CN110457597A - Advertisement recognition method and device - Google Patents

Abstract

The embodiment of the present application provides an advertisement identification method and device, which can use a classification model containing classification expressions to obtain suspected advertisement information from media information released by users; generate the suspected advertisement information according to at least one preset weight factor The weight factor includes the proportion of the length of the suspected advertisement information to the full text of the media information, the proportion of the advertisement information released by the user to all the media information released by the user, and the pictures in the suspected advertisement information One or more of the numbers; determining the suspected advertisement information whose advertisement weight is greater than a preset threshold as advertisement information. The method provided in the embodiment of the present application uses a classification model and advertisement weights to screen media information at two levels, first determines suspected advertisement information, and then determines whether the suspected advertisement information is an advertisement according to the advertisement weight, thereby accurately identifying advertisements and solving the problem of the prior art Problems with advertising within social platforms cannot be identified in a timely and effective manner.

Description

Advertisement recognition method and device

technical field

The present application relates to the technical field of natural language processing, and in particular to an advertisement recognition method and device.

Background technique

Various social platforms, such as Weibo, WeChat, Tieba and forums, are often targeted by advertisers. Advertisers place advertisements by registering a large number of accounts on various social platforms, and generating a large number of advertisement messages, replies, etc., resulting in a large amount of advertisement content interspersed in the normal content of the social platform, reducing the content quality of the social platform, making users in Passive browsing of advertising information when browsing content on social platforms affects user experience.

At present, in order to manage all kinds of advertisements appearing on social platforms, the managers or operators of social platforms usually set up some accounts with audit authority, and the personnel holding these accounts will find out the advertisements by manual inspection and conduct inspections. delete. However, advertisers usually use software to automatically place advertisements in order to increase the amount of advertisements, and the amount of advertisements is huge, which makes it difficult to effectively and timely contain these advertisements by manually inspecting and removing advertisements. Therefore, the problem of advertising on social platforms has not been effectively resolved.

Contents of the invention

The embodiment of the present application provides an advertisement identification method and device to solve the problem that the prior art cannot identify advertisements on social platforms in a timely and effective manner.

In the first aspect, an embodiment of the present application provides an advertisement identification method, including: using a classification model containing classification expressions to obtain suspected advertisement information from media information released by users; generating the advertisement information according to at least one preset weight factor. The advertisement weight of the suspected advertisement information; the weight factors include the proportion of the length of the suspected advertisement information to the full text of the media information, the proportion of the advertisement information released by the user to all the media information released by the user, and the proportion of the suspected advertisement information One or more of the number of pictures in ; determining the suspected advertisement information whose advertisement weight is greater than a preset threshold as advertisement information.

In the second aspect, an embodiment of the present application provides an advertisement recognition device, including: an information acquisition module, configured to obtain suspected advertisement information from media information released by users by using a classification model containing classification expressions; a weight generation module, configured to The advertisement weight of the suspected advertisement information is generated according to at least one preset weight factor; the weight factor includes the proportion of the suspected advertisement information to the length of the full text of the media information, and the proportion of the advertisement information released by the user to the length of the published advertisement information by the user The proportion of all media information, and one or more of the number of pictures in the suspected advertisement information; an advertisement determination module, configured to determine the suspected advertisement information whose advertisement weight is greater than a preset threshold as advertisement information.

It can be seen from the above technical solutions that the embodiments of the present application provide an advertisement identification method and device, which can use a classification model containing classification expressions to obtain suspected advertisement information from media information released by users; according to at least one preset weight factor Generate the advertisement weight of the suspected advertisement information; the weight factors include the proportion of the length of the suspected advertisement information to the full text of the media information, the proportion of the advertisement information released by the user to all the media information released by the user, and the One or more of the number of pictures in the suspected advertisement information; determining the suspected advertisement information whose advertisement weight is greater than a preset threshold as advertisement information. The technical solution provided by the embodiment of this application uses the classification model and advertisement weight to perform two-level screening of media information, first determines the suspected advertisement information, and then determines based on factors such as the character length, content, and user behavior corresponding to the suspected advertisement information. Its advertisement weight, and determine whether the suspected advertisement information is an advertisement according to the advertisement weight, so as to accurately identify the advertisement, and solve the problem that the existing technology cannot timely and effectively identify the advertisement in the social platform.

Description of drawings

In order to illustrate the technical solution of the present application more clearly, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, on the premise of not paying creative labor, Additional drawings can also be derived from these drawings.

FIG. 1 is a flow chart of an advertisement identification method provided by an embodiment of the present application;

Fig. 2 is a flow chart of obtaining the length of suspected advertisement information provided by the embodiment of the present application;

Fig. 3 is a flow chart of providing advertisement weights for generating suspicious advertisement information according to an embodiment of the present application;

FIG. 4 is a flow chart for managing users who publish advertisements provided by the embodiment of the present application;

Fig. 5 is a schematic structural diagram of an advertisement recognition device provided by an embodiment of the present application.

Detailed ways

In order to enable those skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described The embodiments are only some of the embodiments of the present application, but not all of them. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

Various social platforms, such as Weibo, WeChat, Tieba and forums, are often targeted by advertisers. Advertisers place advertisements by registering a large number of accounts on various social platforms, and generating a large number of advertisement messages, replies, etc., resulting in a large amount of advertisement content interspersed in the normal content of the social platform, reducing the content quality of the social platform, making users in Passive browsing of advertising information when browsing content on social platforms affects user experience.

In the prior art, there are at least two forms of advertisements in webpages, one is text advertisements, picture advertisements and pop-up advertisements placed on designated positions in webpages by website service providers, advertisers or various advertising alliances. Blocking is done by filtering web page elements; the other is that on various social platforms, such as Weibo, WeChat, Tieba and forums, advertisers use these platform accounts to post, leave messages, and reply. The advertisement itself is part of the content of the social platform, so it cannot be blocked by the above-mentioned webpage element filtering method.

In this regard, in order to manage all kinds of advertisements that appear on social platforms, managers or operators of social platforms usually set up some accounts with audit authority, and the personnel holding these accounts will find out the advertisements by manual inspection and check them out. to delete. However, advertisers usually use software to automatically place advertisements in order to increase the amount of advertisements, and the amount of advertisements is huge, which makes it difficult to effectively and timely contain these advertisements by manually inspecting and removing advertisements. Therefore, the problem of advertising on social platforms has not been effectively resolved.

The embodiment of the present application provides an advertisement identification method and device to solve the problem that the prior art cannot identify advertisements on social platforms in a timely and effective manner.

The following are method embodiments of the present application.

The method embodiment of the present application provides an advertisement identification method. Fig. 1 is a flow chart of the advertisement identification method. The method can be applied to multiple devices such as servers, PCs (personal computers), tablet computers, and mobile phones.

As shown in Figure 1, the method includes the following steps:

Step S101, using a classification model including classification expressions to obtain suspected advertisement information from media information released by users.

Structurally, the classification model used in the embodiment of the present application includes: at least one classification node, each classification node corresponds to an advertisement category, and each advertisement category corresponds to a category weight; each classification node includes at least one classification expression, The classification expression is used to identify the suspected advertisement information from the media information.

In terms of content, the classification model used in the embodiment of this application consists of three parts: ontology, elements and concepts. Among them, the business expressions that may appear in the advertisement description are constructed into different elements according to their content and semantics, and each element corresponds to a business classification; the general language expressions that may appear in the advertisement description are constructed into different elements according to their content and semantics. Each concept contains at least one concept expression expressing the same semantics; at least one concept expression forms a classification expression by combining various operators such as "and", "or", "not", "distance", "order"; at least A business classification and at least one classification expression in the business classification are integrated to form the ontology of the model.

Table 1 shows the structure of the classification model.

Table 1

As shown in Table 1, the business classification corresponding to the elements includes "high-weight advertisement identification", and multiple classification expressions are included under the "high-weight advertisement identification" category, and the classification expression "c_promoting URL+{0, 30} c_try or communicate" as an example, it includes the concepts of "c_sale URL" and "c_try or communicate", and the AND operator "+" and "{0,30}" A distance operator.

Further, each concept can be provided with a corresponding concept tree, which includes at least one concept expression, which is used to match the content corresponding to the concept from the corpus, for example: the concept of "marketing speech" can include, for example Conceptual expressions such as "gentle.*skin-friendly" and "skin-friendly and rejuvenating" (wherein, "." and "*" are operators, which can be written according to the rules of regular expressions). Therefore, the classification expression "c_sales URL+{0,30}c_attempt or communication" can match: match "c_sales URL" and also match "c _try or communicate" corpus content.

In the embodiment of the present application, the classification node may include multiple advertisement categories, such as "high-weight advertisement identification" and "low-weight advertisement category" in Table 1 (not shown in Table 1). For different advertisement categories, different category weights are set correspondingly in the embodiment of the present application, for example: set the category weight W 1 of "high weight advertisement identification" to 2, and set the category weight W ₁ of "low weight advertisement category" to ₁ .

