TWI776742B - System for analyzing user behavior in information exchange platform - Google Patents

Abstract

A system for analyzing user behavior in an information exchange platform is disclosed. The system includes a platform monitoring module, an online time analysis module, a consumption behavior analysis module, a data grouping module, an analysis result processing module, and a data labeling module. The online time analysis module analyzes the user online time period, the consumer behavior analysis module looks for customer unit price and average discounts, the data grouping module performs grouping for users, and the data labeling module marks the characteristics of each digital footprint. The merchants of this system can obtain information that is very close to the consumer market, so as to carry out precise marketing.

Description

A system for analyzing user behavior in an information exchange platform

The present invention relates to a system for analyzing user behavior, in particular to a system for analyzing user behavior in an information exchange platform.

With the vigorous development of the online consumer market, business models for marketing through various online channels have emerged one after another. The advertising that people often say no longer only refers to the kanban or flat print with fixed screen text, and the multimedia marketing method is the mainstream nowadays. Even though this single-source, multi-channel advertising model is well known to everyone, in order to maximize its effectiveness in practice, it is still necessary to grasp the traffic (viewing volume), which has created a group of service providers that increase commercial interests for customers by bringing traffic. in portals, social platforms, and even communication software. However, their approach is not necessarily the most effective, hence the saying of precision marketing.

Regarding precision marketing, it is mainly to have accurate predictions on the consumption behavior of target consumers, so that suppliers can know what kind of commodity information to provide, so that consumers can accept them to promote transactions. Various applications in techniques for predicting consumer behavior are proposed from time to time. Traditionally, predictions have been based on consumers' past spending behavior, with little reliance on consumers' intentional or unintentional actions. However, the analysis of the consumer's complete behavior (the digital footprint on the Internet) makes it possible to understand the true mind of the consumer, but there is not much research and technology in this area.

The invention starts from the user behavior analysis of the information exchange platform, and further predicts the sales, so as to achieve the purpose of precise marketing.

This text extracts and compiles certain features of the invention. Other features will be disclosed in subsequent paragraphs. It is intended to cover various modifications and similar arrangements within the spirit and scope of the appended claims.

The invention discloses a system for analyzing user behavior in an information exchange platform, which is installed in a server host. The system includes: a platform monitoring module that continuously receives a plurality of digital footprints sent by a management server that executes a service of an information exchange platform, each digital footprint includes a user token and an object operated through the user token , wherein the object contains a plurality of attribute data that can be changed by operation; an online period analysis module performs the following operations: for each digital footprint, according to a trigger time included in the content of the attribute data, confirm that the corresponding user token is in The online periods that appear in a plurality of online periods in a day; and the online periods in which each user token appears the most in a first period of time; a consumption behavior analysis module, which performs the following operations: in a second period of time In the second time period, the content of calculating the relevant attribute data for each user token includes the average amount of each consumption amount as the unit price of one customer; and in the second time period, the content of calculating the relevant attribute data for each user token The discount rate for each consumption amount is an average discount; a data grouping module performs the following operations: Counting all customer unit prices and average discounts related to each user token in a third time period; All relevant customer unit prices of user tokens, using a grouping algorithm to group each user token into one of several customer price groups; and all relevant average discounts for all user tokens, using the grouping algorithm The method is to group each user token into one of several average discount groups; and an analysis result processing module, which analyzes the online period with the most occurrence of a specific user token, the customer unit price, the average discount, and the The clustering results of the data clustering module are transmitted to the specified cloud platform through the specified application program interface.

According to the present invention, the consumption behavior analysis module can further perform the following operations: for each user token in a fourth time period, calculate a last online time interval and an online frequency through the trigger time of the content of the relevant attribute data , and accumulate the total consumption amount; calculate the median of the last online time interval, the online frequency and the total consumption amount calculated above for all user tokens in the fourth time period; for each user token, use the These medians are used as the standard to judge whether the last online time interval, online frequency and total consumption are above or below the median; and according to the combination of the judgment results, they are divided into 8 categories, and each user token is classified as one of them one. The analysis result processing module further transmits the classification result of the specific user token to the designated cloud platform through the designated application program interface.

According to the present invention, the system for analyzing user behavior in the information exchange platform may further include a data labeling module, which monitors the module on the platform through a correspondence table having a plurality of specified attribute data contents and label texts corresponding to each other. From the attribute data contained in the digital footprint received each time, find out at least one specified attribute data content that matches the corresponding table, and compare the label text corresponding to the at least one specified attribute data content with the user token string contained in the digital footprint catch.

According to the present invention, the data grouping module can further perform the following operations: counting the number of times of concatenation of a specified label text concatenated by each user token in a fifth time period; and using the specified label of all user tokens The number of times of concatenation of text and the number of times of concatenation of the designated label text of a designated user token, and the designated user token is grouped by the grouping algorithm. According to the present invention, the analysis result processing module can further transmit all user tokens concatenated with at least one specified label text in a sixth time period to the specified cloud platform through the specified application programming interface, so that the specified cloud The platform sends newsletters, email marketing (EDM) or advertisements to these user tokens.

According to the present invention, the information exchange platform can be an interactive web page, an instant messaging platform or a customer relationship management platform. If the information exchange platform is an instant messaging platform, the content of the attribute data is the name of the clicked data box, the specific sentence or trigger time provided by the instant messaging platform. If the information exchange platform is an interactive web page, the content of the attribute data is the name of the clicked object, the trigger time, the description text or the value with a specific unit. Finally, the clustering algorithm is the k-means clustering algorithm or the Jenks natural breaks optimization algorithm.

The online period analysis module analyzes the online period of users, the consumer behavior analysis module finds customer unit price and average discount, the data grouping module groups users, and the data labeling module marks the characteristics of each digital footprint. Merchants in the system can obtain information that is very close to the consumer market, so as to carry out precise marketing.

The present invention will be described more specifically by referring to the following embodiments.

Please refer to FIG. 1 , which is a simplified block diagram of a system for analyzing user behavior in an information exchange platform (hereinafter referred to as the system) according to an embodiment of the present invention. The system can be installed in a server host 10 and operates by the hardware structure of the server host 10 . The hardware architecture of the server host 10 on which the system is installed is not significantly different from the general server architecture, and may include a central processing unit, a memory, a storage device (such as a hard disk), an I/O unit, and the like. Although these hardwares are not shown in FIG. 1 , they are structures that should be understood by those skilled in the server field. In addition, one of the important hardwares in the server host 10 is the network communication interface 110 , which is an important software and firmware (sometimes also includes program software running on the operating system) connecting the server host 10 and external hardware through the network 30 . The final assembly can include network cards, connecting cables, wireless communication modules and other hardware. The following introduces the technical requirements of the present system for each module of the present invention to use or cooperate with the equipment of the above-mentioned existing server host 10 . Therefore, they can be software, including specific code and data, and run under the operating system on at least a part of the hardware structure (such as the code and related data files are stored in the storage device, in the operating system. Temporarily stored in memory during operation, and executed dynamically for CPU calls). On the other hand, the modules may also be special hardware, such as application-specific integrated circuits (ASICs) or add-in cards, for performing the functions assigned by the modules. What's more, these technical elements can be partly software and partly hardware, and they are effectively integrated according to the needs of product designers, all within the technical scope claimed by this patent.

The system includes a platform monitoring module 210 , an online period analysis module 220 , a consumption behavior analysis module 230 , a data grouping module 240 , an analysis result processing module 250 and a data labeling module 260 . The following describes each module in detail.

The platform monitoring module 210 continuously receives a plurality of digital footprints sent by a management server executing a service of an information exchange platform through the network communication interface 110 . According to the present invention, an information exchange platform refers to a cloud platform that can exchange information among users, or provide information to specific or indefinite users, whether it is a public cloud or a private cloud. The former is, for example, an instant messaging platform, such as LINE ^TM , in this embodiment, an instant messaging management server 420 is used to process all operations; the latter can be an interactive web page or a customer relationship management platform, in this embodiment, a web page is used to manage The server 410 provides interactive web page services, and a customer relationship management server 430 handles related operations. The software running on these servers can provide an Application Programming Interface (API) for the platform monitoring module 210 to use for webhooking, and the servers can obtain the user's information according to trigger events. digital footprint. The so-called digital footprint here not only refers to what actions users have done on the information exchange platform, such as clicking on an icon, uploading data, receiving files, sending messages, etc., but also includes some actions generated by these actions. News. For the description of the digital footprint, please refer to Figure 2, which illustrates the structure of the digital footprint transmitted. Each digital footprint contains a User Token and an object manipulated by the User Token. The user token represents the identity of the user on the information exchange platform, which can be used by the system to distinguish different users but does not need to be linked to the user's personal information. Generally speaking, a user token is a series of meaningless strings with fixed digits but containing numbers, English and/or symbols, and the user's alias on the information exchange platform for the administrator and other users to identify (using name or nickname) are different. The user represented by the user token operates the components on the information exchange platform, thereby producing the digital footprint. The object transmitted from the information exchange platform to the network communication interface 110 along with the user token is actually a combination of some data. The object contains several attribute data that can be changed by operations (time, state, and user token of the digital footprint). These attribute data are encapsulated in JSON (JavaScript Object Notation) format, such as {"Attribute A" in Figure 2). :"2020/2/16 04:11:27 PM"}, before the semicolon is the attribute name, after the semicolon is the attribute value, the two are collectively called attribute data. In practice, if the information exchange platform is an interactive web page, the content of the attribute data can be the name of the clicked object, the trigger time, the description text, the value with a specific unit, etc.; if the information exchange platform is an instant messaging platform, the content of the attribute data It can be the name of the clicked data box, the specific statement provided by the instant messaging platform, the trigger time, etc. It should be noted that the user is different from the merchants mentioned later. The former is an entity user who is active in the information exchange platform, while the latter is an analysis data demander who wants to obtain user consumption behavior analysis through this system.

The online period analysis module 220 performs some operations based on the digital footprint received by the platform monitoring module 210 . The first operation is for each digital footprint, according to a trigger time included in the content of the attribute data, to confirm the online time period that the corresponding user token appears in a plurality of online time periods in a day. For a better understanding of this, please refer to FIG. 3 , which lists the digital footprint transmitted by the web management server 410 . For the convenience of explanation, all digital footprints are sorted in chronological order and given a serial number. In this embodiment, the web page management server 410 is a management server of an online store, and its content is presented in the form of a web page. With the application program interface, when the user clicks the "checkout" icon in the shopping cart, the web management server 410 not only confirms and transmits the transaction-related information, but also sends the relevant information generated by the action to the user. The attribute data format is encapsulated in the object and transmitted to the platform monitoring module 210 together with the user token of the operation. The attribute profile contains the trigger time, which is when the "checkout" icon is clicked. According to the present invention, in order to analyze the consumption time distribution of users, a day is divided into several time periods. There are six time periods in this embodiment, namely 0:00~4:00, 4:00~8:00, 8:00~12:00, 12:00~16:00, 16:00~20 in a day :00 and 20:00~24:00. It should be noted that this classification is just an example, and other classifications (different number of time periods) based on the sales characteristics of goods and services are also possible. Therefore, it can be seen that the online period of the user token corresponding to the digital footprint appears. For example, the digital footprint of serial number 1 corresponds to the user token 00FO_0NNF, which appears during the online period from 0:00 to 4:00 (trigger time is 2021/11/15 00:16:16). The second operation of the online period analysis module 220 is to count the online periods in which each user token appears the most in a first period of time. The first time period and the time periods mentioned below are based on a specific purpose to distinguish the time during which the system operates. Each time period may overlap or may not intersect with each other, depending on the needs of the application. In this embodiment, the first time period is for the manufacturer to collect information on a specific event day, starting at 2021/11/15 00:00:00 and ending at 2021/11/17 00:00:00, a total of two days. In practice, the first time period may also be as long as a week, a month, or even years. In Figure 3, the user represented by the user token b12e-_4ZPG has been online three times in two days, and the online periods are 0:00~4:00, 12:00~16:00 and 12:00~16: 00. Since the online time period is 12:00~16:00 the most, the online time period analysis module 220 determines that the online time period when the user token b12e-_4ZPG appears the most is 12:00~16:00.

The consumption behavior analysis module 230 is a technical component that actually analyzes the user's consumption behavior, and performs the following operations. The first operation is to calculate the content of the relevant attribute data for each user token in a second period of time, including the average amount of each consumption amount as a unit price per customer. Here, the second time period is independent of the first time period, and is to analyze the actual consumption amount of each user within a certain time period, so it is completed by tracking the average value of the consumption amount attribute in the object corresponding to the user token. In this embodiment, the second time period is set to end from 2021/11/15 00:00:00 to 2021/11/16 00:00:00. For the user token ccf2-_5RDF, the unit price is 12999 yuan because it is consumed once; for the user token b12e-_4ZPG, the user consumes two amounts, 10000 yuan and 299 yuan, so the unit price per customer is 5150 yuan. The second operation of the consumption behavior analysis module 230 is to calculate the content of the relevant attribute data for each user token in the second time period, and the discount rate for each consumption amount is an average discount. Compared with the unit price, the discount rate of user token ccf2-_5RDF consumption is 13.33%, and the average discount is 13.33%. For the user token b12e-_4ZPG, the discount on consumption is 0% and 40%, so the average discount is 20%.

The data grouping module 240 is a technical component that performs grouping analysis on the consumer behavior analysis module 230, and performs the following operations. The first operation is to count the unit price and average discount related to each user token in a third time period. Here, the third time period is a continuous or discontinuous time period based on the second time period. If there are many second time periods, such as 2021/11/01 00:00:00 to 2021/11/02 00:00:00, 2021/11/04 00:00:00 to 2021/11/07 00: 00:00 and 2021/11/15 00:00:00 to 2021/11/16 00:00:00, the third time period can include all the second time periods, or one or both of them, and then in In the third time period, the unit price and average discount related to each user token are counted. The second operation of the data grouping module 240 is to use a grouping algorithm to group all relevant customer unit prices of all user tokens into one of several customer price groups for each user token. Grouping is to group all data, group similar data together, each piece of data can have a group, and each group is called a cluster. Therefore, each user token will be assigned to a customer price group based on its related customer price. Similarly, the third job of the data grouping module 240 is to use the grouping algorithm to group all relevant average discounts for all user tokens into one of several average discount groups for each user token. According to the present invention, the aforementioned clustering algorithm may be a k-means clustering algorithm or a Jenks natural breaks optimization algorithm.

The analysis result processing module 250 is a technical component that manages the output of the operation results of the system, and can analyze the online period in which the specific user token appears the most, the customer unit price, the average discount, and the grouping result of the data grouping module 240 through the specified user token. The API is uploaded to the specified cloud platform. Here, because the system is a business application, the cloud platform is usually a cloud receiving hardware device designated by the merchant side of the system, such as a customer relationship management server on the client side.

The data labeling module 260 can, through a correspondence table with several specified attribute data contents corresponding to the label text, find out the attribute data contained in the digital footprint received by the platform monitoring module 210 each time, and find the corresponding data corresponding to the corresponding table. at least one specified attribute data content, and the label text corresponding to the at least one specified attribute data content is concatenated with the user token contained in the digital footprint. For a better understanding of this, please refer to FIG. 4 , which shows the aforementioned correspondence table and the label texts concatenated by the correspondence table for the first six serial numbers in FIG. 3 . In FIG. 4 , the content of the attribute data is taken as an example of the discount rate for description. When the discount rate is 0~10%, the label text is low discount; when the discount rate is 10~20%, the label text is medium; when the discount rate is more than 20%, the label text is high discount. Therefore, the discount rate of user token 00FO_0NNF is 80%, and its concatenated label text is high discount attractiveness; the discount rate of user token b12e-_4ZPG is 0, and its concatenated label text is low discount attractiveness. The aforementioned operation results can also be transmitted to a specified cloud platform through a specified application programming interface through the analysis result processing module 250 .

In order to dynamically analyze the consumption behavior of the user, the consumption behavior analysis module 230 may further perform the following operations. The third operation of the consumption behavior analysis module 230 is to calculate a last online time interval and an online frequency according to the trigger time of the content of the relevant attribute data for each user token in a fourth time period, and accumulate consumption lump sum. For a better understanding of this, please refer to FIG. 5 and FIG. 6 . FIG. 5 lists attribute data of user tokens and corresponding objects received by the platform monitoring module in a fourth time period in another embodiment, and FIG. 6 discloses a classification table and the data in FIG. 5 according to the classification table The result of the classification performed. In Figure 5, the fourth time period is a time period selected from a record history according to the needs of the users of the system, such as 2021/11/02 00:00:00 to 2021/11/03 00:00:00 . Of these, 7 user tokens have transaction history. According to the aforementioned calculation method, the consumption behavior analysis module 230 calculates the last online time interval, online frequency and total consumption for each user token. Taking the user token ccf2-_5RDF as an example, its last online time is 21:03:17, so the last online time interval (from the calculation base point 2021/11/03 00:00:00) is 02: 56:43 (In practice, if the fourth time period lasts for several weeks, "day" can be used as the calculation interval), the online frequency is 3 times, and the total consumption is 100+3500+1000, totaling 4600 yuan. The fourth operation of the consumption behavior analysis module 230 is to calculate the median of the last online time interval, online frequency and total consumption of all user tokens in the fourth time period. Regarding the calculation results of the fourth job, the table is shown in FIG. 6 . The median of the last online time interval was 02:42:14, the median of the online frequency was 3 times, and the median of the total consumption was 4,600 yuan. The fifth operation of the consumption behavior analysis module 230 is to determine, for each user token, whether the last online time interval, online frequency and total consumption are above or below the median using the medians as standards. The sixth operation of the consumption behavior analysis module 230 is to classify into 8 categories according to the combination of the judgment results, and to classify each user token as one of them. In this embodiment, the classification table lists the classification results. For example, when it is above the median of the last online time interval, the median of the online frequency, and the median of the total consumption, it is classified as category 1. Therefore, the classification result table of all user tokens is listed below the classification table in FIG. 6 . When the classification result is released, the analysis result processing module 250 can transmit the classification result of the specific user token to the designated cloud platform through the designated application program interface for analysis and use by the merchants using the system.

According to the present invention, the digital footprint of each user can be labeled with different labels at the same time, that is, the user token possessed by the user can be concatenated with more than one label text at the same time. In order to know the labelled status of each user's digital footprint, the data grouping module 240 may further perform the following operations. The fourth operation of the data grouping module 240 is to count the concatenation times of a specified label text concatenated by each user token in a fifth time period. The fifth time period here is also a specified period of time when the merchants of the system are interested, but the processing data are the same or different from the aforementioned time period. The fifth operation of the data grouping module 240 is to use the number of times of concatenation of the specified label text of all user tokens and the number of times of concatenation of the specified label text of a specified user token to use the grouping algorithm to assign the specified user token. The cards are grouped. For example, for the specified label text "consumption amount is high" in the fifth time period, the number of times of concatenation of user token ccf3-_5RDF is 5 times, the number of times of concatenation of user token p291_9IIJ is 5 times, and the number of times of concatenation of user token p291_9IIJ is 5 times. The concatenation times of qu2d_4Z9M is 20 times, the concatenated times of user token bp2w_9WER is 7 times, the concatenated times of user token b125-_4ZPG is 86 times, the concatenated times of user token zl8p_0SXZ is 33 times, the concatenated times of user token The number of concatenation times of n543_9KLM is 17 times. All user tokens and the designated user token b125-_4ZPG are grouped by the grouping algorithm, and the designated user token is divided into the first group. That is to say, even if the users corresponding to these user tokens are high-consumption groups, there are orders of magnitude difference in the number of consumptions, and further distinguishing them is helpful for accurate marketing.

When a certain user token is concatenated with a label text, the user to which the user token belongs has a relative "consumption" attribute, and the user token can be further precisely marketed. According to the present invention, the analysis result processing module 250 can transmit all the user tokens concatenated with at least one specified label text in a sixth time (the period during which the merchants in the system need to carry out precise marketing) to the specified user token through the specified application program interface. cloud platform, whereby the designated cloud platform sends newsletters, email marketing (EDM) or advertisements to those user tokens. Here, the cloud platform is also the cloud receiving hardware device designated by the merchant.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the technical field can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, The protection scope of the present invention shall be determined by the scope of the appended patent application.

10: Servo host 110: Network communication interface 210: Platform Monitoring Module 220: On-line time period analysis module 230: Consumer Behavior Analysis Module 240: Data Grouping Module 250: Analysis result processing module 260:Data Labeling Module 30: Internet 410: Web Management Server 420: IM Management Server 430: Customer Relationship Management Server

FIG. 1 is a simplified block diagram of a system for analyzing user behavior in an information exchange platform according to an embodiment of the present invention.

Figure 2 shows the architecture of the digital footprint of the transfer.

Figure 3 tabulates the digital footprint transmitted by the web management server.

FIG. 4 shows a correspondence table and the label texts concatenated by the correspondence table for the first six serial numbers in FIG. 3 .

FIG. 5 lists user tokens and attribute data in corresponding objects received by the platform monitoring module in a fourth time period in another embodiment.

FIG. 6 discloses a classification table and the classification results according to the classification table and the data in FIG. 5 .

10: Servo host

110: Network communication interface

210: Platform Monitoring Module

220: On-line time period analysis module

230: Consumer Behavior Analysis Module

240: Data Grouping Module

250: Analysis result processing module

260:Data Labeling Module

30: Internet

410: Web Management Server

420: IM Management Server

430: Customer Relationship Management Server

Claims (9)

A system for analyzing user behavior in an information exchange platform is installed in a server host, and includes: A platform monitoring module continuously receives a plurality of digital footprints sent by a management server executing a service of an information exchange platform, each digital footprint includes a user token and an object operated through the user token, wherein the object Contains a plurality of attribute data that can be changed by operation; As soon as the analysis module goes online, perform the following operations: For each digital footprint, according to a trigger time included in the content of the attribute data, confirm the online time period that the corresponding user token appears in a plurality of online time periods in a day; and Statistics in a first time period, the online period in which each user token appears the most; A consumer behavior analysis module that performs the following tasks: During a second period of time, the content of calculating the relevant attribute data for each user token includes the average amount of each consumption amount as the unit price per customer; and In the second time period, the content of calculating the relevant attribute data for each user token includes a discount rate for each consumption amount as an average discount; A data grouping module that performs the following operations: Statistics on the unit price and average discount related to each user token in a third time period; All relevant customer unit prices for all user tokens are grouped for each user token into one of several customer price groups using a grouping algorithm; and grouping all relevant average discounts for all user tokens into one of several average discount groups for each user token using the clustering algorithm; and An analysis result processing module, which transmits the most-appearing online period of the specific user token, the customer unit price, the average discount, and the grouping result of the data grouping module to the designated cloud platform through the designated application program interface. The system for analyzing user behavior in an information exchange platform according to claim 1, wherein the consumption behavior analysis module further performs the following operations: For each user token in a fourth time period, calculate a last online time interval and an online frequency through the trigger time of the content of the relevant attribute data, and accumulate the total consumption; Calculate the median of the last online time interval, the online frequency and the total consumption of all user tokens in the fourth time period; For each user token, use the medians as the criteria to determine whether the last online time interval, online frequency and total consumption are above or below the median; and According to the combination of judgment results, it is divided into 8 categories, and each user token is classified as one of them. The analysis result processing module further transmits the classification result of the specific user token to the designated cloud platform through the designated application programming interface. The system for analyzing user behavior in an information exchange platform as described in claim 1, further comprising a data labeling module, through a correspondence table having a plurality of specified attribute data contents and label texts, on the platform From the attribute data contained in the digital footprints received by the monitoring module each time, find out at least one specified attribute data content that matches the corresponding table, and compare the label text corresponding to the at least one specified attribute data content with the user contained in the digital footprint. Token concatenation. The system for analyzing user behavior in an information exchange platform according to claim 3, wherein the data grouping module further performs the following operations: Counting the concatenation times of a specified label text concatenated by each user token in a fifth time period; and The designated user tokens are grouped by the grouping algorithm according to the concatenation times of the designated label text of all user tokens and the concatenation times of the designated label text of a designated user token. The system for analyzing user behavior in an information exchange platform according to claim 3, wherein the analysis result processing module further processes all user tokens concatenated with at least one specified label text within a sixth time period through the The designated API is passed to the designated cloud platform, whereby the designated cloud platform sends a newsletter, email marketing (EDM) or advertisement to these user tokens. The system for analyzing user behavior in an information exchange platform according to claim 1, wherein the information exchange platform is an interactive web page, an instant messaging platform or a customer relationship management platform. The system for analyzing user behavior in an information exchange platform according to claim 6, wherein if the information exchange platform is an instant messaging platform, the content of the attribute data is the name of the clicked data box and the specific sentence provided by the instant messaging platform or trigger time. The system for analyzing user behavior in an information exchange platform as described in claim 6, wherein if the information exchange platform is an interactive web page, the content of the attribute data is the name of the clicked object, the trigger time, the description text or the specific text. The value of the unit. The system for analyzing user behavior in an information exchange platform according to claim 1, wherein the clustering algorithm is a k-means clustering algorithm or a Jenks natural breaks optimization algorithm.

Images