JP2000242626A - Method for analyzing electronic business history - Google Patents

Abstract

PROBLEM TO BE SOLVED: To analyze the behavior of a user by extracting an URL from history information on access to the electronic business site, analyzing the action of the user according to the action level of the user in electronic business, and reasoning the amount of business by analogy. SOLUTION: A history input processing 112 takes only history information of some session of some user out of user history information and passes it to an action level determining processing 113. The action level determining process 113 classifies log information passed from the history input processing according to a level determination rule 115. Then the classified action level of the user and the corresponding log information and outputted from a level output processing 114 to a totalizing processing 119. The totalizing processing 119 performs totalization as to the user having the level information and on the basis of the result, a behavior analyzing process 120 analyzes the behavior of the user. Further, a processing 121 which reasons the amount of business by analogy from an access frequency reasons the amount of electronic business by analogy on the basis of the totalization result 122.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for estimating the amount of e-commerce by classifying the behavior of a user in e-commerce using the Internet according to the importance of the e-commerce.

[0002]

2. Description of the Related Art As a method of analyzing the behavior of Internet users, a survey method using a questionnaire is generally known. However, due to poor work efficiency in the questionnaire, it is expected that the access history of the user will be mechanically acquired in some form.

As a known example of a method for collecting information on the Internet, there is Japanese Patent Application Laid-Open No. H10-124428 entitled "Information Collection Method, Storage Medium, and Information Collection System on the Internet". According to the present invention, detailed information is automatically collected based on the statistical processing result of the activity status of the information viewer and supplied to the information provider, which helps the information provider to improve the browsing information such as WWW and electronic mail. It aims to provide information gathering technology on the Internet where possible.
As an implementation method, when an information viewer browses the information provider's WWW in a network in which an information provider's server, an information viewer's client, and an information collector's server are connected on the Internet, The acquisition client program is installed to acquire information such as the URL and time of the WWW browsed by the information viewer in the server of the information collector, and the acquired browse information is classified by access time, access frequency, gender, Statistical processing is performed for each age group and region, and based on the result of the statistical processing, detailed information on the Internet is automatically obtained by an information collector's server.

Further, as an invention for analyzing the behavior of a user, there is JP-A-5-28124, "Behavior Analysis Device". An object of the present invention is to provide a behavior analysis simulation method in a field where it is difficult to obtain general knowledge on the behavior of a person or a system. For example, an emergency evacuation model is used.

[0005]

In the first known example, the analysis of the behavior of the user unique to the electronic commerce is not examined.

Further, in the second known example, behavior involving physical movement is studied, and behavior and behavior characteristics of a user on the Internet are not studied.

Generally, in the electronic commerce using the Internet, there are the following problems in analyzing the behavior of a user.

(1) It has not been possible to determine whether the frequency of access to Web pages on the Internet relates to actual commercial transactions. For example, a user frequently accesses a homepage, but this page is not directly related to the purchase of a product, and thus has a low importance in a commercial transaction. Therefore, if the count is performed only based on the access frequency, the statistical processing is performed even on an act that is not related to the commercial transaction, and an incorrect result is obtained. That is, in electronic commerce, it is impossible to simply classify and analyze the behavior of the user based on the frequency of access to the Web page.

(2) An Internet e-commerce site on the Internet is constructed securely, and it is almost impossible to acquire encrypted communication contents and Web page contents of the site.

In the present invention, only the access history information for the e-commerce site is extracted from the access history information on the Internet, the URL included in the history information is extracted, and a keyword is classified using a predefined table, and each keyword is classified. The purpose is to determine the corresponding user's action level (determined based on the degree of importance related to the actual transaction) and analyze the user's behavior based on this classification result and the user's profile And

It is another object of the present invention to provide a method for estimating an electronic commerce amount based on a statistical analysis result of a user.
It is another object of the present invention to detect https issued at the time of SSL communication and to analyze the behavior at an encrypted commerce site.

[0012]

Means for Solving the Problems The above problems (1) and (2)
In order to solve the above problem, the present invention provides (1) extracting only history information related to electronic commerce from Internet access history information, extracting a URL from the history information, and classifying the activity level using the URL as a keyword. Suggest a method. The action level is provided with four levels based on how much the action performed by the user on the e-commerce site is related to the commercial transaction. According to this method, it is possible to classify the history according to the importance of the commercial transaction, not the access frequency.

(2) For security in electronic commerce, S
SL (Secure Socket Layer: a protocol in which an encryption and authentication function is added to TCP / IP) is frequently used. Utilizing this, SSL information is used in history information.
It is detected that the communication has been issued, and at this time, the user is considered to have completed a certain commercial transaction and is classified into the fourth level.

As described above, using the methods (1) and (2), the URL included in the history information of accessing the electronic commerce site
And a method of analyzing the user's behavior based on the user's behavior level in the electronic commerce and further estimating the transaction amount.

Further, by detecting that https is issued at the time of SSL communication, and classifying this as the behavior level 4, the user's behavior can be controlled with respect to access to an encrypted secure e-commerce site. Is provided.

[0016]

(Embodiment 1) An overall configuration of an electronic commerce history analysis method according to the present invention will be described with reference to FIG. FIG. 1 shows an embodiment of the electronic commerce history analysis method.

The e-commerce history analysis method shown in FIG. 1 includes user history information (110), history input processing (112), action level determination processing (113), and level output processing (11).
4), level determination rule (115), level / keyword correspondence table (116), rule generation processing (11)
7), level information (118), tallying process for each user (119), user behavior analysis process (120), process of estimating transaction volume from the number of accesses (121), tallying result (1)
22) is included.

The user behavior analysis process and the process of estimating the transaction amount from the number of accesses provide services to the user through the Internet. These generally perform the following functions.

The user history information (111) contains the user I
D, log information in which the time when the user accesses the Internet, the access URL, the access status, and the like are automatically collected.

In the history input process (112), only the history information of a certain session of a certain user is extracted from the user history information and passed to the next process.

The action level determination process (113) classifies the log information passed from the history input process based on a level determination rule (115). The level determination rule (115) is a level-keyword correspondence table (11
It is created by a rule creation process (117) based on 6).

In the level output processing (114), log information corresponding to the categorized user's action level is output.
The information (log and user action level) thus obtained is level information (118).

Next, by a tallying process (119) for each user, tallying is performed for users having level information. Based on this result, the behavior of the user is sequentially analyzed in the user behavior analysis process (120). Analysis results are provided via the Internet. Further, based on the tallying result (122) obtained from the tallying process for each user, the e-commerce transaction amount is estimated by the process (121) of estimating the transaction amount from the number of accesses. The analogy is served through the Internet.

FIG. 2 shows a processing flow of the history input processing (112).

The history input processing used in this embodiment is realized by the following processing. First, only the accessed URL is extracted from the user history information (111) (201).
Next, the domain name of this URL is extracted (202). List of domain name and e-commerce site (2
03) is compared (204), and only those that match are read from the database as logs relating to electronic commerce (205).

FIG. 3 shows a processing flow of the action level determination processing (113) and the level output processing (114). The action level determination processing is realized by the following processing. First, FIG.
The log related to the electronic commerce obtained in the history input processing of (1) is read (301). At this point, since the input log information is all related to electronic commerce, the action level 2 is assigned as a default value (302).

The action level is obtained by classifying the user's action in electronic commerce according to its importance, that is, whether or not the user has performed a commercial transaction. Level 2 is when an action is performed (searching, browsing, etc.) on the site, and level 3 is when an action more than selecting a product and putting it into a basket or cart is performed.

URLs are used to classify each. A keyword is determined from English words that appear in the URL. For example, if the level is 1, the URL of the home page of the site is used as a keyword. Level 2 applies most of the time, so UR including e-commerce sites
If it is L, the default value is 2 for all.

Level 3 is a URL relating to an order for a product. These URLs include Order, shopping-basket, shop
Characteristic words such as ping-cart and credit-card are included.
Therefore, such a word is registered as a keyword of level 3. The level / keyword correspondence table of FIG. 4 is obtained as described above.

Further, following the level 3 state,
The case where L (Secure Socket Layer) occurs is regarded as level 4, and it is considered that the commercial transaction has been completed. Specifically, at this time, the URL is http instead of http
, It can be determined that it is SSL. Therefore, in 303, it is determined whether or not the input log information corresponds to level 1, level 3, or level 4. At the time of discrimination, classification is performed using the level / keyword correspondence table (116).

Words serving as keywords are registered in each of level 1 and level 3, and if the keyword is included in the URL of the log information, it is determined that the level is the corresponding level. If it corresponds to each level, the default value is rewritten (304), a pointer to log information corresponding to the determined level is output (305), and the process ends.

FIG. 4 shows an example of the level / keyword correspondence table (116). A domain name, a keyword, and an action level of the user corresponding to the keyword are described.
Each of these can have multiple values. For example, in the case of hitachi.com, the keyword of the level 1, that is, the URL of the homepage is http://www.hitachi.com/
It is. There are two level 3 keywords, shopping-b
asket and order. If these words are included in the URL related to Hitachi.com, the URL is determined to be level 3.

FIG. 5 shows an example of the level information (118). In this example, the format is client, date, "http_
request ", return_code, data_size, and action_level. The Client is an identifier for identifying the accessed machine. From this, the user who is accessing is identified using the pre-registered user profile information (608). Date is the date of access, http_request is the content of the issued http access request and includes the URL, Return_code is the value returned from the system as a result of the access (usually 2
00). Data_size is the size of the accessed information. Action_level is the action level described above.

An example of the first line (one session) will be described. Client is localhost. (This is an example, so user profile information is not specified.) Date and time of next access (May 13), http method is GET
The URL of the access destination is http: //www.h
Itachi.com: 80 / ~ alan / bana / miura.jpg Since this URL is for hitachi.com, the action_level, that is, the user action level is 2. In addition, those with no user action level are blank.

FIG. 6 shows a tallying process for each user (11
9) shows a process flow of a user behavior analysis process (120) and a process (121) of estimating a transaction amount from the number of accesses.
The tallying process (119) for each user is realized by the following process. Focusing on the user ID among the access history information of the users included in the level information, history information of the same user ID is extracted in chronological order (601). The user profile is the user profile information (60
8), when the tabulation is performed, it is possible to classify, for example, what percentage of a woman is and what percentage of a single person is.

At 601, when there is an access history classified as level 3, if an SSL access has occurred continuously, it is regarded as level 4 and the level is rewritten (602). FIG. 7 shows an example of the tallying result (122) for each user.

The user behavior analysis processing (120) is realized by the following processing. The behavior of the user is sequentially analyzed based on the level information collected for each user (603). The relationship between the number of people who visited a certain e-commerce site (level 1), the number of people who were interested (level 2), and the number of people who actually tried to purchase (level 3 + level 4), etc., is presented (604). In addition, the behavior of the user is sequentially analyzed and information on the efficiency of site construction is presented (60).
Examples such as 5) are possible.

Processing for estimating the transaction amount from the number of accesses (1)
21) is realized by the following processing. The behavior of the user is sequentially analyzed based on the level information collected for each user (603). When the user makes a level 3 or level 4 access, it counts as if some kind of electronic commerce was performed (606). This count and 1
A product obtained by multiplying the average purchase price per access by the user is presented as a predicted purchase price (607).

FIG. 7 is a totaling result (122) for each user.
Here is an example. Taking a certain user as an example, this user ID is 238592M, and the corresponding user profile No.
Is 238592. The access date is May 13, 1998. Hereinafter, the access time, the accessed URL, and the action level of the user are described in order for each row.
For example, in the example of the first line, the time is 13:50:12,
The URL is http://www.hitachi.com/ .

Since this URL is the home page of hitachi.com, the user action level was 1. In the example of the sixth line, the time is 14:06:11, and the URL is
http://www.hitachi.com/credit-card . At this time, since the URL starts with https, it is known that SSL communication has been performed.

In the access information on the fifth line immediately before, this user is at the activity level 3, so the SSL communication at this time corresponds to the level 4. Therefore, the user action level is 4. In the next seventh line, the user is accessing another site, which indicates that the user has exited hitachi.com. Therefore, the action level is 0.

FIG. 8 shows an embodiment of a user interface for presenting the result obtained from the user behavior analysis process and the result obtained from the process of estimating the transaction amount from the number of accesses. This is an example of a screen provided on the Internet.

Here, the totaling result for each week is displayed. The date is displayed at the top. Next, it is specified that a list of the top 100 ECs (electronic commerce) is displayed, and that the product category is BOOK. Below that is the access rank list of the e-commerce site. Here, the first place is hitachi.
com, and the Visitor displays the number of visitors to the site, that is, the number of level 1 users (170512). The number of users of level 3 (537) is displayed in Order on the right side. The right end is the estimated trading volume ($ 11,050). Hereinafter, it is displayed in order from the second place.

A graph of the above display result is also shown. The color changed in the bar graph is the bar graph selected by the user by moving the cursor. In this case, seki.com has been selected, and its user profile indicates that 48% of the Orders are women and 27% are singles.

FIG. 9 is a processing flow of the rule generation processing (117). Level / keyword correspondence table (11
6) is created beforehand by using the rule generation process (117). Since the keyword of level 1 is easily obtained from the URL of the homepage, here the level 3
A method for obtaining the keyword of the above will be described.

First, a URL at the time of performing electronic commerce is given as an example (901). The URL is divided by the symbol /, and each divided element is set as a keyword candidate (90
2). It is confirmed whether or not the last element that appears in the keyword candidates is a keyword of level 3 (903). YE
If it is S, synonyms and inflected forms of level 3 keywords are also presented as keyword candidates (904).

At this point, a URL including a keyword of level 3 is extracted from the history information (90).
5). The process returns to 902 again. Also, in the case of NO in the process 903, it is confirmed whether or not each element appearing in order from the rear is used as a keyword of level 3 (906). YE
If it is S, the process enters 904. If all are NO, the process ends.

[0048]

As described above, according to the present invention, the URL included in the history information of accessing the electronic commerce site is extracted, and the behavior of the user is analyzed based on the activity level of the user in the electronic commerce. Another object of the present invention is to provide a method for estimating transaction volume.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an electronic commerce history analysis method according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a history input process according to the present invention.

FIG. 3 is a flowchart showing an action level determination process and a level output process according to the present invention.

FIG. 4 is a diagram showing a level / keyword correspondence table according to one embodiment of the present invention;

FIG. 5 is a diagram showing one embodiment of level information according to the present invention.

FIG. 6 is a flowchart showing a tallying process for each user, a user behavior analysis process, and a process of estimating a transaction amount from the number of accesses according to the present invention.

FIG. 7 is a diagram showing an example of a tally result for each user.

FIG. 8 is a diagram showing an embodiment of a user interface for presenting a result obtained from a user behavior analysis process and a result obtained from a process of estimating a transaction amount from the number of accesses.

FIG. 9 is a flowchart illustrating a rule generation process.

[Explanation of symbols]

111: user history information, 112: history input processing, 11
3 ... action level determination processing, 114 ... level output processing, 1
15: Level determination rule, 116: Level / keyword correspondence table, 117: Rule generation processing, 118: Level information, 119: Total processing for each user, 120: User behavior analysis processing, 121: Transaction volume from the number of accesses Analogous processing.

Continued on the front page (72) Inventor Hiroshi Tsuji 1099 Ozenji Temple, Aso-ku, Kawasaki City, Kanagawa Prefecture F-term in Hitachi, Ltd. System Development Laboratory 5B049 AA02 BB11 CC08 EE02 GG07 GG09

Claims (3)

[Claims] 1. A method of analyzing history information of accessing the Internet, comprising: extracting history information on electronic commerce from the access history; and determining a URL included in the history information on electronic commerce.
A step of classifying and determining the activity level according to the importance of the commercial transaction (whether or not the product is going to be purchased) based on the keyword, and the activity level determined according to the importance of the commercial transaction are separately accumulated. An electronic commerce history analysis method, comprising a step of statistically analyzing a behavior of a user from user profile data. 2. The electronic commerce history analysis method according to claim 1, further comprising a step of estimating and presenting a transaction amount of the commerce site based on an activity level according to the importance of the transaction. 2. The electronic commerce history analysis method according to claim 1, wherein: 3. An SSL (Secu) used in electronic commerce.
re Socket Layer: a protocol in which an encryption and authentication function is added to TCP / IP), and a step of detecting that SSL communication has been issued in the history information. At this time, the user has performed a commercial transaction. 2. The electronic commerce history analysis method according to claim 1, wherein the histories are classified as being considered.