JPH11242679A - Method and apparatus for classifying information based on user's interest, and recording medium storing program for classifying information based on user's interest - Google Patents

Abstract

(57) [Summary] [Problem] In addition to searching for information based on each user's interest, even if the user's interest is dynamically changing, information classification is performed by reflecting the interest at that time. Do it automatically. SOLUTION: The number of pieces of information corresponding to a keyword extracted from all pieces of information whose overall amount can be grasped is compared with the number of pieces of information browsed by a user corresponding to the keyword, and the user is interested in all the keywords. Calculate the degree,
Classification categories are created for the top N keywords in descending order of interest. The degree of relevance of keywords is calculated for all two combinations of keywords, and the two keywords having high degrees of relevance are grouped to create an upper layer of the classification category.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a method for classifying information based on a user's interest.

[0002]

2. Description of the Related Art Information classification by a system does not generate classification categories that are easy to understand for individual users, but rather, the system gives classification categories that are common to all users from the beginning, and the classification categories are classified. The approach was to classify the information according to. Since this does not reflect the individual user's interest, the user needs to search for the classified information in order to use the classified information.

On the other hand, there is also an approach in which a user presents his / her belief interest to a system that classifies information in advance, and the system generates a category of classification based on the interest and classifies the information. there were.

[0004]

With this approach,
Since the user needs effort to present his / her own belief at first, and even if the interest of the user changes, the classification category used by the system is fixed, so the interest may change. There were problems such as inability to follow.

SUMMARY OF THE INVENTION It is an object of the present invention to classify information individually based on the interests of each user, and to reflect the interests at the time even if the interests of the users change dynamically. It is an object of the present invention to provide a method and apparatus for classifying information based on a user's interest, and a recording medium on which an information classification program is recorded, which can automatically perform the classified information.

[0006]

SUMMARY OF THE INVENTION According to the present invention, there is provided a method for classifying information based on a user's interest, wherein the information extracted from information browsed by a user in a group of information whose total amount can be grasped. The number of information corresponding to each keyword representing the characteristic of the information is compared with the number of information corresponding to the keyword extracted from the information group and representing the characteristic of the information. A classification category generation step of generating a classification category, which is a container used for information classification, in a one-to-one correspondence with a keyword having a large number, and classifying information, and a degree of association among the classification categories The method includes a higher-level creation step of grouping the classification categories having higher ranks and creating an upper-level hierarchy.

Further, the apparatus for classifying information based on the user's interest according to the present invention is a device for extracting characteristics of the information extracted from information viewed by the user from a certain information group whose total amount can be grasped. The number of information corresponding to the keyword is compared with the number of information corresponding to the keyword extracted from the information group and representing the characteristic of the information, and the ratio of the number of the former information to the number of the latter information is large. One to one for keywords
A classification category generating means for generating a classification category, which is a container used for classifying information, and classifying information; and grouping classification categories having a high degree of association among the classification categories to form a higher hierarchy Is created.

[0008] A recording medium according to the present invention, in which a program for classifying information based on the user's interest is recorded, the information extracted from the information viewed by the user from a certain information group whose overall amount can be grasped. The number of information corresponding to each keyword representing the feature of the information is compared with the number of information corresponding to the keyword representing the feature of the information extracted from the information group, and the former information is compared with the number of the latter information. A category that is a container used for information classification in a one-to-one correspondence with keywords having a large ratio of the number of keywords, and classifies the information, and the degree of relevance of the classification categories is high. In such a case, the classification categories are further grouped, and the information is classified based on the user's interest in order for the computer to execute an upper layer generation process of creating an upper layer. That contains the program.

[0009] A keyword extracted from all information within a certain range is compared with a keyword extracted from information viewed by a user within the certain range, and a keyword corresponding to information viewed by the user at a high rate is determined. Automatically generates a classification category that corresponds to the keyword on a one-to-one basis. At this time, the degree of relevance of a plurality of keywords of the information viewed at a high rate corresponding to the classification category is calculated, and if the degree of relevance is high, the classification categories corresponding to those keywords are grouped. The category of the upper hierarchy is automatically generated and the classification is hierarchized.

[0010] Therefore, when a user browses information from among information whose total amount can be grasped, a classification category according to the interest of each user can be automatically set without the user preparing a classification category in advance. And the information can be classified. Further, when the user browses new information as time elapses, a new classification category can be automatically generated according to a change in the user's interest, or unnecessary classification categories can be deleted.

[0011]

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 shows a schematic configuration of a system according to an embodiment of the present invention. An information server 1 in which information is stored, and a terminal such as a personal computer used when a user browses the information. 2 and a computer network 3 for arbitrarily connecting them.

The method for classifying information based on the user's interest according to the present embodiment includes a pre-process performed by the information server 1 and the user terminal 2 and a main process performed by the information server 1.

First, the preprocessing will be described with reference to FIGS.

FIG. 3A shows the flow of processing on the information server 1 side. First, information (FIG. 4) is stored in the information server 1.
When registering (1)), information ID (FIG. 4) is assigned to each information.
(2)) is attached (step 11). Next, for each piece of information in the information server 1, a keyword group (FIG. 4 (3)) that represents the feature of the information is extracted (step 12).
Also, this keyword group is attached to each piece of information (step 13, FIG. 4 (4)). Finally, for all information in the information server 1, "a certain keyword, the number of information including the keyword, and the information ID" (FIG. 4 (5))
Is stored in the information server 1 in a table format (step 14).

FIG. 3B shows the flow of processing on the terminal 2 side. A user browses information in the information server 1 via the computer network 3 using the terminal 2. At this time, information indicating which information has been browsed is stored in the terminal 2 (step 21). At this time, a keyword group attached to each piece of information viewed is also stored (FIG. 4 (6)). Next, "a certain keyword, the number of information including the keyword, and the information ID" are displayed for all the information browsed by the user in the information server 1 (FIG. 4 (7)).
Is stored in the terminal 2 in a table format (step 22).

Next, this processing will be described with reference to FIGS. 2 and 3 (3). In this processing, (I) one-to-one for the keyword
, And (II) generation of an upper hierarchy by grouping the categories, respectively, are performed by the classification category generation unit 5 and the upper hierarchy creation unit 6, respectively.

(I) Generation of Classification Categories Corresponding to Keywords One-to-one is based on the degree of interest indicating how much a user is interested in a certain keyword. The calculation of the degree of interest includes “a certain keyword and the number of pieces of information including the keyword and the information ID” for all pieces of information in the information server 1 described above, and “the information browsed by the user in the information server 1”. "A keyword and the number of information including the keyword and its information ID" for all the information. "The number of information including a certain keyword for all information viewed by the user in the information server 1" is divided by the "number of information including the keyword for all information in the information server 1". Let the result be the degree of interest E (0 ≦ E ≦ 1). The degree of interest E is calculated for all keywords (step 31),
For the top N keywords (N is a variable and an integer) in order from the keyword having the largest value, the category of the classification corresponding to the keyword in one-to-one correspondence is generated (step 32).

(II) Generation of an upper hierarchy by performing grouping using the categories generated in (I) Based on the classification categories generated by the previous operation, classification of the classification categories is performed. Do. The criterion for classifying the categories into categories is a degree of association that measures how much the user considers the keywords A and B to be related.
For the calculation of the degree of relevance, the information regarding the keywords A and B of the above-mentioned "information ID of a certain keyword and information including the keyword for all information browsed by the user in the information server" is used. . A set of "information including keyword A" for all information viewed by the user in the information server and a set of "information including keyword B" for all information viewed by the user in the information server Is greater, the degree of association is weighted higher. The overlap (number) of the set of “information including the keyword” for all the information browsed by the user in the information server of the two keywords A and B is set to “all the information browsed by the user in the information server”. "Number of information containing keyword A" for the information of
The number of pieces of information including the keyword B "is determined as the degree of association (step 33). Two keywords for which at least one of the degrees of relevance is higher than the threshold value X are grouped to generate an upper layer of the classification category (step 34). For three or more keywords, for example, keyword A, keyword B, keyword C
About keyword A and keyword B, keyword B
If the degree of association between the keyword A and the keyword A is higher than the threshold X, the categories of the classifications generated from these three keywords are grouped to generate an upper layer.

In accordance with an increase in the information browsed by the user, the generation of the classification categories and the hierarchical classification of the classification categories are periodically and automatically performed. At this time, even if the category is a category that has been generated before, the degree of interest is relatively low, and, for example, the category of the category corresponding to the keyword that is no longer included in the top 20 categories is deleted. In addition, even if the category is a category that has not been generated before, the degree of interest is relatively high, and a category that corresponds to the top 20 keywords is newly generated. The deletion and generation of categories are performed in the same manner.

By using this method, when the categories and hierarchies of the generated classifications are presented to the user, they can be presented separately for each category of the generated classifications. At this time, the information classified into the category is divided into information that the user has already browsed and information that has not been browsed, and the information that has not been browsed is provided in a recommended form.

Further, since the user's interest structure can be grasped by using this method, a community can be created by drawing together users having similar interest structures.

The processing section shown in FIG. 2 may be provided in the terminal 2. The above-described processes of generating and erasing the classification categories and hierarchizing and erasing the classification categories are recorded as information classification programs on a recording medium such as an FD, a CD-ROM, or a semiconductor memory, and are read and executed by a computer. You can also.

[0024]

As described above, according to the present invention,
Assuming that a plurality of users are browsing information from the same information server, it is possible to perform information classification individually suitable for each user based on their interests. Further, even if the user's interest is dynamically changing, it is possible to automatically perform information classification reflecting the interest at that time.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a system to which the present invention is applied.

FIG. 2 is a configuration diagram of a part that performs this processing.

FIG. 3 is a flowchart illustrating a method of classifying information based on a user's interest according to the present invention.

FIG. 4 is an explanatory diagram of information generated according to the method of FIG. 3;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Information server 2 Terminal 3 Network 5 Classification category generation part 6 Upper layer creation part 11-14, 21,22,31-34 Step

Claims (15)

[Claims] 1. A method of classifying information based on a user's interest, wherein each of information representing a feature of the information extracted from information viewed by the user from a certain information group whose total amount can be grasped. The number of information corresponding to the keyword is compared with the number of information corresponding to the keyword extracted from the information group and representing the characteristic of the information, and the ratio of the number of the former information to the number of the latter information is large. A classification category generation step of generating a classification category, which is a container used for information classification, for one-to-one correspondence with keywords and classifying information, and a classification category having a high degree of association among those classification categories. A method of classifying information based on a user's interest, comprising an upper layer creation step of grouping and creating an upper layer. 2. The method according to claim 1, wherein the classification category generation step includes the step of determining the number of pieces of information corresponding to each keyword representing a feature of the information extracted from the information browsed by the user in a certain information group whose total amount can be grasped. Dividing by the number of pieces of information corresponding to the keyword extracted from the information group and representing characteristics of the information, and generating a classification category corresponding to one-to-one with respect to the top N keywords from the largest value. Item 1
The described method. 3. The method according to claim 1, wherein the upper hierarchy generation step includes, for all two combinations of keywords corresponding to the generated classification categories, the number of pieces of information viewed by the user including both keywords. The method according to claim 1, wherein the category is divided by the number of information viewed by the user, and the classification categories corresponding to keywords whose at least one of them exceeds a preset threshold X are grouped. 4. The classification category generating step, as a result of performing the division again, deletes a classification category corresponding to a keyword that is not included in the top N keywords among keywords that are currently generating a classification category. 3. The method of claim 2. 5. The upper layer generation step, wherein, as a result of performing the division again, the upper layer corresponding to the keyword whose value is equal to or smaller than the threshold value X among the keywords that are currently generating the classification category of the upper layer. 4. The method according to claim 3, wherein the classification categories are deleted and treated as individual classification categories. 6. An apparatus for classifying information based on an interest of a user, wherein each apparatus is extracted from information viewed by the user from a certain group of information whose total amount can be grasped, and represents characteristics of the information. The number of information corresponding to the keyword is compared with the number of information corresponding to the keyword extracted from the information group and representing the characteristic of the information, and the ratio of the number of the former information to the number of the latter information is large. Classification category generating means for generating a classification category, which is a container used for information classification, in one-to-one correspondence with keywords, and classifying information, and a classification category having a high degree of association among the classification categories. An apparatus for classifying information based on a user's interest, comprising an upper layer creating means for grouping and creating an upper layer. 7. The classification category generation unit determines the number of pieces of information corresponding to each keyword representing a feature of the information extracted from information viewed by a user in a certain information group whose total amount can be grasped. Dividing by the number of pieces of information corresponding to the keyword extracted from the information group and representing characteristics of the information, and generating a classification category corresponding to one-to-one with respect to the top N keywords from the largest value. Item 6
The described device. 8. The method according to claim 1, wherein the upper hierarchy generation means uses, for each of the two combinations of keywords corresponding to the generated classification category, the number of pieces of information browsed by the user including both keywords. The apparatus according to claim 6, wherein the category is divided by the number of pieces of information viewed by the user, and the classification categories corresponding to the keywords whose at least one of them exceeds a preset threshold X are grouped. 9. The classification category generating means, as a result of performing the division again, erases a classification category corresponding to a keyword that is not included in the top N keywords among keywords that are currently generating a classification category. An apparatus according to claim 7. 10. The upper hierarchy generating means, as a result of performing the division again, as a result of the upper hierarchy corresponding to the keyword whose value is equal to or less than the threshold X among the keywords for which the classification category of the upper hierarchy is currently generated. 9. The apparatus according to claim 8, wherein the classification categories are deleted and treated as individual classification categories. 11. A program for classifying information based on a user's interest, wherein each program is extracted from information viewed by a user from a group of information whose total amount can be grasped, and represents a characteristic of the information. The number of information corresponding to the keyword is compared with the number of information corresponding to the keyword extracted from the information group and representing the characteristic of the information, and the ratio of the number of the former information to the number of the latter information is large. 1 for keywords
Classification category generation processing for generating a classification category, which is a container used for information classification in correspondence with one-to-one, to classify information, and grouping the classification categories having a high degree of association among the classification categories, and For causing a computer to execute a higher hierarchy generation process for creating a hierarchy,
A recording medium storing a program for classifying information based on a user's interest. 12. The classification category generation processing includes the step of determining the number of pieces of information corresponding to each keyword representing a characteristic of the information extracted from information viewed by a user from a certain information group whose total amount can be grasped. Dividing by the number of pieces of information corresponding to the keyword extracted from the information group and representing characteristics of the information, and generating a classification category corresponding to one-to-one with respect to the top N keywords from the largest value. Item 12. The recording medium according to Item 11. 13. The method according to claim 13, wherein the upper layer generation processing includes, for each of the two combinations of keywords corresponding to the generated classification category, the number of pieces of information browsed by the user including both keywords. 12. The recording medium according to claim 11, wherein the category is divided by the number of pieces of information browsed, and the classification categories corresponding to the keywords in which at least one of the values exceeds a preset threshold X are grouped. 14. The classification category generating step, as a result of performing the division again, deletes a classification category corresponding to a keyword that is not included in the top N keywords among keywords that are currently generating a classification category, The recording medium according to claim 12. 15. The upper layer generation processing includes, as a result of performing the division again, an upper layer corresponding to a keyword whose value is equal to or smaller than the threshold value X among keywords that are currently generating a classification category of an upper layer. 14. The recording medium according to claim 13, wherein said category is deleted and treated as an individual category.