JP4938725B2 - User behavior analysis method and user behavior analysis system - Google Patents

Description

  The present invention relates to a user behavior analysis method and a user behavior analysis system.

  In online businesses, understanding user behavior is extremely important. In terms of usability, analysis of user behavior is an important source for user feedback for online services. In terms of marketing, the user trend, that is, the above-described user behavior analysis is a measure for understanding the user's needs and requirements. Furthermore, the user behavior analysis is a basis for solving the resource allocation problem in the system management aspect.

  In the conventional technology, various performance indexes are used as a method for analyzing such user behavior. That is, the user's behavior is analyzed by a performance index such as the number of page views, the number of unique visitors, or the number of queries. Most of these performance indicators can be obtained by counting the user behavior that occurs on each page on the web.

  FIG. 1 is an example for explaining a conventional method of analyzing user behavior. That is, in the conventional technology, the number of occurrences is counted each time a user activity 102 on the website such as login, game, purchase, event participation, etc. of the user 101 occurs, and the game index, sales index, web usage index are counted. Various indexes 103 such as these can be measured, and such an index 103 can be used to analyze the behavior of the user 101 using the above-described performance index.

  However, the conventional user behavior analysis method has a problem that it is difficult to understand the detailed activity on the user's website that causes the performance index. For example, with the above-mentioned indices such as the game index, sales index, and web usage index alone, what changes will occur in the behavior pattern after the user has used all of the game money on the game site, or on a specific purchase route It is difficult to understand the detailed activities of the user, such as whether a change in sales occurs. The difficulty in understanding the detailed activity of the user, that is, the difficulty in analyzing the user behavior leads to a problem that makes it difficult to improve the service of the website using the performance index or the like.

  The present invention has been devised to solve the above-described problems, and proposes a new technique relating to a user behavior analysis method and a user behavior analysis system.

  Therefore, the present invention provides a user behavior analysis method and user behavior analysis that can easily and easily perform time-series analysis on user behavior by arranging user behavior by user or time order via a user log. The purpose is to provide a system.

  In addition, the present invention analyzes the user's behavior in a time series to analyze the cause of the usage status of the website, which is difficult to grasp with existing technologies such as the purchasing route and the member behavior pattern for a specific case. It is an object of the present invention to provide a user behavior analysis method and a user behavior analysis system that are capable of doing so.

  In addition, the present invention provides a user behavior analysis method and a user behavior analysis system capable of understanding a user's behavior by analyzing the cause of the current usage status of the website and thereby improving the website service. With the goal.

  Furthermore, the present invention can solve the problem of traffic increase by using a plurality of collection servers included in the log analysis server, and can shorten the analysis time for user behavior by distributed processing using a plurality of computer clusters. An object is to provide a user behavior analysis method and a user behavior analysis system.

  In order to achieve the above object and solve the above-mentioned problems of the prior art, a user behavior analysis method according to an embodiment of the present invention collects user logs via a log collection server LCS (Log Collection Server). Converting the user log into a flow field format to generate a flow field set, and processing the flow field set in response to an input of a processing command for the flow field set. At this time, the flow field format is a predetermined format of the flow field, and the flow field is classified into a user identification field for identifying a user and an action record field for recording the action of the user or the system, The attribute includes an attribute name that classifies the attribute and an attribute value that defines a user's action.

  According to one aspect of the present invention, the step of collecting a user log via a log collection server includes the step of receiving the user log via a web page provided by a web server, and a URL included in the user log. And changing to ST name (Service Tree name) using STDB (Service Tree Data Base). At this time, the step of receiving the user log through a web page provided by a web server receives the user log using a user's web browser and a Javascript tag, or collects the log in the case of a client. Receiving the user log directly at a server;

  According to another aspect of the present invention, the step of converting the user log into a flowing field format to generate a flowing field set includes converting each of the user logs into the flowing field format to generate a respective flowing field. And generating the flow field set by arranging the flow fields in chronological order according to users.

  According to still another aspect of the present invention, the processing instruction includes (1) a draw operator that extracts a flow field in an input time range, and (2) a flow field that conforms to an input pattern. And a set operator that measures the total or number of the attributes according to the input attributes, and the step of processing the fluid field set by inputting a processing instruction for the fluid field set (1) The corresponding flow field is extracted by the draw operator or the filter operator, or (2) the attribute corresponding to the set operator is selected from the flow field set or the extracted flow field. It is the step of measuring the total or number.

  In another embodiment of the present invention, a user behavior analysis system includes: a log collection server that collects user logs; a fluid field set generation unit that generates a fluid field set by converting the user log into a fluid field format; A fluid field set processing unit for processing the fluid field set by inputting a processing command for the fluid field set.

  ADVANTAGE OF THE INVENTION According to this invention, the time series analysis with respect to a user's action can be performed easily and simply by arranging a user's action according to a user or a time order via a user log.

  In addition, according to the present invention, by analyzing the user's behavior in time series, the cause of the current usage status of the website, which is difficult to grasp with existing technology such as the purchase route and the member behavior pattern for a specific case, is analyzed. It becomes possible to do.

  Further, according to the present invention, it is possible to understand the user's behavior by analyzing the cause of the current usage status of the website, thereby improving the website service.

  Furthermore, according to the present invention, the problem of traffic increase is solved by using a plurality of collection servers that can be included in the log analysis server, and analysis time for user behavior is shortened by distributed processing using a plurality of computer clusters. can do.

  Hereinafter, various embodiments according to the present invention will be described in detail with reference to the accompanying drawings.

  In web services, patterns of user behavior, more specifically sequences of user behavior, can be used in many ways. For example, depending on user behavior, how a visitor connects to a given page on the site, what is the most frequently used route for the user to connect to the page, and the user purchases the product “A” After that, it is possible to obtain time-series information such as what the purchased product is.

  Since such statistical material is obtained by tracking the behavior of each user, it is different from the above-described performance index in the prior art such as simply measuring the number of visitors who entered each page. Such statistical material can be collected via user logs, ie web logs.

  FIG. 2 is an example for generally explaining the user behavior analysis method according to the present invention.

  Here, FIG. 2 shows a user action 201 defined by an analysis purpose and a scenario such as login, game, all-in, purchase, logout, a user “A” 202 and a user “B” on a web server that provides a game. Each user's action with respect to 203 is shown. Referring to FIG. 2 in detail, the user “A” 202 can know that the game has progressed after login and that the user has logged out after all-in of the game money, and the user “B” 203 is the same as the user “A” 202. It is possible to know that the player has taken action, purchased the game money after the game money all-in, played the game again, and then logged out.

  That is, if the user's actions can be known for each user or each time (in order of time) using the user log, the user's actions can be understood. In other words, it is difficult to understand the detailed behavior of the user simply by the number of logins, the number of purchases, and the purchase price, but it is possible to know not only the various outcome indicators 204 such as the game index and sales index but also the order of user behavior. If it is possible, the detailed behavior of the user can be easily analyzed and understood.

  For this reason, it is necessary to know the user behavior of the user and the attributes at the time of the user behavior, and such information can be acquired via the user log.

  FIG. 3 is an example for generally explaining the user behavior analysis method according to the present invention.

  310 indicates the time points when four users purchase game money. However, in 310, only the time when game money is purchased is shown, and the behavior before and after the purchase behavior cannot be known at all. For this reason, only the sales amount for each user, that is, only result-centric data is obtained.

  On the other hand, 320 can obtain behaviors executed at each time point for four users, that is, time-series information of user behaviors. Then, it is possible to obtain the user behavior pattern for each situation of “game” again, and by analyzing this, it becomes possible to grasp the change between the user behavior and the attribute.

  In this way, when it is possible to check user behavior by user or in time order, it is possible to grasp the cause of the user behavior, thereby making it possible to analyze the purchase route, which is difficult to grasp only with the conventional performance index, and the specific case It is possible to analyze user behavior patterns after an (event) occurs. Such a method and system will be described in more detail with reference to FIGS.

  FIG. 4 is a flowchart illustrating a user behavior analysis method according to an embodiment of the present invention.

  In step S410, the user behavior analysis system collects user logs via the log collection server. Here, the user log is an action such as a purchase of game money generated by an event such as a mouse click from the user or a keyboard input, an action such as entering a game, an action of a system generated on a web server by an all-in of game money, or the like. Can contain all. In other words, the user behavior analysis system can collect user logs without requiring a separate behavior from the user via the log collection server (without the user being aware).

  At this time, the user log may include a web server log generated by the web server, and the step of collecting the user log via the log collection server generates the web server log by the web server included in the log collection server. Configured as a step. That is, the user behavior analysis system can include a web server, and the web server log generated by the web server can be used as the user log.

  The user behavior analysis system can also be configured to collect user logs via the user's browser by the log collection server. In this case, as illustrated in FIG. 4, the user behavior analysis system can be executed including steps S411 to S413 in order to collect user logs.

  In step S <b> 411, the user behavior analysis system manages and maintains the STDB by using a service tree management system (STMS). Here, the STDB stores a URL and an ST name corresponding to the URL in association with each other. In other words, the STDB includes various URLs related to the website provided to the user when the user accesses through the web server, and stores the ST name related to the URL in association with the URL.

  In step S412, the user behavior analysis system receives a user log through a web page provided by the web server. That is, the user behavior analysis system accesses the web server via the user or the web server, and is defined by the analysis purpose and scenario such as meaningful behavior, ie, login, logout, game money purchase, or all-in processing procedure. The generated user log can be received when the performed user action occurs. At this time, the user behavior analysis system can receive the user log by using the user's web browser and JavaScript tag, or can receive the user log directly by the log collection server in the case of the client.

  In step S413, the user behavior analysis system changes the URL included in the user log to the ST name using the STDB. This is for easily grasping the route through which the user action occurs via the URL, and for removing unnecessary elements from the URL. Also, since the analyst cannot easily grasp the web page related to the URL via the URL, the URL is changed to the ST name so that the analyst can easily grasp the result of the user behavior analysis system. You can also

  In step S420, the user behavior analysis system converts the user log into a flow field format to generate a flow field set. At this time, the fluid field format is a predetermined format of the fluid field, and the fluid field includes a field name and a plurality of attributes.

  The flow field is classified into a user identification field for identifying the user and an action record field for recording the action of the user or the system. The user identification field is divided into (1) a login field indicating a user classified by login name and (2) a permanent cookie field indicating a user classified according to a permanent cookie assigned for each browser. being classified. The action record field includes (1) a page view field indicating a visit to a specific page of the user, (2) a click field indicating a click on the specific link of the user, and (3) an action field indicating a system action for the user. are categorized. At this time, the user identification field and the action record field can be configured to be classified into more detailed fields as necessary.

  The attribute is composed of an attribute name that classifies the attribute and an attribute value that defines the user's behavior.

  Note that the flow field can include the occurrence time of the user action included in the user log as an attribute. That is, all the flow fields can include the occurrence time of the user action associated with the flow field.

  Step S420 for generating such a fluid field set will be described in more detail with reference to the flowchart of FIG. 6 and the examples of FIGS.

  In step S430, the user behavior analysis system processes the fluid field set by inputting a processing command for the fluid field set. The flow field set can be divided and merged as needed, which will be described in more detail with reference to FIG. Further, the processing instruction can include (1) a typical analysis question (analysis query) and (2) a predetermined operator or analysis script.

  If the processing instruction is a typical analysis question, in step S430 of processing the flow field set, the user behavior analysis system applies the corresponding flow based on the time range, pattern, statistical function, or group division criterion included in the analysis question. A field can be extracted and a processing result for an attribute included in the flow field can be provided.

  The processing instruction is input to a predetermined operator, that is, (1) a draw operator that extracts a flow field in the input time range, and (2) a filter operator that extracts a flow field that matches the input pattern. And a set operator that measures the total or number of the attributes based on the attribute, in step S430, the user behavior analysis system (1) extracts a corresponding flow field based on the draw operator or the filter operator. (2) It is possible to measure the total or number of attributes corresponding to the set operator in the flow field set and the extracted flow fields.

  Finally, when the processing instruction is an analysis script, in step S430, the user behavior analysis system extracts (1) the corresponding flow field according to the time range, pattern, or attribute processing instruction according to the input analysis script, (2) It is possible to measure the total or number of attributes corresponding to the processing command among the attributes included in the fluid field set or the extracted fluid field.

  As described above, the method for processing the flow field set will be described in more detail with reference to an example of FIG.

  FIG. 5 is an example for explaining a user log collection method according to the present invention.

  A user can request a web page from the web server 502 via the browser 501, and the web server 502 can send the web page browser 501 including a Javascript tag. That is, the web server 502 can set a login permanent cookie in the browser 501 via the JavaScript tag, and the user behavior analysis system receives the user log via the URL of the log collection server 503 included in the login permanent cookie. (collect.

  Thereafter, when a user action (user action) occurs via the web page, the log collection server 503 uses the URL included in the login permanent cookie with the Javascript tag to store the user log transmitted via the browser 501. Can be received. At this time, the log collection server 503 can identify the user by a method of identifying the browser 501 when the user does not log in. For this purpose, the log collection server 503 can set a permanent cookie in the browser 501. That is, the permanent cookie can be used to identify a user who does not log in via the browser 501.

  FIG. 6 is a flowchart illustrating a method for generating a flow field set in an embodiment of the present invention. As shown in FIG. 6, steps S601 and S602 are included in step S420 described with reference to FIG. 4 and executed.

  In step S601, the user behavior analysis system converts each user log into a flow field format and generates a flow field. That is, the user log includes user behavior, and the user behavior analysis system can convert such a user log into a fluid field format that can be used by the user behavior analysis system. As described above, since the fluid field format is a predetermined format of the fluid field, when the user log is converted to the fluid field format, the user log is converted as a fluid field. That is, a flow field can be generated.

  In step S <b> 602, the user behavior analysis system generates a flow field set by arranging the flow fields in time order by user. Since each of such flow fields does not have a large meaning as the flow field itself, the flow field set is generated by aligning the occurrence time of the user action included in the flow field with the user as a reference. Such flow field set and flow field will be described in more detail with reference to FIGS.

  FIG. 7 is an example for explaining a flow field set and a flow field according to the present invention. FIG. 7 shows a plurality of flow fields arranged in the order of the generation time 702 of each user 701 or user action. At this time, the flow field includes a field name indicating the type of the field and an attribute based on the user's action. An attribute is composed of an attribute name and an attribute value, and may be composed of a set of a plurality of attribute names and attribute values.

  That is, in the example of FIG. 7, the field name “L” 704 included in the flow field 703 is a login name, which is a field for classifying users, and when a user logs in to a website via a browser, Generated through the generated user log. The attribute “id = xxxxxxxx ts = 116163168755” 705 includes an attribute name “ts” 706 and an attribute value “116316168755” 707. In the example of FIG. 7, a flow field 703 having two attributes is shown. At this time, the first attribute may mean that the login ID of the user is “xxxxxxx”, and the second attribute may mean that the login occurrence time is “116316168755”.

  At this time, since one row of the flow field set is an arrangement of user actions for one user in time order (time-series order), the first element of the row, that is, the first attribute is This is an attribute for identifying the user.

  FIG. 8 is an example of a flow field set according to the present invention. Here, reference numeral 800 indicates a plurality of flow fields having a plurality of attributes. That is, the example of FIG. 8 includes user behavior from the moment when a user having a login ID “xxxxxxxx” logs in to the game server (at the time of login) to the moment of logout (at the time of logout).

  FIG. 9 is an example for explaining division and merging between flow field sets according to the present invention.

  As described above, the flow field set is an enumeration of user actions that are a series of corresponding information sorted by user or in time order. When trying to analyze such a flow field set, there is a case where a user's one-week user behavior pattern is analyzed or a period of one month or more is analyzed. In such a case, a flow field set must be generated and collected for the user for a period of one month or more. However, for storage or technical reasons, such a whole flow is set as one flow. Rather than generating as a set, it is preferable to dynamically generate a flow field set for an arbitrary period.

  That is, it is possible to generate a plurality of fluid field sets for a user in a short time unit, and combine the fluid field sets by using a join operator or a union operator as necessary. It is possible to obtain a flow field set for the user's behavior within a unit of time. For example, in the flow field sets 901 and 902, the flow field set 903 is generated by merging via the join operator, or the flow field sets 901 and 903 are merged via the union operator to generate the flow field set 905. can do. In other words, the merged flow field sets 903 and 905 can be generated by merging the flow field sets 901, 902, and 904.

  On the contrary, it is also possible to divide (divide) a large unit flow field set into small unit flow field sets.

  The user behavior analysis system includes (1) a processing command input in the form of a typical analysis question, (2) a processing command including a predetermined operator for easily processing a typical analysis question, 3) A flow field set can be processed by analyzing a processing instruction using an analysis script that can be expressed in a typical form of analysis and question while including a predetermined operator. The relationship between such typical analysis questions, predetermined operators, and analysis scripts will be described in more detail with reference to FIG.

  FIG. 10 is an example of processing a flow field set according to the present invention.

  First, an example of a method for processing a fluid field set through processing instructions including predetermined operators described above in the user behavior analysis system, that is, a draw operator, a filter operator, and a set operator will be described.

  First, when a draw operator 1001 is input, a record of user behavior within a predetermined time range included in the draw operator 1001, that is, a flow field corresponding to the predetermined time range is extracted from the flow field set 1002. be able to. In addition, when the filter operator 1003 is input, a flow field corresponding to the pattern included in the filter operator 1003 can be extracted from the extracted flow field. Finally, for the set operator 1004, the user behavior analysis system provides the analyst who measures the total or number of attributes corresponding to the attributes included in the set operator in the flow field extracted by the filter operator 1003. can do.

  At this time, as shown in FIG. 10, for example, when only the draw operator 1001 and the filter operator 1003 are input by the input operator, the flow field extracted after executing the filter operator 1003 Can also be provided to the analyst. In addition to this, it is possible to continuously execute the same operator under different conditions or repeatedly execute the same operator under the same conditions. That is, FIG. 10 shows an example of processing a flow field set, and various modifications are possible depending on the order of operators input from an analyst and the values such as the time range or pattern included in the operators.

  FIG. 11 is an example for explaining a processing instruction according to the present invention.

  As described above, the user behavior analysis system includes (1) a processing command input in the form of a typical analysis question and (2) a predetermined operator for easily processing the typical analysis question. It is possible to process the flow field set by either one of a processing instruction and (3) a processing instruction using an analysis script that can be expressed in the form of a typical analysis question that is a predetermined operator.

  1110 of FIG. 11 is an example of a typical analysis question. Such a typical analytical question is a filter operator in which processing for a time range is expressed by “regex” through the predetermined operator described above, that is, through a draw operator expressed by “draw”. A process for a pattern can be processed through, or a statistical function or a group partition criterion can be processed through a set operator represented by “aggr”.

  Reference numeral 1120 in FIG. 11 shows an example in which typical analysis questions are processed using the draw operator 1121, the filter operator 1122, and the set operator 1123. At this time, reference numerals 1124 to 1126 indicate processing results obtained by the draw operator 1121, the filter operator 1122, and the set operator 1123, respectively.

  Reference numeral 1130 in FIG. 11 is an example in which typical analysis questions are expressed by the above-described analysis script. The analysis script includes a predetermined operator generated to easily process a typical analysis question, but expresses it in the form of a typical analysis question outside the mechanical framework of the predetermined operator. This is a script generated so that humans can easily recognize it.

  That is, the user behavior analysis system can process the fluid field set by a processing instruction including any one of the above-described typical analysis question, predetermined operator, or analysis script.

  As described above, by using the user behavior analysis method according to the present invention, it is possible to easily and simply perform time-series analysis on user behavior by arranging user behavior by user or in time order via a user log. In addition, by analyzing the user's behavior in time series, it becomes possible to analyze the cause of the current usage of the website, which has been difficult to grasp, such as the purchase route and the membership behavior pattern for a specific incident. .

  In addition, by analyzing the cause of the usage status of the website, it is possible to understand the user's behavior and thereby improve the website service. By using multiple collection servers included in the log analysis server, traffic increases. The analysis time for the user's action can be shortened by distributed processing using a plurality of computer clusters.

  FIG. 12 is a block diagram for explaining the internal configuration of the user behavior analysis system in the embodiment of the present invention. As illustrated in FIG. 12, the user behavior analysis system 1200 may include a log collection server 1210, a fluid field set generation unit 1220, and a fluid field set processing unit 1230.

  The log collection server 1210 collects user logs. Here, the user log indicates game money purchases caused by events such as clicks from the user, keyboard input, etc., actions such as game admission, system actions that occur on the web server due to game money all-in, etc. All can be included. In other words, the log collection server 1210 can collect user logs without requiring a user to take separate actions. At this time, as illustrated in FIG. 12, the log collection server 1210 can include a user log reception unit 1211, a URL change unit 1212, an STDB 1213, and an STMS 1214 in order to collect user logs.

  The user log receiving unit 1211 receives a user log via a web page provided by a web server. That is, the user log receiving unit 1211 accesses (connects) the web server via the user or the web server, and performs meaningful actions of the user, that is, login, logout, purchase of game money, or all-in processing procedures. Thus, when the user behavior defined by the analysis purpose and the scenario occurs, the generated user log can be received. At this time, the user log receiving unit 1211 can receive the user log using the user's web browser and Javascript tag, or, in the case of a client, can directly receive the user log at the log collection server. In other words, in the user log, the web browser and the Javascript tag detect a user action that matches the defined user action, generate a corresponding user log, and transmit the generated user log to the user log receiving unit 1211. .

  The URL changing unit 1212 changes the URL included in the user log to an ST name using the STDB 1213. That is, referring to the URL associated with the ST name stored in the STDB 1213, the ST name corresponding to the URL included in the user log is extracted. This is for easily grasping the route through which the user action occurs via the URL, and for removing unnecessary elements from the URL. Also, since the analyst cannot easily grasp the web page related to the URL via the URL, the URL is changed to the ST name so that the analyst can easily grasp the result of the user behavior analysis system. You can also.

  The STMS 1214 manages and maintains the STDB 1213. Here, the STMS 1214 stores a URL and an ST name corresponding to the URL in association with each other. That is, the STDB 1213 includes various URLs related to the website provided to the user when the user accesses via the web server, and stores the ST name related to the URL in association with the URL.

  The fluid field set generation unit 1220 converts the user log into a fluid field format to generate a fluid field set. At this time, the fluid field format is a predetermined format of the fluid field, and the fluid field includes a field name and a plurality of attributes.

  The flow field is classified into a user identification field for identifying the user and an action record field for recording the action of the user or the system. The user identification field is divided into (1) a login field indicating a user classified by login name and (2) a permanent cookie field indicating a user classified according to a permanent cookie assigned for each browser. being classified. The action record field includes (1) a page view field indicating a visit to a specific page of the user, (2) a click field indicating a click on the specific link of the user, and (3) an action field indicating a system action for the user. are categorized.

  Further, the attribute is configured with an attribute name for classifying the attribute and an attribute value for defining the user's behavior.

  Note that the flow field can include the occurrence time of the user action included in the user log as an attribute. That is, all the flow fields can include the occurrence time of the user action associated with the flow field.

  Such a fluid field set generation unit 1220 may include a fluid field generation unit 1221 and an alignment unit 1222, as shown in FIG.

  The fluid field generation unit 1221 converts each acquired user log into a fluid field format to generate each fluid field. That is, the user log includes user behavior, and the user behavior analysis system converts such a user log into a fluid field format that can be used by the user behavior analysis system. As described above, the fluid field format is a predetermined format of the fluid field, and when the user log is converted into the fluid field format, the user log is converted into the fluid field. That is, the flow field is generated from the user log.

  The aligning unit 1222 sorts the flow fields by user in time order to generate a flow field set. Since each of such flow fields does not have a large meaning as the flow field itself, the flow field set is generated by being aligned based on the occurrence time of the user action included in the flow field and the user.

  The fluid field set processing unit 1230 processes the fluid field set by inputting a processing command for the fluid field set. The flow field set can be divided and merged (integrated) as required, and the processing instructions can include (1) typical analysis questions and (2) predetermined operators or analysis scripts.

  First, when the processing instruction is a typical analysis question, the flow field set processing unit 1230 that processes the flow field set corresponds to the time range, pattern, statistical function, or group division criterion included in the analysis question. A flow field can be extracted and a processing result for an attribute included in the flow field can be provided.

  Second, the processing instruction is a predetermined operator, that is, (1) a draw operator that extracts a flow field within a predetermined time range that has been input, and (2) a flow field that matches the input pattern. In the case of including a filter operator and (3) a set operator that measures the total or number of attributes according to the input attributes, the fluid field set processing unit 1230 corresponds to (1) a draw operator or a filter operator. Or (2) the total or number of attributes corresponding to the set operator in the flow field set or the extracted flow fields can be measured.

  Finally, when the processing instruction is an analysis script, the fluid field set processing unit 1230 extracts (1) the corresponding fluid field by the time range, pattern, or attribute processing instruction according to the input analysis script, (2) It is possible to measure the total or number of attributes corresponding to the processing command among the attributes included in the fluid field set or the extracted fluid field.

  Further, in the user behavior analysis system 1200, the log collection server can be configured by a plurality of collection servers for distributed data processing, and the user log can be configured to be distributed and collected by a plurality of collection servers. it can. In this case, the fluid field set processing unit 1230 may include a plurality of computer clusters (not shown) that process the fluid field set.

  That is, the flow field for the same user can be processed according to the processing command in the same computer cluster designated among the plurality of computer clusters, and the flow field set processing unit 1230 is distributed according to the processing command in the plurality of computer clusters. A commander that integrates the processed results may further be included. Such distributed processing will be described in more detail with reference to FIG.

  FIG. 13 is an example for explaining the distributed data processing method according to the present invention.

  The increase in traffic (traffic) that increases the amount of usage per unit time may cause an overload to the log collection server when a very large number of user logs are input within the unit time, There is a problem in that there is a possibility of causing the loss of logs, and an increase in data that increases the amount of data that must be recorded and analyzed leads to an increase in time for user behavior analysis.

  Therefore, in order to process a user's action in time, a structure like the example of FIG. 13 can be used.

  FIG. 13 realizes distributed processing of user logs and data by increasing the number of log collection servers (LCS) and computer units (computing units; computer.units), that is, computer clusters, to solve the problems caused by traffic increase and data increase. And it can be solved. For example, if the user log increases about 10 times only at a specific time such as midnight or a specific event period, but there is no special difference (change) in the amount of user log for the entire day, The number can be left as it is, and only the log collection server number can be increased to solve the problem of traffic increase.

  That is, the log collection server includes a plurality of collection servers, and user logs are distributed and collected on the plurality of collection servers. In addition, since the fluid field set includes at least one fluid field generated through the user log and classified by user, the fluid field for the same user is designated among a plurality of computer clusters that process the fluid field set. Such a problem can be solved by processing according to the processing instruction in the same computer cluster.

  More specifically, the user behavior analysis system records user logs in a plurality of collection servers, each of which operates in the same manner as the log collection server, and generates a uniform distribution such as a hash function. By specifying a computer cluster for each user through a function, user behavior for a user can be processed through one computer cluster. As described above, since all data for one user is processed by one computer cluster, communication with an external host is unnecessary, and a flow field set is processed by a plurality of computer clusters. Time is reduced. Moreover, the traffic problem can be solved by using a plurality of collection servers.

In the operation of simply counting the number of times and the number of users, a number is counted for each computer cluster, and the numbers are summed up for each corresponding group, so that the total number is calculated, this is the operation for counting the number (count) And the summation operation (sum) is attributed to the property of keeping the exchange law and the coupling law. That is, in the case of obtaining the unique visitor number by the existing method, it was necessary to maintain a set consisting of all users as a member in order to confirm whether the same user was counted. There is an effect that it is not necessary to maintain the corresponding set by storing in each computer cluster for each user.

  Further, all the operations by the above-described draw operator, filter operator, and set operator are executed in the computer cluster. At this time, in order to merge the results of the computer clusters, the set operator can be used once again by using the commander shown in FIG.

  As described above, by using the user behavior analysis system according to the present invention, it is possible to easily and simply perform time-series analysis on user behavior by arranging user behaviors by user or in time order via a user log. In addition, by analyzing the user's behavior in time series, it becomes possible to analyze the cause of the usage status of the website, which has been difficult to grasp, such as the purchase route and the member behavior pattern for a specific case.

  In addition, by analyzing the cause of the usage status of the website, it is possible to understand the behavior of the user, thereby improving the website service, and using a plurality of collection servers that the log analysis server can include Thus, the problem of traffic increase can be solved, and analysis time for user behavior can be shortened by distributed processing using a plurality of computer clusters.

  Note that the user behavior analysis method according to the embodiment of the present invention can also be provided as a program executable on a computer. Specifically, a program for executing each step of the PPP method is stored in a storage unit (not shown) of the computer, and each step of the PPP method is realized by executing each program by a control unit such as a CPU. Is done. The program is composed of one program including all the steps or a plurality of programs for each step, and the computer that executes the program unevenness can also be composed of one or a plurality of computers.

  The present invention can also be provided as a computer-readable recording medium storing the above-described program that causes a computer to implement the user behavior analysis method. The recording medium may include program instructions, data files, data structures, etc. alone or in combination, and the recording medium and program instructions may be specially designed and configured for the purposes of the present invention, It may be known and usable by those skilled in the computer software art. Examples of computer-readable recording media include magnetic media such as hard disks, floppy (registered trademark) disks and magnetic tapes, optical recording media such as CD-ROMs and DVDs, and magnetic-lights such as floppy disks. A medium and a hardware device specially configured to store and execute program instructions such as ROM, RAM, flash memory, and the like are included. The recording medium is also a transmission medium such as an optical or metal line or a waveguide including a carrier wave that transmits a signal for storing program instructions, data structures, and the like. Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that are executed by a computer using an interpreter or the like. The hardware elements described above can be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

  As described above, the present invention has been described according to the preferred embodiments. However, for those skilled in the art to which the present invention belongs, the technical idea of the present invention and its areas described in the claims are described. It will be understood that various modifications and changes can be made to the present invention without departing from the scope of the invention. In other words, the technical scope of the present invention is defined based on the claims, and is not limited by the best mode for carrying out the invention.

It is an example for demonstrating the method to measure a user action pattern in a prior art. It is an example for demonstrating generally the user behavior analysis method which concerns on this invention. It is another example for demonstrating generally the user behavior analysis method which concerns on this invention. 5 is a flowchart illustrating a user behavior analysis method according to an exemplary embodiment of the present invention. It is an example for demonstrating the user log collection method which concerns on this invention. 4 is a flowchart illustrating a method for generating a flow field set in an embodiment of the present invention. It is an example for demonstrating the fluid field set and fluid field which concern on this invention. It is an example of a flow field set concerning the present invention. It is an example for demonstrating the division | segmentation and merge between the flow field sets which concern on this invention. 6 is an example of processing a flow field set according to the present invention. It is an example for demonstrating the processing instruction which concerns on this invention. In other embodiment of this invention, it is a block diagram for demonstrating the internal structure of a user action analysis system. It is an example for demonstrating the distributed processing method of the data which concerns on this invention.

Explanation of symbols

1200: User behavior analysis system 1210: Log collection server 1211: User log receiving unit 1212: URL changing unit 1213: STDB
1214: STMS
1220: Flow field set generation unit 1221: Flow field generation unit 1222: Alignment unit 1230: Flow field set processing unit

Claims (16)

A user behavior analysis method comprising: a user log including at least information for identifying the user and an occurrence time of the user behavior, provided from a log collection server that collects log information related to the user behavior via a web page; The steps performed by the computer using the user behavior analysis process are:
Converting the user log into a flow field format to generate a flow field set;
Processing the flow field set when an analysis processing instruction related to the user behavior is input to the flow field set, and
The flow field format is a predetermined format of a flow field composed of a field name and a plurality of attributes. The flow field includes a user identification field for identifying a user and an action for recording an action of the user or the system. Including record fields,
The attribute includes an attribute name that classifies the attribute and an attribute value that defines a user's behavior, and includes the occurrence time of the user behavior included in the user log as the attribute,
The step of converting the user log into a flow field format to generate a flow field set includes:
Converting each user log individually provided from the log collection server with respect to the occurrence of each user action into the fluid field format to generate each fluid field;
Arranging the flow field by user in time order to generate the flow field set;
Processing the flow field set when an analysis processing command related to the user behavior is input to the flow field set;
Extract the flow field corresponding to the analysis processing instruction from the flow field set, measure the total or number of attributes included in the flow field ,
The analysis processing instruction related to the user behavior is:
(1) a draw operator that extracts a flow field in an input time range;
(2) a filter operator that extracts a flow field that conforms to an input pattern and a set operator that measures the total or number of the attributes based on the input attributes;
The step of processing the fluid field set when there is an input of an analysis processing command related to the user behavior for the fluid field set,
(1) Extracting a corresponding flow field based on the draw operator or the filter operator,
(2) A user behavior analysis method characterized by measuring a total or number of attributes corresponding to the set operator in the flow field set and the extracted flow fields . The analysis processing instruction related to the user behavior includes analysis questions,
The step of processing the fluid field set when there is an input of an analysis processing command related to the user behavior for the fluid field set,
2. The corresponding flow field is extracted based on a time range, a pattern, a statistical function, or a set partition criterion included in the analysis question, or a total or number of attributes included in the flow field is measured. The user behavior analysis method described in 1. The analysis processing instruction related to the user behavior includes an analysis script composed of a time range, a pattern, or an attribute processing instruction,
The step of processing the fluid field set when there is an input of an analysis processing command related to the user behavior for the fluid field set,
Based on the time range based on the input analysis script, the pattern, or the attribute processing instruction,
(1) Extract the relevant flow field,
(2) The user behavior analysis according to claim 1 or 2 , wherein the total or the number of attributes corresponding to the processing command among the attributes included in the set of flowing fields and the extracted flowing fields is measured. Method. Receiving the user log via a web page provided by the log collection server on a web server;
The log collection server changing a URL included in the user log to the predetermined name with reference to a database storing a predetermined name corresponding to the URL;
User behavior analysis method according to one of claims 1 to 3, which comprises a. The step of receiving the user log via a web page provided by the log collection server on a web server includes:
5. The user behavior according to claim 4 , wherein the log collection server receives the user log using a user's web browser and a Javascript tag, or receives the user log directly from the client terminal of the user. Analysis method. The user according to claim 4 , wherein the log collection server further includes a service tree management system that manages the URL and the predetermined name corresponding to the URL in association with each other and maintains the database. Behavior analysis method. The user log includes a web server log generated by a web server,
The log collecting server, a user behavior analysis method according to any one of claims 1 6, characterized in that collecting the web server logs generated by the web server. The log collection server is composed of a plurality of collection servers,
The user log is a user behavior analysis method according to any one of claims 1 7, characterized in that the dispersed collected by the plurality collection server. The flow field set includes at least one flow field generated through the user log and classified by user, and the computer receives an analysis processing instruction related to the user behavior for the flow field set. A plurality of computer clusters that perform the steps of processing the flow field set, if any;
The same computer cluster designated among the plurality of computer clusters extracts a flow field corresponding to the user action analysis processing instruction for the same user, or measures the total or number of attributes included in the flow field. 9. The user behavior analysis method according to claim 8 , wherein A computer-readable recording medium in which a program for causing a computer to execute each step of the user behavior analysis method according to any one of claims 1 to 3 is recorded. A user behavior analysis system,
A log collection server that collects a user log that includes log information related to user behavior via a web page and includes at least information for identifying the user and the occurrence time of the user behavior;
A fluid field set generation unit that generates a fluid field set by converting the user log into a fluid field format;
A fluid field set processing unit that processes the fluid field set when an analysis processing command related to the user behavior is input to the fluid field set;
The flow field format is a predetermined format of a flow field composed of a field name and a plurality of attributes, and includes a user identification field for identifying a user and an action record field for recording the action of the user or the system,
The attribute includes an attribute name that classifies the attribute and an attribute value that defines a user's behavior, and includes the occurrence time of the user behavior included in the user log as the attribute,
The flow field set generation unit includes:
A fluid field generation unit that converts each user log provided individually from the log collection server to the occurrence of each user behavior into the fluid field format to generate each fluid field;
A flow field aligning unit that generates the flow field set by aligning the flow fields according to users in time order, and
The flow field set processing unit includes:
Extract the flow field corresponding to the analysis processing instruction from the flow field set, measure the total or number of attributes included in the flow field ,
The analysis processing instruction related to the user behavior is:
(1) a draw operator that extracts a flow field in an input time range;
(2) a filter operator that extracts a flow field that conforms to an input pattern and a set operator that measures the total or number of the attributes according to the input attributes;
The flow field set processing unit includes:
(1) Extract the corresponding flow field by the draw operator or the filter operator,
(2) The user behavior analysis system characterized by measuring the total or number of attributes corresponding to the set operator in the flow field set and the extracted flow fields . The log collection server
A user log receiver for receiving the user log via a web page provided by a web server;
A URL changing unit that changes the URL included in the user log to the predetermined name with reference to a database that stores a predetermined name corresponding to the URL;
The user behavior analysis system according to claim 11 , comprising: The log collection server
The user behavior analysis system according to claim 12 , further comprising a service tree management system that manages the URL and the predetermined name corresponding to the URL in association with each other and maintains the database. The log collection server is composed of a plurality of collection servers,
The user behavior analysis system according to any one of claims 11 to 13 , wherein the user log is distributed and collected by the plurality of collection servers. The flow field set includes at least one flow field generated through the user log and classified by user,
15. The user behavior analysis system according to claim 14 , wherein the fluid field set processing unit includes a plurality of computer clusters that process the fluid field set. The same computer cluster designated among the plurality of computer clusters extracts a flow field corresponding to an analysis processing instruction related to the user behavior for the same user, or measures the total or number of attributes included in the flow field. ,
The flow field set processing unit includes:
A commander for integrating the results of distributed processing in the plurality of computer clusters;
The user behavior analysis system according to claim 15 , further comprising: