method and device |
| US10346448B2 (en) * | 2016-07-13 | 2019-07-09 | Google Llc | System and method for classifying an alphanumeric candidate identified in an email message |
| WO2019040279A1 (en) * | 2017-08-22 | 2019-02-28 | Gali Health, Inc. | Personalized digital health system using temporal models |
-
2019
- 2019-04-11 CN CN201910289668.5A patent/CN110059189B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110059189A (en) | 2019-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107515873B (en) | Junk information identification method and equipment | |
| CN109977416B (en) | Multi-level natural language anti-spam text method and system | |
| CN108009249B (en) | Spam comment filtering method for unbalanced data and fusing user behavior rules | |
| CN110704715B (en) | Network overlord ice detection method and system | |
| CN101184259A (en) | Keyword automatic learning and updating method in spam messages | |
| CN104317891B (en) | A kind of method and device that label is marked to the page | |
| CN114092742B (en) | Multi-angle-based small sample image classification device and method | |
| CN111177367B (en) | Case classification method, classification model training method and related products | |
| CN113849598B (en) | Social media false information detection method and detection system based on deep learning | |
| CN107169106A (en) | Video retrieval method, device, storage medium and processor | |
| CN115982350A (en) | False news detection method based on multi-mode Transformer | |
| CN111507350A (en) | Text recognition method and device | |
| CN114896305A (en) | Smart internet security platform based on big data technology | |
| CN110059189B (en) | Game platform message classification system and method | |
| CN109299251A (en) | A method and system for identifying abnormal spam short messages based on deep learning algorithm | |
| CN107357895A (en) | A kind of processing method of the text representation based on bag of words | |
| CN105224955A (en) | Based on the method for microblogging large data acquisition network service state | |
| CN107832307B (en) | Chinese word segmentation method based on undirected graph and single-layer neural network | |
| CN110110087A (en) | A kind of Feature Engineering method for Law Text classification based on two classifiers | |
| CN116739408A (en) | Power grid dispatching safety monitoring method and system based on data tag and electronic equipment | |
| CN115186095A (en) | Juvenile text recognition method and device | |
| CN113537272B (en) | Deep learning-based semi-supervised social network abnormal account detection method | |
| CN101329668A (en) | A method and device for generating information rules, and a method and system for judging information types | |
| CN109783807A (en) | A kind of user comment method for digging for APP software defect | |
| CN110362828B (en) | Network information risk identification method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |