CN110134770A - A device and method for false detection of online commodity reviews based on capsule network - Google Patents

Abstract

The invention discloses a false detection device for naval commodity reviews based on a capsule network, which relates to the technical field related to the capsule network, including a review acquisition unit, a review detection unit, a false review confirmation unit, a data sharing set unit, and an output terminal of the review acquisition unit The input end of the comment detection unit is electrically connected, and the output end of the comment detection unit is electrically connected to the input end of the false comment confirmation unit. The invention also discloses a method for false detection of online commodity reviews based on the capsule network, which includes five steps. The present invention is conducive to improving the accuracy of identifying false reviews, has superior detection, analysis and identification capabilities for business reviews, can handle most of the detection tasks, has good versatility, avoids false reviews that mislead other users to consume, and can screen And identify false reviews such as illegal reviews, prevent false reviews from maliciously attacking good merchants, and provide merchants with an equal and safe online environment.

Description

A device and method for false detection of online commodity reviews based on capsule network

technical field

The present invention relates to the technical field related to the capsule network, in particular to a device and method for false detection of naval commodity reviews based on the capsule network.

Background technique

With the development of Internet business, in order to allow users to decide their purchase intentions based on their own intentions and the evaluations of other consumers, major e-commerce platforms have developed various user feedback mechanisms, among which the product evaluation system is the most popular one. Since many users tend to read relevant product reviews before making a purchase decision on the e-commerce platform, some criminals use the evaluation system to inject a large number of false reviews on the e-commerce platform to influence the sales of products for the purpose of profit. E-commerce has generated a large amount of review data, coupled with the uneven quality of reviews, it will be a huge challenge for consumers to consult and analyze these data, which has far exceeded the information processing ability of an ordinary consumer. Massive commodity reviews contain not only real and valuable consumer experiences and opinions, but also a considerable number of fake reviews. Behind the false product reviews is the drive of huge commercial interests, which is a typical unfair commercial competition. Therefore, the proliferation of false product reviews will inevitably seriously affect the healthy development of the e-commerce ecosystem.

Therefore, it is necessary to propose a fake detection device and method based on capsule network to solve the above problems.

Contents of the invention

(1) Solved technical problems

Aiming at the deficiencies of the prior art, the present invention provides a device and method for false detection of online product reviews based on a capsule network, which solves the problems raised in the above-mentioned background technology.

(2) Technical solutions

To achieve the above object, the present invention is achieved through the following technical solutions:

A capsule network-based false detection device for water army product reviews, including a review acquisition unit, a review detection unit, a false review confirmation unit and a data sharing set unit, the output of the review acquisition unit is electrically connected to the input of the review detection unit, The output end of the comment detection unit is electrically connected to the input end of the false comment confirmation unit, and the comment detection unit is electrically connected to the data sharing set unit bidirectionally, and the comment detection unit includes a comment content attribute detection unit, a reviewer attribute detection unit and Review item attribute detection unit.

Optionally, the comment content attribute detection unit includes a natural language similarity test module, a comment semantic artificial intelligence recognition test module and a comment style artificial intelligence recognition test module, and the output terminal of the natural language similarity test module is electrically connected to the comment semantic The input end of the artificial intelligence identification test module, the output end of the comment semantic artificial intelligence identification test module is electrically connected to the input end of the comment style artificial intelligence identification test module.

Optionally, the commenter attribute detection unit includes a comment time and frequency comparison module, a comment IP and MAC address comparison module, a commenter geographical location comparison module, a commenter active time monitoring module, an associated mobile terminal device model comparison module and a comment Person name analysis module.

Optionally, the output end of the comment time and frequency comparison module is electrically connected to the input end of the comment IP and MAC address comparison module, and the output end of the comment IP and MAC address comparison module is electrically connected to the input of the reviewer's geographic location comparison module end, the output end of the reviewer's geographic location comparison module is electrically connected to the input end of the reviewer's active time monitoring module, and the output end of the reviewer's active time monitoring module is electrically connected to the input end of the associated mobile terminal device model comparison module, so The output end of the model comparison module of the associated mobile terminal device is electrically connected to the input end of the reviewer name analysis module.

Optionally, the reviewer name analysis module includes a user name regularity analysis module, a time interval analysis module for submitting information, and an analysis module for multiple reviews of the same product.

Optionally, the output end of the user name pattern analysis module is electrically connected to the input end of the submitted information time interval analysis module, and the output end of the submitted information time interval analysis module is electrically connected to the input end of the commodity multi-comment analysis module.

Optionally, the review product attribute detection unit includes a product actual description matching module and a product actual sales ranking matching module, the output end of the product actual description matching module is electrically connected to the input end of the product actual sales ranking matching module.

A method for false detection of water army commodity reviews based on capsule network, comprising the following steps:

A: Comment acquisition: Obtain comment messages received during a preset period of time;

B: Based on review content detection: According to the lexical features used in the review, the similarity of the review content and style from different reviewers, and the semantics of the review, the authenticity of the review is identified by comparing it with the data sharing set database, and using natural language understanding and artificial intelligence for identification;

C: Based on commenter attribute detection: According to the time and frequency of comment publishing, the commenters involved, the IP or MAC address used by the comment, the geographical location of the commenter, the active time, the mobile terminal model, etc., the abnormal behavior of the comment is found through comparative analysis , so as to identify the falsity of the review, and then analyze the name of the reviewer, according to the rules of the user name, the time interval of submitting information, and the multiple reviews of the same product. When there are similar or similar user names, it can be judged as Fake reviews, when the time interval for submitting information is very short, it can be judged as a fake review, and when there are multiple reviews on the same product, it can be judged as a fake review;

D: Product feature detection based on reviews: According to the actual description of the product, the actual sales ranking and other information, compare and identify fake reviews. If a product sells well but there are obviously derogatory reviews, the review is a fake review;

E: False comment confirmation: After obtaining and analyzing the shared set of comment comparison data, the confirmed false comment data is obtained.

(3) Beneficial effects

The present invention provides a false detection device and method based on a capsule network for naval commodity reviews, which has the following beneficial effects:

(1), the present invention is conducive to improving the accuracy of identifying false reviews, has superior detection, analysis and identification capabilities for merchant reviews, is capable of most detection tasks, has good versatility, and avoids false reviews that mislead other users to consume .

(2) The present invention can screen and identify false reviews such as illegal reviews, prevent false reviews from maliciously attacking excellent merchants, and provide merchants with an equal and safe online environment.

Description of drawings

Fig. 1 is a device block diagram of the present invention;

Fig. 2 is a block diagram of the comment detection unit of the present invention;

Fig. 3 is a block diagram of the comment content attribute detection unit of the present invention;

Fig. 4 is a block diagram of reviewer attribute detection unit of the present invention;

Fig. 5 is a block diagram of a review product attribute detection unit in the present invention.

In the figure: review acquisition unit 1, review detection unit 2, false review confirmation unit 3, data sharing set unit 4, review content attribute detection unit 5, reviewer attribute detection unit 6, review commodity attribute detection unit 7, natural language similarity Test module 8, comment semantic artificial intelligence recognition test module 9, comment style artificial intelligence recognition test module 10, comment time and frequency comparison module 11, comment IP and MAC address comparison module 12, commenter geographic location comparison module 13, commenter active Time monitoring module 14, associated mobile terminal device model comparison module 15, reviewer name analysis module 16, user name rule analysis module 17, submission information time interval analysis module 18, same product multiple comment analysis module 19, product actual description matching module 20 , The actual sales ranking matching module 21 of commodities.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

In describing the present invention, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", " Back", "Left", "Right", "Vertical", "Horizontal", "Top", "Bottom", "Inner", "Outer", "Clockwise", "Counterclockwise", "Axial" , "radial", "circumferential" and other indicated orientations or positional relationships are based on the orientations or positional relationships shown in the drawings, which are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying the referred device or Elements must have certain orientations, be constructed and operate in certain orientations, and therefore should not be construed as limitations on the invention.

In the present invention, unless otherwise clearly specified and limited, terms such as "installation", "installation", "connection", "connection" and "fixation" should be understood in a broad sense, for example, it can be a fixed connection or It is a detachable connection; it may be a mechanical connection; it may be a direct connection or an indirect connection through an intermediary. Those of ordinary skill in the art can understand the specific meanings of the above terms in the present invention according to specific situations.

In addition, the terms "first", "second", etc. are used for descriptive purposes only, and should not be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, a feature defined as "first" and "second" may explicitly or implicitly include one or more of these features.

As shown in Figures 1-5, the present invention provides a technical solution:

A capsule network-based device for detecting false reviews of navy commodities, including a review acquisition unit 1, a review detection unit 2, a false review confirmation unit 3, and a data sharing set unit 4, and the output end of the review acquisition unit 1 is electrically connected to the review detection unit 2 The input terminal of the comment detection unit 2 is electrically connected to the input terminal of the false comment confirmation unit 3, and the comment detection unit 2 and the data sharing set unit 4 are electrically connected bidirectionally, and the comment detection unit 2 includes a comment content attribute detection unit 5, a reviewer Attribute detection unit 6 and comment commodity attribute detection unit 7;

The comment content attribute detection unit 5 includes a natural language similarity test module 8, a comment semantic artificial intelligence recognition test module 9 and a comment style artificial intelligence recognition test module 10, and the output terminal of the natural language similarity test module 8 is electrically connected to the comment semantic artificial intelligence recognition The input end of the test module 9, the output end of the comment semantic artificial intelligence recognition test module 9 is electrically connected to the input end of the comment style artificial intelligence recognition test module 10, and the reviewer attribute detection unit 6 includes a comment time and frequency comparison module 11, comment IP and MAC address comparison module 12, reviewer's geographic location comparison module 13, reviewer's active time monitoring module 14, associated mobile terminal equipment model comparison module 15 and reviewer's name analysis module 16, comment time and the output terminal of frequency comparison module 11 are electrically connected Comment on the input of the IP and MAC address comparison module 12, the output of the comment IP and the MAC address comparison module 12 is electrically connected to the input of the reviewer's geographic location comparison module 13, and the output of the reviewer's geographic location comparison module 13 is electrically connected to the reviewer The input end of the active time monitoring module 14, the output end of the reviewer active time monitoring module 14 is electrically connected to the input end of the associated mobile terminal device model comparison module 15, and the output end of the associated mobile terminal device model comparison module 15 is electrically connected to the reviewer name analysis The input end of module 16, reviewer's name analysis module 16 comprises user name rule analysis module 17, submission information time interval analysis module 18 and same commodity many comments analysis module 19, the output terminal of user name rule analysis module 17 is electrically connected to submit information time The input end of interval analysis module 18, the output end of submitting information time interval analysis module 18 is electrically connected with the input end of commodity multi-comment analysis module 19;

The review product attribute detection unit 7 includes a product actual description matching module 20 and a product actual sales ranking matching module 21 , the output end of the product actual description matching module 20 is electrically connected to the input end of the product actual sales ranking matching module 21 .

A method for false detection of water army commodity reviews based on capsule network, comprising the following steps:

A: Comment acquisition: Obtain comment messages received during a preset period of time;

B: Based on review content detection: According to the lexical features used in the review, the similarity of the review content and style from different reviewers, and the semantics of the review, the authenticity of the review is identified by comparing it with the data sharing set database, and using natural language understanding and artificial intelligence for identification;

C: Based on commenter attribute detection: According to the time and frequency of comment publishing, the commenters involved, the IP or MAC address used by the comment, the geographical location of the commenter, the active time, the mobile terminal model, etc., the abnormal behavior of the comment is found through comparative analysis , so as to identify the falsity of the review, and then analyze the name of the reviewer, according to the rules of the user name, the time interval of submitting information, and the multiple reviews of the same product. When there are similar or similar user names, it can be judged as Fake reviews, when the time interval for submitting information is very short, it can be judged as a fake review, and when there are multiple reviews on the same product, it can be judged as a fake review;

D: Product feature detection based on reviews: According to the actual description of the product, the actual sales ranking and other information, compare and identify fake reviews. If a product sells well but there are obviously derogatory reviews, the review is a fake review;

E: False comment confirmation: After obtaining and analyzing the shared set of comment comparison data, the confirmed false comment data is obtained.

It should be noted that, in the present invention, unless otherwise specified and limited, the first feature may be in direct contact with the first feature or the first feature and the second feature "on" or "under" the second feature. Features come into contact indirectly through intermediaries. Moreover, "above", "above" and "above" the first feature on the second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. "Below", "beneath" and "beneath" the first feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is less horizontally than the second feature.

Finally, it should be noted that: the above is only a preferred embodiment of the present invention, and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, it still It is possible to modify the technical solutions recorded in the foregoing embodiments, or to perform equivalent replacements on some of the technical features. Any modifications, equivalent replacements, improvements, etc. within the spirit and principles of the present invention shall be included in the within the protection scope of the present invention.

Claims (8)

1. A false detection device for water army commodity reviews based on a capsule network, comprising a review acquisition unit (1), a review detection unit (2), a false review confirmation unit (3) and a data sharing set unit (4), characterized in that : The output end of the comment acquisition unit (1) is electrically connected to the input end of the comment detection unit (2), and the output end of the comment detection unit (2) is electrically connected to the input end of the false comment confirmation unit (3), and the comment Detection unit (2) and data sharing set unit (4) are bidirectionally electrically connected, and described comment detection unit (2) comprises review content attribute detection unit (5), reviewer attribute detection unit (6) and comment product attribute detection unit ( 7). 2. A capsule network-based fake detection device for water army product reviews according to claim 1, characterized in that: The comment content attribute detection unit (5) includes a natural language similarity test module (8), a comment semantic artificial intelligence recognition test module (9) and a comment style artificial intelligence recognition test module (10), and the natural language similarity test The output end of the module (8) is electrically connected to the input of the comment semantic artificial intelligence recognition test module (9), and the output end of the comment semantic artificial intelligence recognition test module (9) is electrically connected to the comment style artificial intelligence recognition test module (10) input terminal. 3. A capsule network-based fake detection device for water army product reviews according to claim 1, characterized in that: Described reviewer attribute detection unit (6) comprises comment time and frequency comparison module (11), comment IP and MAC address comparison module (12), reviewer geographic location comparison module (13), reviewer active time monitoring module (14 ), an associated mobile terminal device model comparison module (15) and a reviewer name analysis module (16). 4. A capsule network-based fake detection device for water army product reviews according to claim 3, characterized in that: The output end of the comment time and frequency comparison module (11) is electrically connected to the input end of the comment IP and the MAC address comparison module (12), and the output end of the comment IP and the MAC address comparison module (12) is electrically connected to the geographic location of the reviewer. The input end of the position comparison module (13), the output end of the reviewer's geographic location comparison module (13) is electrically connected to the input end of the reviewer's active time monitoring module (14), and the reviewer's active time monitoring module (14) The output end of the associated mobile terminal equipment model comparison module (15) is electrically connected to the input end of the associated mobile terminal equipment model comparison module (15), and the output end of the associated mobile terminal equipment model comparison module (15) is electrically connected to the input end of the reviewer name analysis module (16). 5. A capsule network-based fake detection device for water army product reviews according to claim 4, characterized in that: The reviewer's name analysis module (16) includes a user name regularity analysis module (17), a time interval analysis module for information submission (18) and a multi-comment analysis module for the same commodity (19). 6. A capsule network-based fake detection device for online commodity reviews according to claim 5, characterized in that: The output end of the user name rule analysis module (17) is electrically connected to the input of the submission information time interval analysis module (18), and the output end of the submission information time interval analysis module (18) is electrically connected to the multi-comment analysis module of the commodity (19) input terminal. 7. A capsule network-based fake detection device for online commodity reviews according to claim 1, characterized in that: The review product attribute detection unit (7) includes a product actual description matching module (20) and a product actual sales ranking matching module (21), and the output terminal of the product actual description matching module (20) is electrically connected to the product actual sales ranking matching module. Input terminal of module (21). 8. A method for false detection of water army product reviews based on capsule network, characterized in that: comprising the following steps: A: Comment acquisition: Obtain comment messages received during a preset period of time; B: Based on review content detection: According to the lexical features used in the review, the similarity of the review content and style from different reviewers, and the semantics of the review, the authenticity of the review is identified by comparing it with the data sharing set database, and using natural language understanding and artificial intelligence for identification; C: Based on commenter attribute detection: According to the time and frequency of comment publishing, the commenters involved, the IP or MAC address used by the comment, the geographical location of the commenter, the active time, the mobile terminal model, etc., the abnormal behavior of the comment is found through comparative analysis , so as to identify the falsity of the review, and then analyze the name of the reviewer, according to the rules of the user name, the time interval of submitting information, and the multiple reviews of the same product. When there are similar or similar user names, it can be judged as Fake reviews, when the time interval for submitting information is very short, it can be judged as a fake review, and when there are multiple reviews on the same product, it can be judged as a fake review; D: Product feature detection based on reviews: According to the actual description of the product, the actual sales ranking and other information, compare and identify fake reviews. If a product sells well but there are obviously derogatory reviews, the review is a fake review; E: False comment confirmation: After obtaining and analyzing the shared set of comment comparison data, the confirmed false comment data is obtained.