TWI634497B - Method, system, computer program product and readable recording medium for detecting fraud bidder - Google Patents

Abstract

A method, system, computer program product and recordable recording medium for exploring fraudulent bidding users, the system comprising a processing unit, a database group and a transaction record analysis subsystem, wherein the database group stores information and transactions of the user terminal Record, the transaction record contains the record of the commodity item that the user has participated in the bidding, and then perform the following steps: the computer system compares the commodity item that the specific user has participated in the bidding with the similar or identical commodity item that the other user has participated in the bidding. The number of intersections, and the ratio of the number of intersections to the number of commodity items that the particular client has participated in the bidding is calculated to analyze the degree of association between the specific client and the other client, and the foregoing is determined when the degree of association is lower than the rated value. The specific client is a suspected fraudulent bidding client.

Description

Method, system, computer program product and readable for exploring fraudulent bidding users Recording media

The invention relates to a method, a system, a computer program product and a readable record medium for exploring fraudulent bidding users, in particular, the computer system compares the commodity items of a specific user end with the similar or identical commodity items of other users. The quantity is used to analyze the degree of association between a specific client and one of the other clients to determine whether the client has the possibility of fraudulent bidding.

Online auction is one of the most popular methods of e-commerce business. Consumers can display consignment items on the auction website. Bidding users can choose according to their needs and prices. The acceptance of the auctioned goods, and then enjoy the interest of the product bidding.

With the increasing scale of auction transactions, it is accompanied by many transaction-related issues. For example, some bidders are only trying to raise the price of the auctioned goods, and they are not actually interested in buying. In addition to the real price of the auctioned goods, their bidding behavior. It also seriously jeopardizes the fairness of transactions in the auction market. In the "Report on Internet Fraud Crimes" published by the Federal Trade Commission on October 31, 2000, the seller disguised as other consumers and joined the price of bidding to raise the price of the auctioned goods. One of the main types of auction fraud.

Although the current auction website has the function of giving the credit evaluation by the user's purchase record or opening the function to the seller and the buyer to give negative comments and praise to each other, the purchase record is pinched. The situation of building and watering is difficult to judge the basis of bidding bidding users. Therefore, there is still no effective way for the auction website to find out the users who have the possibility of defrauding bidding from multiple bidding users.

Therefore, in order to find out the users who have the possibility of disguising the bidding from among the plurality of bidding users, in order to prevent the price of the auctioned goods from being pushed up, the inventor is devoted to research and proposes a method for exploring the fraudulent bidding users. a computer system, wherein the computer system stores a plurality of client data and a transaction record corresponding to each client, the transaction record includes a record of at least one commodity item that the user has participated in the bidding, and then performs the following steps: the computer The system compares the number of intersections of the commodity items that the at least one specific user has participated in the bidding with the commodity items that the other users have participated in the bidding, and calculates the ratio of the number of intersections to the number of commodity items that the specific user has participated in the bidding. To analyze the degree of association between the specific client and the other client, and determine that the specific client is a suspected fraudulent client when the degree of association is lower than a preset value.

Further, the record of the commodity item further includes a record of the at least one purchased commodity item purchased by the user terminal after participating in the bidding and a record of at least one unpurchased commodity item, and the other user end compares the aforementioned purchased commodity item with the foregoing Any one or combination of purchased commodity items and unpurchased commodity items of a specific user end to calculate the aforementioned intersection quantity.

Further, the computer system first calculates a proportional value of the purchased item items of the other user terminals relative to the number of other user terminals, and extracts the purchased item items whose ratio value exceeds a set value as the information of a related commodity item, and then Each of the related item items is intersected with the purchased item of the specific user end to calculate the intersection quantity.

The computer system further analyzes the actual bidding intention value B of one of the specific users according to the foregoing correlation degree, so as to determine that the specific client is a fraudulent bidding possibility when the actual bidding intention value B is lower than a threshold: B=(P+R)/q; P is the number of purchased commodity items of the specific client, R is the intersection number, and q is the number of commodity items of the specific client.

Further, the computer system uses a cluster algorithm to group the product similarity according to one of the purchased product items of each of the user terminals, and then calculates the related degree, and the similarity of the purchased product is the purchased product item and the specific user of the other user end. The ratio of the quantity value of the purchase item to the intersection quantity and the quantity value of the union.

The invention is also a method for exploring a fraudulent bidding user system, which is a method for deceiving a bidding user by using the foregoing method, comprising: a processing unit; a database group connected to the processing unit, including a user database and a transaction record database The user database stores the data of the client, the transaction record database stores the transaction record; a transaction record analysis subsystem, connected to the processing unit, comprising a comparison module and an operation module, The computer system compares the number of intersections of the commodity items of the specific client with the commodity items of other clients, and the computing module is used to calculate the ratio of the number of intersections to the number of commodity items of a specific client, to analyze the operation. The specific client is associated with one of the other clients, and when the degree of association is lower than a rated value, it is determined that the specific client is a user with the possibility of fraudulent bidding.

Further, the transaction record analysis subsystem further includes a user grouping module for grouping according to the similarity degree of the purchased item items of the foregoing user terminals, the similarity is the purchase of the commodity item and the specific user end of the other user end. The quantity value of the commodity item after the intersection.

The effect of the invention is:

1. The present invention compares the quantity of a commodity item of a specific client with a commodity item of another client to obtain a degree of association, and can determine a specific user when the degree of association of the specific client is lower than a preset value set in advance. The end has the possibility of fraudulent bidding.

2. The present invention uses a clustering algorithm to group a plurality of users according to the similarity of purchased item items, so as to more accurately analyze the degree of association between the item of the specific client and other users.

(100) ‧‧‧Exploration of fraudulent bidding user systems

(1) ‧‧‧Processing unit

(2) ‧ ‧ database group

(21)‧‧‧ User Database

(22) ‧‧‧Transaction Records Database

(3) ‧ ‧ transaction record analysis subsystem

(31)‧‧‧ Alignment module

(32)‧‧‧ Computing Module

(33) ‧‧‧User grouping module

[First figure] is a schematic flow chart of the exploration step of the embodiment of the present invention.

[Second figure] is a system architecture diagram of an embodiment of the present invention.

[Third Diagram] is a schematic diagram of a usage interface of an embodiment of the present invention.

[Fourth Diagram] is a schematic diagram of a usage interface for presenting an operation result according to an embodiment of the present invention.

[Fifth Figure] is a schematic diagram showing the results of actual exploration test using virtual transaction records in the embodiment of the present invention.

In summary of the above technical features, the main effects of the inventive deceptive bidding user system (100) will be clearly presented in the following embodiments.

Referring to the first figure, a system architecture diagram of the system for discovering fraudulent bidding users (100) of the present invention is disclosed. The system for detecting fraudulent bidding users (100) includes a processing unit (1), a database group (2), and A transaction record analysis subsystem (3), wherein: the processing unit (1) is a function of logically computing, controlling, and receiving and outputting instructions of other hardware components.

The database group (2) is connected to the processing unit (1), and includes a user database (21) and a transaction record database (22), wherein the user database (21) stores a plurality of files. The data of the client, the transaction record database (22) stores a transaction record corresponding to each client, and the transaction record includes a record of at least one commodity item that the user has participated in the bidding.

The transaction record analysis subsystem (3) is connected to the processing unit (1), and includes a comparison module (31) and an operation module (32), wherein the comparison module (31) is configured to compare at least one specific The computing module (32) is configured to calculate a ratio of the number of intersections to the quantity of the commodity items of the at least one specific user terminal, to analyze the operation, the number of intersections of the commodity items of the user end is similar to or the same as the commodity items of other users. The specific client is associated with one of the other clients, and determines that the at least one specific client is a fraudulent bidding possibility when the degree of association is lower than a preset value. Preferably, the transaction record analysis subsystem (3) further includes a user grouping module (33) for grouping according to the commodity items in which the respective users have participated in the bidding.

The computer system further analyzes the actual bidding intention value B of one of the specific users according to the foregoing correlation degree, so as to determine that the specific client is a fraudulent bidding possibility when the actual bidding intention value B is lower than a threshold: B=(P+R)/q; P is the number of purchased commodity items of the specific client, R is the intersection number, and q is the number of commodity items of the specific client.

Continuing to refer to the second figure and referring to the first figure, the present invention is also a method for exploring fraudulent bidding users. The program product can be constructed in a program manner, and the program product is stored in a recording medium. For computer systems (such as: desktop computers, notebook computers, smart phones, tablets, personal digital assistants (PDAs)] to read, in addition to the server, for online download and installation; the computer The system first stores the data of the plurality of client terminals and the transaction record corresponding to each of the client terminals. The transaction record includes the record of the at least one commodity item that the user has participated in the bidding, and then performs the following steps: The computer system compares the matching module (31) of the transaction record analysis subsystem (3) with the number of intersections of at least one commodity item in which the specific user has participated in the bidding and the commodity item in which the other user has participated in the bidding. And calculating, by the operation module (32) of the transaction record analysis subsystem (3), the ratio of the number of intersections to the number of commodity items that the specific user has participated in the bidding, to analyze and calculate the degree of association between the specific client and the other client. And determining that the specific client is a suspected fraudulent bidding client when the degree of association is lower than a preset rating.

More specifically, the record of the commodity item further includes a record of at least one purchased product item purchased by the client after participating in the bidding and a record of at least one unpurchased item, and the other user ends with the purchased item. The foregoing intersection quantity is calculated by comparing any one or combination of the purchased commodity item and the unpurchased commodity item of the specific specific terminal.

Preferably, the computer system first calculates a proportional value of each purchased item of the other client relative to the number of other users, and extracts the purchased item item whose ratio exceeds a preset value as a related item item. The data is then combined with the purchased item items of the specific user end to calculate the intersection quantity. For example, the transaction data of a bidding client is [a, b, c; a, b, c, d, e], and the commodity items that have been bidding for are a, b, and c. e, f, the actual bidding willingness value B = {3 + ef The number of de}}/5=4/5=80%.

When the number of the user terminals is large, the computer system preferably first divides the group by using the user grouping module (33) to purchase the product similarity according to one of the purchased product items of the foregoing user terminals, and then calculates the related degree, and the purchase is performed. The product similarity is the ratio of the quantity value of the purchase item item of another client to the purchase item item of a specific client and the quantity value of the union.

The similarity of the purchased product can also be expressed as: purchase similarity = (specific transaction data The number of items in the group center point / the number of items in all transaction data.

For example, a transaction data is [a, b, c, d (purchasing commodity items); a, b, c, d, e, f (commodity items that have been bidding)], a group center point is [a, c , d (purchasing merchandise items); a, b, c, d, e (commercial items that have been bidding)], where {a, b, c, d, e, f} are collections of all merchandise items, then purchases are similar Degree = 3/4 = 75%.

The following is an introduction to the execution screen of the deceptive bidding user system and its test results: the deceptive bidding user system of the present invention is a programming language written by C#, and the transaction data is from the IBM Data Mining website (http://www.almaden.ibm). .com/) Download the data simulation program to generate the transaction data needed to evaluate the experiment. Each item generated by the simulation program, which contains items, is regarded as a commodity item that has been screamed, and then randomly generates the commodity items that have been purchased, and then combines them into transaction data of each bidder. The system generates a transaction database D2 containing 10,000 transaction data as an example. As shown in Table 2, the meanings of the main parameter values are: n represents the number of commodity items, ntran is the number of transaction records, and np is the type. The number of combinations, tl is the average number of items in the transaction items that have been bid for each transaction data, and the remaining parameters are set by default values.

In the transaction record database D2, the commodity items are denoted by numbers 1, 2, 3, ..., 1000, respectively, and the transaction records of the bidding users are indicated by the numbers T1, T2, T3, ..., T10000, and the exploration using the creation is performed. Method, design and build a system for exploring fraudulent bidding users. Please refer to the third figure for revealing the exploration screen of the system of the present invention, and input the values of "purchasing product similarity" and "association degree" respectively, and input the number of bidding users to be detected in "number of bidding users", respectively, to 100. For example, the bidding user indicates that the system will randomly select 100 bidding users as the detection target.

Through the exploration calculation process, the group after grouping can be displayed in the "Group" field. Then display the results found in the "Suspected fraudulent bidding users" field (see Figure 4). For the continuation, please refer to the fifth figure, which shows the results of the exploration of the fraudulent bidding user system of the present invention for different number of bidding users. The result shows that the number of suspected fraudulent bidding users can be kept stable. value.

In view of the foregoing description of the embodiments, the operation and the use of the present invention and the effects of the present invention are fully understood, but the above described embodiments are merely preferred embodiments of the present invention, and the invention may not be limited thereto. Included within the scope of the present invention are the scope of the present invention.

Claims (7)

A method for detecting fraudulent bidding users is applied to a computer system, which stores data of a plurality of client terminals and a transaction record corresponding to each client end, the transaction record including at least one product of the foregoing user side participating in the bidding. The record of the project, the following steps are performed: the computer system compares the quantity of the intersection of the commodity item that the at least one specific user has participated in the bidding with the commodity item that the other client has participated in the bidding, and calculates the number of the intersection to be relatively specific. The proportion of the number of commodity items that the user has participated in the bidding, to analyze and calculate the degree of association between the specific client and the other client, and determine that the specific client is one when the degree of association is lower than a preset value. Suspected fraudulent bidding client, wherein the computer system uses the cluster algorithm to group the product similarity according to one of the purchased product items of each of the foregoing users, and then calculates the associated quantity, and the similarity of the purchased product is other users. Purchase item items and purchase items of specific users The ratio of the number value set to the number of values taken union. The method for detecting a fraudulent bidding user according to the first aspect of the patent application, wherein the record of the commodity item further includes a record of the at least one purchased commodity item purchased by the user terminal after participating in the bidding and at least one unpurchased commodity item. The other user end calculates the foregoing intersection quantity by comparing any one or combination of the purchased commodity item and the unpurchased commodity item of the specific specific terminal with the aforementioned purchased commodity item. For example, the method for exploring a fraudulent bidding user according to the second aspect of the patent application, wherein the computer system first calculates a ratio of the number of purchased items of the other client to the number of other users, and extracts the ratio value by more than one. The set value of the purchased item is used as the item of the related item, and the related item of the item is intersected with the purchased item of the specific user to calculate the number of intersections. For example, in the method of claiming the fraudulent bidding user according to the second or the third aspect of the patent application, wherein the computer system further analyzes the actual bidding intention of one of the specific users according to the foregoing correlation degree. a value B, to determine that the specific client is a fraudulent bidding possibility when the actual bidding intention value B is lower than a threshold: B=(P+R)/q; P is the specific user terminal The number of items purchased, R is the aforementioned number of intersections, and q is the number of commodity items of the aforementioned specific client. A computer program product installed in a computer system for causing the computer system to perform the method of exploring fraudulent bidding users as claimed in any one of claims 1 to 4. A readable record medium storing a computer program product as described in claim 5 of the patent application. A method for exploring fraudulent bidding user system, which is a method for detecting fraudulent bidding users according to any one of claims 1 to 4, comprising: a processing unit; a database group connected to the processing unit, including a user database and a transaction record database, the user database is used to store the data of the client, the transaction record database is to store the transaction record; a transaction record analysis subsystem is connected to the processing unit, including a comparison The module and a computing module are configured to compare the number of intersections of the commodity item of the specific user end with the commodity item of the other user end, and the computing module is used to calculate the quantity of the intersection relative to the commodity item of the specific user end. The ratio of the quantity is used to analyze the degree of association between the specific client and the other client, and when the degree of association is lower than the foregoing rating, the specific client is determined to be a user with the possibility of fraudulent bidding; The grouping module is used for grouping according to the commodity items in which the foregoing users have participated in the bidding.