In the embodiment of the present application, the above classification model is used to perform content matching on the media information posted by the user. If the media information matches an expression in the classification model, the media information is suspected advertisement information. The classification model can divide the suspected advertisement information into "high weight advertisement information" and "low weight advertisement information" according to the preset "advertisement category". , freight information, etc.; "low-weight advertisement information" refers to suspected advertisement information that only contains salesmanship.

For example, high-weight advertisement information:

Follow me, select stocks efficiently in 3 minutes, and benefit endlessly for a lifetime. QQ: XXXXXX, WeChat: XXXXXX

For example, low-weight advertisement information:

Gentle and clean, skin-friendly and rejuvenating, without hurting the stratum corneum, it is a rare good cleanser

Before obtaining the suspected advertisement information from the media information, the embodiment of the present application further preprocesses the media information, so as to improve the accuracy of obtaining the suspected advertisement information. The preprocessing process includes removing specific characters from the media information, performing character conversion on the media information, and performing one or more of converting Chinese characters to numbers on the media information.

For example, some advertisement publishers will deliberately use some special characters to process the advertisement content in order to evade review. Therefore, the preprocessing process must first remove the special characters, for example:

If you want to change your current situation, you might as well take out your mobile phone and add [WeChat] [Letter] abc111111 for consultation!

After removing special characters:

If you want to change your current situation, you might as well take out your mobile phone and add WeChat abc111111 to consult and consult!

In addition to special characters such as "【" and "】" shown in the above examples, special characters that can be removed also include: ! ~#$^&<>[]{}()*? /,.[], etc., which will not be repeated in the embodiments of this application.

After the special characters are removed, the next step is to perform character conversion on the media information, such as converting wide-body characters to narrow-body characters, converting traditional characters to simplified characters, converting uppercase letters to lowercase letters, and so on.

Take the conversion of traditional Chinese characters to simplified Chinese characters as an example:

Before conversion: Part-time job at home, daily settlement, hourly hundred yuan

After conversion: part-time job at home, daily settlement, 100 yuan per hour

After converting the characters, the next step is to normalize the numbers, specifically to convert the numbers expressed by Chinese characters or homophonic characters into Arabic numerals, for example:

Chongqing Iressa Purchasing_Iressa Price_India Iressa Purchasing Service Hotline: 139225XXXXX

translates to:

Chongqing Iressa Purchasing_Iressa Price_India Iressa Purchasing Service Hotline: 139225XXXXX

After the above preprocessing, the media information becomes relatively regular, which is convenient for identification using classification expressions, thereby improving the accuracy of extracting suspected advertisement information.

Step S102, generating the advertisement weight of the suspected advertisement information according to at least one preset weight factor; the weight factor includes the proportion of the suspected advertisement information to the length of the full text of the media information, the proportion of the advertisement information published by the user to the user One or more of the proportion of all published media information, and the number of pictures in the suspected advertisement information.

In the embodiment of this application, at least three weighting factors are determined, including:

1. The proportion of suspected advertising information in the length of the full text of media information. For example, if a user posts 50 characters in a post on a forum, 20 of which are identified as suspected advertising information, then, in this reply, the proportion of suspected advertising information in the length of the full text of the media information X ₂ =20/50=0.4.

Fig. 2 is a flow chart of obtaining the length of suspected advertisement information provided by the embodiment of the present application.

In one embodiment, as shown in Figure 2, obtaining the length of suspected advertisement information specifically includes the following steps:

Step S201, acquiring the start position and end position of the suspected advertisement information matched by each classification expression respectively.

Exemplarily, the media information (corpus) is:

Gentle and clean, skin-friendly and rejuvenating, without hurting the stratum corneum, it is a rare good cleanser

It matches all concept expressions and positions as shown in Table 2:

concept name concept expression matching content starting point end position 1 sales pitch Gentle.* Skin friendly Gentle and clean, skin-friendly 0 8 2 sales pitch skin rejuvenation skin rejuvenation 6 10

Table 2

It can be seen from Table 2 that the starting position of the matching content of the concept expression in line 1 is 0 and the ending position is 8; the starting position of the matching content of the concept expression in line 2 is 6 and the ending position is 10.

Step S202, judging whether the positions of the suspected advertisement information overlap according to the start position and the end position.

Combined with Table 2, since the starting position of row 1 is smaller than the starting position of row 2, and the ending position of row 1 is greater than the starting position of row 2, and the ending position of row 1 is smaller than the ending position of row 2, the row There is an intersection between the content matched by line 1 and line 2.

Step S203, if the positions of the suspected advertisement information overlap, the difference between the maximum end position value and the minimum start position value in the suspected advertisement information is used as the length of the suspected advertisement information.

Combined with Table 2, since the positions of the suspected advertisement information overlap, and the maximum value of the end position of the suspected advertisement information is 10, and the minimum value of the start position is 0, the length of the suspected advertisement information is 10 at the end position of line 2 The starting position 0 of line 1 is subtracted, that is, 10-0=10; and the total length of the text is 28, so X ₂ =10/28.

2. The proportion of the advertising information released by the user to all the media information released by the user. For example, user behavior tracking and statistics can be performed according to identity information such as account ID, user name, and IP address used by users to post media information, including counting the number C1 of all media information and the number C2 of advertisement information released by users. Based on the above statistical information, the proportion of the advertisement information released by the user to all the media information released by the user is X ₃ =C2/C1.

3. The number of pictures in the suspected advertisement information. For example, if a user's reply in a certain forum is identified as containing suspected advertisement information, and the reply contains 3 pictures, then the number of pictures in the suspected advertisement information X ₄ =3.

Fig. 3 is a flowchart of providing advertisement weights for generating suspected advertisement information according to an embodiment of the present application.

As shown in Figure 2, based on the above three weight factors, the embodiment of the present application adopts the following steps to obtain the advertisement weight of suspected advertisement information:

Step S301, multiplying each weight factor by a corresponding weight coefficient to obtain the weight of each weight factor.

Wherein, suspected advertisement information accounts for the length proportion X ₂ of the full text of the media information corresponding to the first weight, and this first weight is the product of the length proportion X ₂ and the first weight coefficient W ₂ that the suspected advertisement information accounts for the full text of the media information, namely: W ₂ X ₂ . Wherein, the first weight coefficient W ₂ is a parameter preset in the embodiment of the present application, and its value is greater than 0.

In addition, the proportion X3 of the advertisement information released by the user to all the media information released by the user corresponds to the second weight, and the second weight is the proportion _{X3 of} the advertisement information released by the user to all the media information released by the user The product with the second weight coefficient W ₃ , namely: W ₃ X ₃ . Wherein, the second weight coefficient W ₃ is a parameter preset in the embodiment of the present application, and its value is greater than 0.

In addition, the number X ₄ of pictures in the suspected advertisement information corresponds to a third weight, and the third weight is the product of the number X ₄ of pictures in the suspected advertisement information and the third weight coefficient W ₄ , that is, X ₄ W ₄ . The third weight coefficient W ₄ is a parameter preset in the embodiment of the present application, and its value is greater than 0.

It should be added that the values of the above-mentioned first weight coefficient W ₂ , second weight coefficient W ₃ , and third weight coefficient W ₄ can be determined flexibly, for example: when it is necessary to focus on monitoring the behavior of some users placing advertisements multiple times , the second weight coefficient W ₃ can be set to a relatively large value; when some forums, social platforms, etc. appear to place advertisements mainly based on image advertisements, the third weight coefficient W ₄ can be set to a relatively large value.

Step S302, adding the category weight of the suspected advertisement information to the weight of each of the factor weights to obtain the advertisement weight.

In the embodiment of the present application, the advertisement weight indicates the possibility that certain media information (including postings on forums, various dynamics of social platforms, etc.) is an advertisement. If the advertisement weight is represented by X, then:

X＝W ₁ +W ₂ X ₂ +W ₃ X ₃ +X ₄ W ₄

Step S103, determining the suspected advertisement information whose advertisement weight is greater than a preset threshold as advertisement information.

In the embodiment of the present application, a threshold M is preset, and the numerical value is compared with the above advertisement weight X. If the advertisement weight X is greater than or equal to the threshold M, it is determined that the media information is an advertisement, and if the advertisement weight X is less than the threshold M, then it is determined Media information is not an advertisement.

The threshold M can be flexibly determined according to the advertisement identification strategy. If a strict advertisement identification strategy needs to be implemented, the threshold M can take a smaller value. At this time, advertisements and media information with advertising tendencies can be identified together, regardless of manslaughter; if a loose advertisement identification strategy needs to be implemented, the threshold M can take a larger value. At this time, only relatively obvious advertisements are recognized, and omissions are not considered. When a forum or social media is operating, it will determine the strength of the advertising filtering method based on various factors such as management costs and the impact on user experience. Therefore, the threshold M can be determined by the forum or social media operator in practice. It is determined by itself, and is not specifically limited in the embodiments of the present application.

Fig. 4 is a flow chart of managing users posting advertisements provided by the embodiment of the present application.

As shown in Figure 4, in one embodiment, after determining that the suspected advertisement information is an advertisement, the embodiment of the present application further includes the following steps to manage users who publish advertisements:

Step S401, updating the quantity of advertisement information published by the user.

In the embodiment of the present application, for each user, the quantity of advertisement information released by the user is counted, and if it is recognized that the user has released new advertisement information, the counted quantity is updated. Among them, the embodiment of the present application can distinguish whether the user who publishes the advertisement is the same user according to the user's IP address, mobile phone number, mailbox number, and real-name authentication information. , thereby preventing some users from registering multiple accounts to publish advertisements in order to avoid management.

Step S402, judging whether the quantity of advertisement information published by the user is greater than a quantity threshold.

Step S403, when the quantity of advertisement information published by the user is greater than the quantity threshold, ban the user's IP address and delete the user's account information.

In the embodiment of this application, deleting the user's account information can prevent the user who sent the advertisement information from continuing to log in to the account, and blocking the user's IP address is used to prevent the user from re-registering the account, thereby fundamentally eliminating the phenomenon of the user posting advertisements.

In some embodiments, when counting the number of advertisements posted by users, the number of advertisements posted by users in the latest preset time period can be counted in segments, for example: counting the number of advertisements posted by users within one hour Or the number of advertisements posted by the user in a day, etc. At the same time, a quantity threshold is set for each time period, for example, the quantity threshold corresponding to an hour is 3, the quantity threshold corresponding to a day is 5, the quantity threshold corresponding to a week is 10, and so on. In step S402, when judging whether the quantity of advertisement information published by the user is greater than the quantity threshold, as long as the quantity of advertisement information published by the user in any time period is greater than the corresponding threshold, step S403 will be triggered to block the user's IP address and delete the The user's account information.

It can be seen from the above technical solutions that the embodiment of the present application provides an advertisement identification method, including: using a classification model containing classification expressions to obtain suspected advertisement information from media information released by users; generating advertisement information according to at least one preset weight factor The advertisement weight of the suspected advertisement information; the weight factors include the proportion of the length of the suspected advertisement information to the full text of the media information, the proportion of the advertisement information released by the user to all the media information released by the user, and the proportion of the suspected advertisement information One or more of the number of pictures in the advertisement information; determining the suspected advertisement information whose advertisement weight is greater than a preset threshold as advertisement information. The method provided in the embodiment of the present application uses the classification model and advertisement weight to perform two-level screening of media information, first determines the suspected advertisement information, and then determines the suspected advertisement information based on factors such as character length, content, and user behavior corresponding to the suspected advertisement information. Advertisement weight, and determine whether the suspected advertisement information is an advertisement according to the advertisement weight, so as to accurately identify the advertisement, and solve the problem that the existing technology cannot timely and effectively identify the advertisement in the social platform.

The following is the device embodiment of the present application, which provides an advertisement recognition device that can be used to implement the method embodiment of the present application. For the undisclosed technical details in the device embodiment of the present application, please refer to the method of the present application Example.

Fig. 5 is a schematic structural diagram of an advertisement recognition device provided by an embodiment of the present application.

As shown in Figure 5, the device includes:

An information acquisition module 501, configured to acquire suspected advertisement information from media information released by users by using a classification model containing classification expressions;

The weight generation module 502 is configured to generate the advertisement weight of the suspected advertisement information according to at least one preset weight factor; the weight factor includes the proportion of the length of the suspected advertisement information in the full text of the media information, and the One or more of the proportion of advertising information in all media information released by the user, and the number of pictures in the suspected advertising information;

An advertisement determining module 503, configured to determine the suspected advertisement information whose advertisement weight is greater than a preset threshold as advertisement information.

It can be seen from the above technical solutions that the embodiment of the present application provides an advertisement recognition device, which is used to: use a classification model containing classification expressions to obtain suspected advertisement information from media information released by users; according to at least one preset The weight factor generates the advertisement weight of the suspected advertisement information; the weight factor includes the proportion of the length of the suspected advertisement information to the full text of the media information, the proportion of the advertisement information released by the user to all the media information released by the user, and One or more of the number of pictures in the suspected advertisement information; determining the suspected advertisement information whose advertisement weight is greater than a preset threshold as advertisement information. The device provided in the embodiment of the present application uses the classification model and advertisement weight to perform two-level screening of media information, first determines the suspected advertisement information, and then determines the suspected advertisement information based on factors such as character length, content, and user behavior corresponding to the suspected advertisement information. Advertisement weight, and determine whether the suspected advertisement information is an advertisement according to the advertisement weight, so as to accurately identify the advertisement, and solve the problem that the existing technology cannot timely and effectively identify the advertisement in the social platform.

Other embodiments of the application will be readily apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. This application is intended to cover any modification, use or adaptation of the application, these modifications, uses or adaptations follow the general principles of the application and include common knowledge or conventional technical means in the technical field not disclosed in the application . The specification and examples are to be considered exemplary only, with a true scope and spirit of the application indicated by the following claims.

It should be understood that the present application is not limited to the precise constructions which have been described above and shown in the accompanying drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the application is limited only by the appended claims.

Claims (10)

1. a kind of advertisement recognition method characterized by comprising

Doubtful advertising information is obtained from the media information that user issues using the disaggregated model comprising expression formula of classifying；

The advertisement weight of the doubtful advertising information is generated according to preset at least one weight；The weight includes The doubtful advertising information accounts for the length specific gravity of the media information full text, and the announced advertising information of user occupies family and issued Total medium information specific gravity and one or more of the picture number in the doubtful advertising information；

The great doubtful advertising information in preset threshold of the advertising rights is determined as advertising information.

2. the method according to claim 1, wherein the use comprising classify expression formula disaggregated model from Before obtaining doubtful advertising information in the media information of family publication, further includes:

The media information is pre-processed；The pretreatment includes removing specific character from the media information, to described Media information carries out character conversion, and the one or more of Chinese character revolution word is carried out to the media information.

3. the method according to claim 1, wherein

The disaggregated model includes at least one class node, each corresponding advertisement classification of the class node, Mei Geguang Accuse the corresponding class weight of classification；Each class node includes at least one classification expression formula, the classification expression formula For identifying the doubtful advertising information from the media information.

4. according to the method described in claim 3, it is characterized in that, described generate institute according to preset at least one weight State the advertisement weight of doubtful advertising information, comprising:

By each described weight multiplied by corresponding weight coefficient, the weight of each weight is obtained；

The class weight of the doubtful advertising information is added with the weight of each factor weight, obtains the advertising rights Weight.

5. according to the method described in claim 4, it is characterized in that,

The weight of the weight includes the first weight, and first weight is that the doubtful advertising information accounts for the media letter Cease the length specific gravity of full text and the product of the first weight coefficient.

6. according to the method described in claim 4, it is characterized in that,

The weight of the weight includes the second weight, and second weight is that the announced advertising information of user has occupied family The specific gravity of the total medium information of publication and the product of the second weight coefficient.

7. according to the method described in claim 4, it is characterized in that,

The weight of the weight includes third weight, and the third weight is the picture number in the doubtful advertising information With the product of third weight coefficient.

8. the method according to claim 1, wherein the length of the doubtful advertising information is obtained by following steps It takes:

Initial position and the end position of the doubtful advertising information that each classification expression formula is matched to are obtained respectively；

Judge the position of the doubtful advertising information with the presence or absence of intersection according to the initial position and end position；

If the position of the doubtful advertising information there are intersection, by the doubtful advertising information end position maximum value with Length of the difference of initial position minimum value as the doubtful advertising information.

9. the method according to claim 1, wherein described that advertising rights are great in the described doubtful of preset threshold Advertising information is determined as after advertising information, further includes:

Update the quantity of the announced advertising information of user；

Judge whether the quantity of the announced advertising information of user is greater than amount threshold；

When the quantity of the announced advertising information of user is greater than amount threshold, closes the IP address of user and delete the account of user Number information.

10. a kind of advertisement identification device characterized by comprising

Data obtaining module is doubted for using the disaggregated model comprising expression formula of classifying to obtain from the media information that user issues Like advertising information；

Weight generation module, for generating the advertising rights of the doubtful advertising information according to preset at least one weight Weight；The weight includes the length specific gravity that the doubtful advertising information accounts for the media information full text, and user is announced Advertising information occupies in the specific gravity of the announced total medium information in family and the picture number in the doubtful advertising information It is one or more；

Advertisement determining module, for the great doubtful advertising information in preset threshold of the advertising rights to be determined as advertisement letter Breath.