CN117557347B - E-commerce platform user behavior management method - Google Patents

E-commerce platform user behavior management method Download PDF

Info

Publication number
CN117557347B
CN117557347B CN202410039932.0A CN202410039932A CN117557347B CN 117557347 B CN117557347 B CN 117557347B CN 202410039932 A CN202410039932 A CN 202410039932A CN 117557347 B CN117557347 B CN 117557347B
Authority
CN
China
Prior art keywords
entity
bid
representing
data
entities
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
Application number
CN202410039932.0A
Other languages
Chinese (zh)
Other versions
CN117557347A (en
Inventor
张涛
杜晔
匡建宇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Huadian E Commerce Technology Co ltd
Original Assignee
Beijing Huadian E Commerce Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Huadian E Commerce Technology Co ltd filed Critical Beijing Huadian E Commerce Technology Co ltd
Priority to CN202410039932.0A priority Critical patent/CN117557347B/en
Publication of CN117557347A publication Critical patent/CN117557347A/en
Application granted granted Critical
Publication of CN117557347B publication Critical patent/CN117557347B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0609—Buyer or seller confidence or verification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2431—Multiple classes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Marketing (AREA)
  • Artificial Intelligence (AREA)
  • Strategic Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Business, Economics & Management (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Technology Law (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention relates to the technical field of big data processing, and discloses a user behavior management method of an electronic commerce platform, which comprises the following steps: extracting data of a business platform; analyzing and analyzing based on the data of the business platform to obtain entities and relationship paths among the entities, wherein the entities are divided into a plurality of categories, and the relationship paths are divided into a plurality of categories; encoding data of different modes contained in the entity to obtain an encoded vector, and then aggregating the encoded vector to obtain an initial feature vector of the entity; inputting the initial feature vector of the entity into a data analysis model, and outputting a value representing whether the user has abnormal behaviors; the invention associates different services with transaction information by analyzing entities and defining relationship paths, and then identifies abnormal associated transaction behavior of users by finding rules of the abnormal associated transaction behavior of users through artificial intelligence and deep learning.

Description

E-commerce platform user behavior management method
Technical Field
The invention relates to the technical field of big data processing, in particular to a user behavior management method of an electronic commerce platform.
Background
The invention with the bulletin number of CN113935782A discloses a user abnormal behavior detection system for electronic commerce, which is used for identifying abnormal product comments by comprehensively considering user characteristics of a target user, behavior characteristics of the target user when purchasing a target product and text characteristics of the target product comments, and solves the defects of low abnormal comment identification efficiency and low identification accuracy caused by mining abnormal comments by analyzing comment text tendency in the prior art; the method only extracts commodity transaction information, and for a comprehensive electronic commerce platform comprising bidding business, purchasing business and commodity transaction, the method only can identify abnormal behaviors of a user through the related information of commodity transaction, and can not identify abnormal related transaction behaviors of other businesses.
Disclosure of Invention
The invention provides a user behavior management method of an e-commerce platform, which solves the technical problem that abnormal associated transaction behaviors of bidding purchasing business cannot be identified in the related technology.
The invention provides a user behavior management method of an e-commerce platform, which comprises the following steps:
step 101, extracting data of a business platform;
step 102, analyzing and analyzing based on the data of the business platform to obtain entities and relationship paths among the entities, wherein the entities are divided into a plurality of categories, and the relationship paths are divided into a plurality of categories;
step 103, coding data of different modes contained in the entity to obtain a coding vector, and then aggregating the coding vector to obtain an initial feature vector of the entity;
step 104, inputting the initial feature vector of the entity into a data analysis model, wherein the data analysis model comprises a first hidden layer, a second hidden layer, a third hidden layer, a fourth hidden layer and an output layer, and the calculation formula of the first hidden layer is as follows:
；
；
wherein,representing the representation of entity i under the entity relation of the mth category,/for example>Represents the first hidden layer weight parameter, T represents the transpose,/->、/>、/>Initial features representing the i, j, k th entities, respectivelyVector (S)>Representing a set of neighboring entities of the ith entity under the relationship path of the mth category, the set including the ith entity itself,/-within>Representing vector stitching;
the calculation formula of the second hidden layer is as follows:
；
；
wherein the method comprises the steps ofRepresenting the analytical characteristics of the ith entity, V representing the total number of categories of the relationship path, q,/-and so on>Respectively represent ninth and tenth weight parameters, T represents transpose, < ->Representing a ninth bias parameter; />，/>，/>D is a superparameter and c is +.>Dimension of (2);
the output layer inputs analysis features representing the user's entities and then outputs a value representing whether the user has abnormal behavior.
Further, the categories of the entities include users, products, bid-inviting projects, technical terms, bid-scoring committees, bid-scoring files and bid-scoring experts.
Further, the categories of the relationship path include purchasing the same product, bidding for the same bidding project, commentary for the same bidding project, co-authors of the commentary file, co-formulators of the bidding file.
Further, the neighboring entity of the ith entity refers to an entity to which it is directly connected through an entity relationship.
Further, the aggregation method of the encoding vectors of the entities is completed through an aggregation model, and the aggregation model calculates the formula of the initial feature vector of one entity as follows:
；
；
；
；
；
；
；
；
；
；
；
wherein,a t-th encoding vector representing an entity; />、/>、/>、/>、/>、/>、/>、Respectively represent the first, second, third, fourth, fifth, sixth, seventh and eighth intermediate characteristics,/->Representing a t-th aggregate output feature; />Representing an initial feature direction of an entityThe amount, Z, represents the total number of encoded vectors for the entity;
、/>、/>、/>、/>、/>、/>、/>the representation represents the first, second, third, fourth, fifth, sixth, seventh, eighth weight parameter,/-respectively>、/>、/>、/>、/>、/>Representing the first, second, third, fourth, fifth, sixth, seventh, eighth bias parameter,/->Represents dot product->Representing an activation function->Representing a hyperbolic tangent function.
Further, the first output layer is fully connected and mapped to the classification space, and the classification space of the first output layer comprises two labels which respectively indicate that the user has abnormal behaviors and the user does not have abnormal behaviors.
Further, the definition of whether the user has abnormal behavior is: the user performs an abnormal commodity transaction for bidding.
The invention provides a user behavior management system of an e-commerce platform, which comprises the following steps:
the data storage module is used for storing data of the business platform;
the data extraction module is used for extracting data of the business platform;
the data preprocessing module is used for analyzing and analyzing based on the data of the business platform to obtain entities and relationship paths among the entities, wherein the entities are divided into a plurality of categories, and the relationship paths are divided into a plurality of categories;
the data initial module is used for coding data of different modes contained in the entity to obtain a coding vector, and then aggregating the coding vector to obtain an initial feature vector of the entity;
the data analysis module inputs the initial feature vector of the entity into the data analysis model and outputs a value representing whether the user has abnormal behaviors or not;
and the user management module is used for limiting or canceling the access rights of the users with abnormal behaviors.
The invention provides a user behavior management system of an e-commerce platform, which further comprises:
the bid management module is used for inputting bid information;
the bidding information comprises bidding documents, bidding document formulators, bidding units and the like;
the bid information can be processed and analyzed by the data preprocessing module to obtain an entity, and the bid information input by the bid management module is input into the data preprocessing module to obtain a newly added entity;
the new adding module is used for establishing a relation path between the new adding entity and the existing entity and between the new adding entity and the new adding entity;
the re-execution module is used for inputting the initial feature vectors of the newly added entity and the existing entity into the data analysis model, the data analysis model further comprises an expert matching layer, and the calculation formula of the expert matching layer is as follows:
wherein,analysis features of the entity representing the o-th comment expert,/->Analysis features of entities representing bid items obtained by analysis of entered bid information, ++>Represents an eleventh weight parameter, ++>Representing tenth bias parameter, ++>Representing vector concatenation->The matching vector representing the o-th bid evaluation expert and the input bid information comprises two components, wherein if the first component is larger than the second component, the o-th bid evaluation expert can meet the technical knowledge requirement of the bid evaluation task corresponding to the bid information, otherwise, the matching vector does not meet the technical knowledge requirement;
the bid evaluation committee composition module is used for randomly extracting the number of people who enter the bid information requirement from bid evaluation experts meeting the technical knowledge requirement of the bid evaluation task corresponding to the bid information to form the bid evaluation committee.
The present invention provides a storage medium storing non-transitory computer readable instructions that, when executed by a computer, are capable of performing the steps of a method of managing user behavior of an e-commerce platform as described above.
The invention has the beneficial effects that: the invention associates different services with transaction information by analyzing entities and defining relationship paths, and then identifies abnormal associated transaction behavior of users by finding rules of the abnormal associated transaction behavior of users through artificial intelligence and deep learning.
Drawings
FIG. 1 is a flow chart of a method for managing user behavior of an e-commerce platform according to the present invention;
FIG. 2 is a block diagram of a system for managing user behavior of an e-commerce platform according to the present invention.
In the figure: the system comprises a data storage module 201, a data extraction module 202, a data preprocessing module 203, a data initialization module 204, a data analysis module 205 and a user management module 206.
Detailed Description
The subject matter described herein will now be discussed with reference to example embodiments. It is to be understood that these embodiments are merely discussed so that those skilled in the art may better understand and implement the subject matter described herein and that changes may be made in the function and arrangement of the elements discussed without departing from the scope of the disclosure herein. Various examples may omit, replace, or add various procedures or components as desired. In addition, features described with respect to some examples may be combined in other examples as well, and combinations between different embodiments are not excluded.
In one embodiment of the present invention, a method for managing user behavior of an e-commerce platform is provided, as shown in fig. 1, including the following steps:
step 101, extracting data of a business platform;
step 102, analyzing and analyzing based on the data of the business platform to obtain entities and relationship paths among the entities, wherein the entities are divided into a plurality of categories, and the relationship paths are divided into a plurality of categories;
in one embodiment of the invention, the categories of the entities comprise users, products, bid-inviting projects, technical terms, bid-scoring committees, bid-scoring files and bid-scoring experts;
the types of entities are not exhaustive and may be found or defined according to the needs for data analysis.
Although the foregoing refers to the category of the entity, an entity is not referred to by a name, and an entity refers to a collection of data within the meaning defined by the entity, for example, an entity of a product type may include a collection of data such as a name, descriptive text, and a product picture of a certain product; for example, an entity of a technical term may contain a word attributed to the technical term;
in one embodiment of the invention, the categories of the relationship path include purchasing the same product, participating in bidding for the same bidding project, participating in commentary for the same bidding project, co-authors of the commentary documents, co-formulators of the bidding documents;
the category of relationship paths is not exhaustive, and represents relationship paths representing relationships of entities requiring data analysis, and the relationship paths can be based on one relationship path to link entities under one specific relationship system;
for example, a simple relationship path for purchasing the same product may be described as "user-product-user", a relationship path for a co-formulator of a simple bidding document may be described as "expert-bidding document-expert", and entities under the relationship path are connected by physical relationships, such as the physical relationship between the user and the product being a purchase or an evaluation or collection.
Step 103, coding data of different modes contained in the entity to obtain a coding vector, and then aggregating the coding vector to obtain an initial feature vector of the entity;
in one embodiment of the present invention, the modes of the data include images, text, audio, etc., and the mode for coding these modes is conventional technical means, for example, the images are coded by convolution, the category data are coded by one-hot coding, and the text is coded by Bag-of-words model, etc., and the present invention is not limited to a specific coding mode.
The partial coding mode needs to be realized by means of a mathematical model, wherein parameters to be trained are contained in the mathematical model, and at the moment, the mathematical model required by coding and the data analysis model can be trained together, namely, the mathematical model and the data analysis model participate in the back propagation process of the data analysis model, and the internal parameters of the mathematical model are updated.
In one embodiment of the present invention, the method of aggregating the encoded vectors of the entities is to average the encoded vectors of the entities as the initial feature vectors of the entities.
In one embodiment of the present invention, the method of aggregating the encoded vectors of an entity is accomplished by an aggregation model that calculates an initial feature vector of an entity as follows:
wherein,a t-th encoding vector representing an entity; />、/>、/>、/>、/>、/>、/>、Respectively represent the first, second, third, fourth, fifth, sixth, seventh and eighth intermediate characteristics,/->Representing a t-th aggregate output feature; />Representing an initial feature vector of the entity, Z representing a total number of encoded vectors of the entity;
、/>、/>、/>、/>、/>、/>、/>the representation represents the first, second, third, fourth, fifth, sixth, seventh, eighth weight parameter,/-respectively>、/>、/>、/>、/>、/>Representing the first, second, third, fourth, fifth, sixth, seventh, eighth bias parameter,/->Represents dot product->Representing an activation function->Representing a hyperbolic tangent function.
The aforementioned weight and bias parameters may be updated by training in common with the data analysis model, as with the model required for encoding the data.
Step 104, inputting the initial feature vector of the entity into a data analysis model, wherein the data analysis model comprises a first hidden layer, a second hidden layer, a third hidden layer, a fourth hidden layer and an output layer, and the calculation formula of the first hidden layer is as follows:
wherein,representing the representation of entity i under the entity relation of the mth category,/for example>Represents the first hidden layer weight parameter, T represents the transpose,/->、/>、/>Initial eigenvectors representing the i, j, k th entity, respectively, +.>Representing a set of neighboring entities of the ith entity under the relationship path of the mth category, the set including the ith entity itself,/-within>Representing vector concatenation.
The neighbor entity of the ith entity refers to the entity to which it is directly connected through an entity relationship.
The calculation formula of the second hidden layer is as follows:
wherein the method comprises the steps ofRepresenting the analytical characteristics of the ith entity, V representing the total number of categories of the relationship path, q,/-and so on>Respectively represent ninth and tenth weight parameters, T represents transpose, < ->Representing a ninth bias parameter; />，/>，/>D is a superparameter and c is +.>Is a dimension of (c).
The output layer inputs analysis characteristics of entities representing users and then outputs a value representing whether the users have abnormal behaviors or not;
in one embodiment of the invention, the first output layer is fully connected and mapped to the classification space, and the classification space of the first output layer comprises two labels which respectively indicate that the user has abnormal behaviors and the user does not have abnormal behaviors;
in one embodiment of the invention, for the recognition of abnormal behavior of a user of an e-commerce platform fusing bidding and general e-commerce functions, the definition of whether the user has abnormal behavior is:
the user performs abnormal commodity transaction for bidding;
including but not limited to abnormal behavior of the user caused by the invisible condition of the bid for certain commodity transactions.
In one embodiment of the present invention, as a general user anomaly recognition function, the definition of whether a user has an anomalous behavior is: the user has access behaviors that interfere with the proper operation of the e-commerce platform.
In one embodiment of the present invention, there is provided an electronic commerce platform user behavior management system, as shown in fig. 2, including:
a data storage module 201 for storing data of the commerce platform;
a data extraction module 202 for extracting data of the business platform;
the data preprocessing module 203 analyzes and analyzes based on the data of the business platform to obtain entities and relationship paths among the entities, wherein the entities are divided into a plurality of categories, and the relationship paths are divided into a plurality of categories;
the data initial module 204 encodes data of different modes contained in the entity to obtain a code vector, and then aggregates the code vector to obtain an initial feature vector of the entity;
the data analysis module 205 inputs the initial feature vector of the entity into the data analysis model and outputs a value indicating whether the user has abnormal behavior;
a user management module 206 for limiting or canceling access rights of users having abnormal behaviors;
in one embodiment of the present invention, there is provided an electronic commerce platform user behavior management system, further including:
the bid management module is used for inputting bid information;
the bidding information comprises bidding documents, bidding document formulators, bidding units and the like;
the bid information can be processed and analyzed by the data preprocessing module to obtain an entity, and the bid information input by the bid management module is input into the data preprocessing module to obtain a newly added entity;
the new adding module is used for establishing a relation path between the new adding entity and the existing entity and between the new adding entity and the new adding entity;
the existing entity refers to an entity which is obtained by being processed by the data preprocessing module before the bid information is input;
the re-execution module is used for inputting the initial feature vectors of the newly added entity and the existing entity into the data analysis model, the data analysis model further comprises an expert matching layer, and the calculation formula of the expert matching layer is as follows:
wherein,analysis features of the entity representing the o-th comment expert,/->Sign information analysis and acquisition representing inputAnalytical features of the entity of the resulting bid item, < ->Represents an eleventh weight parameter, ++>Representing tenth bias parameter, ++>Representing vector concatenation->The matching vector representing the o-th bid evaluation expert and the input bid information comprises two components, wherein if the first component is larger than the second component, the o-th bid evaluation expert can meet the technical knowledge requirement of the bid evaluation task corresponding to the bid information, otherwise, the matching vector does not meet the technical knowledge requirement;
the bid evaluation committee composition module is used for randomly extracting the number of people who enter the bid information requirement from bid evaluation experts meeting the technical knowledge requirement of the bid evaluation task corresponding to the bid information to form the bid evaluation committee.
The system is characterized in that the system comprises a management system, wherein the management system is used for managing the evaluation committee, and the management system is used for ensuring the quality of the evaluation committee, wherein the management system is used for ensuring the fairness of the evaluation committee and not intervening in artificial selection, and the management system can be used for ensuring the evaluation committee while avoiding artificial interference with the selection of the evaluation committee members through the composition flow of the evaluation committee.
The output layer can be isolated when the data analysis model comprising the expert matching layer is trained, the output result during training is compared with the evaluation result of the bid evaluation expert, for example, the o-th bid evaluation expert is output to meet the technical knowledge requirement of the bid evaluation task corresponding to bid information, the o-th bid evaluation expert evaluates whether the o-th bid evaluation expert can meet the technical knowledge requirement of the bid evaluation task corresponding to bid information or not to perform comparison, and due to the adoption of supervised training, the influence of comparison on the fairness of evaluation by the comparison of few marking results can be reduced by matching with randomly extracted training samples in the mode.
In one embodiment of the present invention, a storage medium is provided that stores non-transitory computer readable instructions that, when executed by a computer, are capable of performing the steps of an e-commerce platform user behavior management method as described above.
The embodiment has been described above with reference to the embodiment, but the embodiment is not limited to the above-described specific implementation, which is only illustrative and not restrictive, and many forms can be made by those of ordinary skill in the art, given the benefit of this disclosure, are within the scope of this embodiment.

Claims (9)

1. The electronic commerce platform user behavior management method is characterized by comprising the following steps:
step 101, extracting data of a business platform;
step 102, analyzing and analyzing based on the data of the business platform to obtain entities and relationship paths among the entities, wherein the entities are divided into a plurality of categories, and the relationship paths are divided into a plurality of categories;
step 103, coding data of different modes contained in the entity to obtain a coding vector, and then aggregating the coding vector to obtain an initial feature vector of the entity;
step 104, inputting the initial feature vector of the entity into a data analysis model, wherein the data analysis model comprises a first hidden layer, a second hidden layer, a third hidden layer, a fourth hidden layer and an output layer, and the calculation formula of the first hidden layer is as follows:
；
；
wherein,representing the representation of entity i under the entity relation of the mth category,/for example>Represents the first hidden layer weight parameter, T represents the transpose,/->、/>、/>Initial eigenvectors representing the i, j, k th entity, respectively, +.>Representing a set of neighboring entities of the ith entity under the relationship path of the mth category, the set including the ith entity itself,/-within>Representing vector stitching;
the calculation formula of the second hidden layer is as follows:
；
；
wherein the method comprises the steps ofRepresenting the analytical characteristics of the ith entity, V representing the category of the relationship pathTotal, q, < >>Respectively represent ninth and tenth weight parameters, T represents transpose, < ->Representing a ninth bias parameter;
the output layer inputs analysis characteristics of entities representing users and then outputs a value representing whether the users have abnormal behaviors or not;
the bid information is input, the bid information can be processed and analyzed by the data preprocessing module to obtain an entity, and the bid information input by the bid management module is input into the data preprocessing module to obtain a newly added entity;
establishing a relation path between the newly-added entity and the existing entity and between the newly-added entity and the newly-added entity;
the existing entity refers to an entity which is obtained by being processed by the data preprocessing module before the bid information is input;
the initial feature vectors of the newly added entity and the existing entity are input into a data analysis model, the data analysis model further comprises an expert matching layer, and the calculation formula of the expert matching layer is as follows:
；
wherein,analysis features of the entity representing the o-th comment expert,/->Analysis features of entities representing bid items obtained by analysis of entered bid information, ++>Represents an eleventh weight parameter, ++>Representing tenth bias parameter, ++>Representing vector concatenation->The matching vector representing the o-th bid evaluation expert and the input bid information comprises two components, wherein if the first component is larger than the second component, the o-th bid evaluation expert can meet the technical knowledge requirement of the bid evaluation task corresponding to the bid information, otherwise, the matching vector does not meet the technical knowledge requirement;
the bid evaluation committee composition module is used for randomly extracting the number of people who enter the bid information requirement from bid evaluation experts meeting the technical knowledge requirement of the bid evaluation task corresponding to the bid information to form the bid evaluation committee.
2. The method for managing user behavior of an e-commerce platform according to claim 1, wherein the categories of entities include users, products, bid items, technical terms, bid committees, bid documents, bid experts.
3. The method of claim 1, wherein the categories of relationship paths include purchasing the same product, participating in bidding on the same bidding item, participating in bid evaluation on the same bidding item, co-authors of bid evaluation documents, co-formulators of bid evaluation documents.
4. The method for managing user behavior of e-commerce platform according to claim 1, wherein the neighboring entity of the i-th entity is an entity directly connected to the i-th entity through an entity relationship.
5. The method for managing user behavior of electronic commerce platform according to claim 1, wherein the method for aggregating the coded vectors of the entities is accomplished by an aggregation model, and the formula for calculating the initial feature vector of an entity by the aggregation model is as follows:
；
；
；
；
；
；
；
；
；
；
；
wherein,a t-th encoding vector representing an entity; />、/>、/>、/>、/>、/>、/>、/>Respectively represent the first, second, third, fourth, fifth, sixth, seventh and eighth intermediate characteristics,/->Representing a t-th aggregate output feature; />Representing an initial feature vector of the entity, Z representing a total number of encoded vectors of the entity;
、/>、/>、/>、/>、/>、/>、/>the representation represents the first, second, third, fourth, fifth, sixth, seventh, eighth weight parameter,/-respectively>、/>、/>、/>、/>、/>Representing the first, second, third, fourth, fifth, sixth, seventh, eighth bias parameter,/->Represents dot product->Representing an activation function->Representing a hyperbolic tangent function.
6. The method for managing user behavior of an e-commerce platform according to claim 1, wherein the first output layer is fully connected and mapped to the classification space, and the classification space of the first output layer includes two labels respectively indicating that the user has abnormal behavior and that the user does not have abnormal behavior.
7. The method for managing user behavior of an e-commerce platform according to claim 1 or 6, wherein the definition of whether the user has abnormal behavior is: the user performs an abnormal commodity transaction for bidding.
8. An electronic commerce platform user behavior management system, comprising:
the data storage module is used for storing data of the business platform;
the data extraction module is used for extracting data of the business platform;
the data preprocessing module is used for analyzing and analyzing based on the data of the business platform to obtain entities and relationship paths among the entities, wherein the entities are divided into a plurality of categories, and the relationship paths are divided into a plurality of categories;
the data initial module is used for coding data of different modes contained in the entity to obtain a coding vector, and then aggregating the coding vector to obtain an initial feature vector of the entity;
the data analysis module inputs the initial feature vector of the entity into the data analysis model and outputs a value representing whether the user has abnormal behaviors or not;
a user management module for limiting or canceling access rights of users having abnormal behaviors;
further comprises:
the bid management module is used for inputting bid information;
the bid information can be processed and analyzed by the data preprocessing module to obtain an entity, and the bid information input by the bid management module is input into the data preprocessing module to obtain a newly added entity;
the new adding module is used for establishing a relation path between the new adding entity and the existing entity and between the new adding entity and the new adding entity;
the re-execution module is used for inputting the initial feature vectors of the newly added entity and the existing entity into the data analysis model, the data analysis model further comprises an expert matching layer, and the calculation formula of the expert matching layer is as follows:
；
wherein,analysis features of the entity representing the o-th comment expert,/->Analysis features of entities representing bid items obtained by analysis of entered bid information, ++>Represents an eleventh weight parameter, ++>Representing tenth bias parameter, ++>Representing vector concatenation->The matching vector representing the o-th bid evaluation expert and the input bid information comprises two components, wherein if the first component is larger than the second component, the o-th bid evaluation expert can meet the technical knowledge requirement of the bid evaluation task corresponding to the bid information, otherwise, the matching vector representsNot satisfied;
the bid evaluation committee composition module is used for randomly extracting the number of people who enter the bid information requirement from bid evaluation experts meeting the technical knowledge requirement of the bid evaluation task corresponding to the bid information to form the bid evaluation committee.
9. A storage medium storing non-transitory computer readable instructions which, when executed by a computer, are capable of performing the steps of a method of user behavior management for an e-commerce platform according to any one of claims 1 to 7.
CN202410039932.0A 2024-01-11 2024-01-11 E-commerce platform user behavior management method Active CN117557347B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202410039932.0A CN117557347B (en) 2024-01-11 2024-01-11 E-commerce platform user behavior management method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202410039932.0A CN117557347B (en) 2024-01-11 2024-01-11 E-commerce platform user behavior management method

Publications (2)

Publication Number Publication Date
CN117557347A CN117557347A (en) 2024-02-13
CN117557347B true CN117557347B (en) 2024-04-12

Family

ID=89815101

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202410039932.0A Active CN117557347B (en) 2024-01-11 2024-01-11 E-commerce platform user behavior management method

Country Status (1)

Country Link
CN (1) CN117557347B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110555749A (en) * 2019-07-26 2019-12-10 阿里巴巴集团控股有限公司 credit behavior prediction method and device based on neural network
CN111091259A (en) * 2019-09-20 2020-05-01 上海同在互联网科技有限公司 Electronic bidding purchasing abnormal behavior identification method, system, equipment and storage medium
CN113128229A (en) * 2021-04-14 2021-07-16 河海大学 Chinese entity relation joint extraction method
KR102520146B1 (en) * 2022-08-29 2023-04-11 주식회사 디노마드 method of predicting success rate of bid notice using big data collected from bidding server, bidding experience data of target company and machine learning-based artificial intelligence algorithm
CN116166806A (en) * 2023-03-16 2023-05-26 四川大学 False shadow evaluation detection method based on graph attention neural network
CN117011006A (en) * 2023-08-07 2023-11-07 陈涵 Electronic bidding supervision method based on big data mining

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110555749A (en) * 2019-07-26 2019-12-10 阿里巴巴集团控股有限公司 credit behavior prediction method and device based on neural network
CN111091259A (en) * 2019-09-20 2020-05-01 上海同在互联网科技有限公司 Electronic bidding purchasing abnormal behavior identification method, system, equipment and storage medium
CN113128229A (en) * 2021-04-14 2021-07-16 河海大学 Chinese entity relation joint extraction method
KR102520146B1 (en) * 2022-08-29 2023-04-11 주식회사 디노마드 method of predicting success rate of bid notice using big data collected from bidding server, bidding experience data of target company and machine learning-based artificial intelligence algorithm
CN116166806A (en) * 2023-03-16 2023-05-26 四川大学 False shadow evaluation detection method based on graph attention neural network
CN117011006A (en) * 2023-08-07 2023-11-07 陈涵 Electronic bidding supervision method based on big data mining

Also Published As

Publication number Publication date
CN117557347A (en) 2024-02-13

Similar Documents

Publication Publication Date Title
CN110532542B (en) Invoice false invoice identification method and system based on positive case and unmarked learning
US11809519B2 (en) Semantic input sampling for explanation (SISE) of convolutional neural networks
CN110955750A (en) Combined identification method and device for comment area and emotion polarity, and electronic equipment
Siering " Boom" or" Ruin"--Does It Make a Difference? Using Text Mining and Sentiment Analysis to Support Intraday Investment Decisions
WO2021174699A1 (en) User screening method, apparatus and device, and storage medium
CN107133862A (en) Dynamic produces the method and system of the detailed transaction payment experience of enhancing credit evaluation
CN115099310A (en) Method and device for training model and classifying enterprises
CN112102049A (en) Model training method, business processing method, device and equipment
CN112990443B (en) Neural network evaluation method and device, electronic device, and storage medium
CN113378090B (en) Internet website similarity analysis method and device and readable storage medium
CN114118816A (en) Risk assessment method, device and equipment and computer storage medium
CN117557347B (en) E-commerce platform user behavior management method
CN113592512A (en) Online commodity identity uniqueness identification and confirmation system
CN112966503A (en) Aspect level emotion analysis method
Altaei Detection of Deep Fake in Face Images Using Deep Learning
CN112463966B (en) False comment detection model training method, false comment detection model training method and false comment detection model training device
CN113792541B (en) Aspect-level emotion analysis method introducing mutual information regularizer
CN115760454A (en) Financial fraud identification method based on cycle width learning
Toloo et al. On ranking discovered rules of data mining by data envelopment analysis: some new models with applications
CN110570301B (en) Risk identification method, device, equipment and medium
CN110472680B (en) Object classification method, device and computer-readable storage medium
CN109829109B (en) Recommendation method based on co-occurrence analysis
CN111325419A (en) Method and device for identifying blacklist user
US20220092634A1 (en) Information processing device and non-transitory computer readable medium
CN115545467B (en) Risk commodity identification model based on graphic neural network

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