JP4535765B2 - Content navigation program, content navigation method, and content navigation apparatus - Google Patents

Description

  The present invention relates to a content navigation program, a content navigation method, and a content navigation device for supporting information search, and in particular, a content navigation program, a content navigation method, and a content navigation device that can present a keyword related to a keyword selected by a user. About.

  With the development of high-bandwidth network environments such as FTTH (Fiber To The Home) and ADSL (Asymmetric Digital Subscriber Line), the Internet has become an indispensable infrastructure for people's lives. Under such circumstances, a technique for finding information from a large amount of contents on the Internet is essential, and there are many service providers that provide such functions to users.

  Information search functions are provided on many portal sites. For example, there is a service (directory search) that supports information search by a user by classifying information into a hierarchical structure, and a service (keyword search) that lists information according to an input keyword. These services are used by many users.

  The amount of information on the Internet is enormous. Therefore, in the keyword search, the number of pieces of information that match the input keyword may be large. In view of this, a technique has been proposed in which each information is ranked based on the citation relationship between the information, and the information is preferentially listed from the information of higher rank (for example, see Patent Document 1).

Also, when performing a keyword search, it is important to input an appropriate keyword. Therefore, if a synonym dictionary is created in advance with a plurality of keywords having the same meaning, the keyword input by the user can be complemented with the synonym and a search without omission can be performed. Therefore, a technique of automatically creating a synonym dictionary that defines the degree of association between words has also been considered (see, for example, Patent Document 2).
US Pat. No. 6,526,440 JP 11-31168 A

  Currently, as the value of the Internet as an information transmission medium that follows TV and newspapers is increasing, there are not a few examples in which topics and trends are formed via personal Web pages, BLOGs (Web logs), and bulletin boards. It is important for operators of websites that conduct electronic commerce to quickly capture the attention and interest of users on the Internet and apply them to marketing.

  However, there is a problem that there is no efficient method for reflecting the contents of new contents that are released every day in search results. For example, since directory search manually classifies information, directory maintenance for enormous contents cannot keep up.

  Also, in keyword search, an optimum result cannot be obtained unless an appropriate search keyword is input for the information to be sought. For example, when the latest technology is developed, a new name is given to the technology. When searching for contents related to the latest technology from the Internet, if the user does not input the name assigned to the latest technology as a keyword, the target information may be buried in a huge amount of irrelevant information.

  In the above-mentioned patent document 1, each information is ranked according to the citation relationship between the information so that useful information is preferentially presented. However, it is expected that content that has just been released will have few citations from other content, even if the content is important. In this case, even if the content matches the keyword search, the priority is lowered and the possibility that the user overlooks is increased.

Moreover, in the said patent document 2, since the relevance degree of words is defined using only the word which the user input, the information obtained as a search result is not utilized effectively.
The present invention has been made in view of these points, and provides a content navigation program, a content navigation method, and a content navigation apparatus that can easily search for information that is currently of interest to many users. For the purpose.

  In order to solve the above-described problems, the present invention provides a content navigation program for supporting content search as shown in FIG. When the computer executes the content navigation program according to the present invention, the computer realizes the following functions.

  Each time the user performs a search based on the keyword 6a and selects an arbitrary content 7b from the search result 6b, the storage unit 1a stores the search keyword 6a and the identification information of the selected content 7b in association with each other. Store in means 1b. The grouping unit 1c determines the relevance between the keywords stored in the storage unit 1b based on the correspondence between the keyword 6a and the selected content 7b, and groups a plurality of related keywords. The related keyword output unit 1d outputs another keyword belonging to the same group as the selected representative keyword when an arbitrary representative keyword 8a is selected.

  According to the computer that executes such a content navigation program, whenever the user performs a search based on the keyword 6a and selects an arbitrary content 7b from the search result 6b, the storage means 1a causes the search keyword 6a. And the selected content 7b are stored in the storage unit 1b in association with each other. Thereafter, the grouping unit 1c determines the relevance between the keywords stored in the storage unit 1b based on the correspondence between the keyword and the selected content, and groups a plurality of related keywords. When an arbitrary representative keyword 8a is selected, the related keyword output unit 1d outputs another keyword belonging to the same group as the selected representative keyword.

  In addition, in order to solve the above-mentioned problem, in a content navigation method for assisting a search for content by a computer, a storage unit performs a search based on a keyword and selects arbitrary content from search results. The search keyword and the identification information of the selected content are associated with each other and stored in the storage means, and the grouping means is configured to store the storage means based on the correspondence between the keyword and the selected content. A plurality of related keywords are grouped, and when the related keyword output means selects an arbitrary representative keyword, the same group as the selected representative keyword The other content keyword belonging to is output. Geshon method is provided.

  According to such a content navigation method, every time a user performs a search based on a keyword and selects any content from the search results, the storage keyword associates the search keyword with the selected content. Stored in the storage means. Thereafter, the grouping means determines the relevance between the keywords stored in the storage means based on the correspondence between the keyword and the selected content, and a plurality of related keywords are grouped. When an arbitrary representative keyword is selected, the related keyword output unit outputs another keyword belonging to the same group as the selected representative keyword.

  Further, in order to solve the above problem, in the content navigation device for supporting content search, whenever a user performs a search based on a keyword and selects any content from the search results, the search Based on the correspondence between the keyword and the selected content, the storage means for associating the keyword with the identification information of the selected content and storing it in the storage means, between the keywords stored in the storage means Grouping means for determining relevance and grouping a plurality of related keywords, and when an arbitrary representative keyword is selected, another keyword belonging to the same group as the selected representative keyword is output And a related keyword output means. There is provided.

  According to such a content navigation apparatus, every time a user performs a search based on a keyword and selects any content from the search results, the storage keyword associates the search keyword with the selected content. Stored in the storage means. Thereafter, the grouping means determines the relevance between the keywords stored in the storage means based on the correspondence between the keyword and the selected content, and a plurality of related keywords are grouped. When an arbitrary representative keyword is selected, the related keyword output unit outputs another keyword belonging to the same group as the selected representative keyword.

  As described above, in the present invention, keywords are grouped based on the relationship between the keywords and the content selected by the user from the search result based on the keywords. As a result, keywords related to content that many users are currently interested in are grouped together. Then, by outputting other keywords in the same group as the selected representative keyword, the keywords input by other users with similar interest targets can be presented to the user.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
First, the outline of the invention applied to the embodiment will be described, and then the specific contents of the embodiment will be described.

  FIG. 1 is a conceptual diagram of the invention applied to the embodiment. The content navigation apparatus 1 according to the present invention is connected to a client 2, a search server 3, a content server 4, and a client 5 used by a user who uses a content navigation service. The content navigation apparatus 1 includes a storage unit 1a, a storage unit 1b, a grouping unit 1c, and a related keyword output unit 1d.

Each time the user performs a search based on the keyword 6a and selects an arbitrary content 7b from the search result 6b, the storage unit 1a associates the search keyword 6a with the selected content 7b and stores it in the storage unit 1b. Store. Specifically, when the keyword 6a is output from the client 2, the search based on the keyword 6a is performed by the search server 3. Then, the search result 6b is returned to the client 2. When the user using the client 2 selects any content from the search result 6b, a content acquisition request 7a is output to the content server 4. The content server 4 returns the content 7b corresponding to the content acquisition request 7a to the client 2. At this time, the storage unit 1a collects the identification information of the content 6b indicated by the keyword 6a and the content acquisition request 7a, and stores them in the storage unit 1b in association with each other.

  The grouping unit 1c determines the relevance between the keywords stored in the storage unit 1b based on the correspondence between the keyword and the selected content, and groups a plurality of related keywords. For example, the grouping unit 1c groups keywords having common contents associated with each other. Specifically, when the two keywords are compared, the grouping unit 1c can group the two keywords if the number of commonly associated contents is equal to or greater than a predetermined value. Furthermore, the grouping unit 1c compares the two keywords when the proportion of the content associated with at least one of the two keywords is equal to or greater than a predetermined value. Two keywords can be grouped.

  The related keyword output unit 1 d accepts input of representative keywords from the client 5. Then, the related keyword output unit 1d outputs other keywords belonging to the same group as the selected representative keyword to the client 5 when an arbitrary representative keyword is selected.

  According to such a content navigation apparatus, whenever the user performs a search based on the keyword 6a and selects an arbitrary content 7b from the search result 6b, the storage means 1a selects the keyword 6a for search. The content 7b is associated and stored in the storage unit 1b. Thereafter, the grouping unit 1c determines the relevance between the keywords stored in the storage unit 1b based on the correspondence between the keyword and the selected content, and groups a plurality of related keywords. When an arbitrary representative keyword 8a is selected, the related keyword output unit 1d outputs another related keyword 8b belonging to the same group as the selected representative keyword 8a.

  As described above, the keywords are grouped on the basis of the relationship between the keyword and the content selected by the user from the search result by the keyword, so that it is related to the content that many users are currently interested in. Keywords can be grouped together. As a result, by outputting other keywords in the same group as the selected representative keyword, it is possible to present the keywords input by other users with similar interest targets to the user.

  The technique of the present invention as shown in FIG. 1 can be used for content navigation in various networks such as the Internet and an intranet. In particular, the convenience of the Internet can be improved by applying it to the Internet where a large amount of content is disclosed.

  Note that user identification information can also be used as a grouping determination factor. In that case, in addition to the content identification information, the storage unit 1a stores the user identification information of the user who entered the keyword in the storage unit 1b in association with the keyword. Then, the grouping unit 1c groups keywords having the same associated user identification information. Specifically, when the two keywords are compared, the grouping unit 1c groups the two keywords when the number of commonly associated user identification information is greater than or equal to a predetermined value. Further, when the grouping means 1c compares two keywords, the proportion of the user identification information commonly associated with the user identification information associated with at least one of the two keywords is a predetermined value. In the above case, two keywords can be grouped. By grouping using the user identification information in this way, it is possible to group keywords that are input by the same user of interest.

  Furthermore, the grouping range can be expanded by repeating the grouping. Specifically, the grouping unit 1c groups one keyword included in the storage unit 1b as a target word, another keyword related to the target word as a corresponding word, and groups the target word and the corresponding word. . Furthermore, the grouping means 1c includes other keywords related to the corresponding word in the same group.

  Further, an unnecessary word storage unit is added to the content navigation device 1 to store unnecessary keywords stored in the unnecessary word storage unit as unnecessary words from the user, and the related keyword output unit 1d stores the unnecessary word storage unit in the unnecessary word storage unit. Keywords other than the unnecessary words may be output. Specifically, the unnecessary word storage means determines the relationship between the unnecessary word selected by the user and another keyword based on the correspondence relationship between the keyword and the selected content, and relates to the unnecessary word. The other keywords (associated with the same content) are stored in the unnecessary word storage unit as new unnecessary words. By registering unnecessary words in this way, it is possible to prevent unnecessary related keywords (related keywords that are related only to the same content as other related keywords) from being output.

  In addition, content associated with each of a plurality of grouped keywords can be collected as association with one keyword (degeneration process). Specifically, the grouping unit 1c sets one keyword included in the storage unit 1b as a target word, and sets other keywords related to the target word as corresponding words. Then, the grouping unit 1c registers the content associated with each corresponding word in the storage unit 1b in association with the target word and deletes the corresponding word from the storage unit 1b. After performing the degeneration processing in this way, the grouping means 1c determines the relevance between the keywords stored in the storage means 1b, and groups a plurality of related keywords. That is, degeneration processing and grouping are performed alternately. Thus, even when grouping is repeated and the range to be grouped is expanded, it is possible to prevent a large amount of keywords from being output at once as related keywords.

  It is also possible to combine a plurality of already generated groups into a tree structure. Specifically, when the two groups generated by grouping include a common keyword, the grouping unit 1c connects the other group to a tree structure under one group via the common keyword. Create a new group.

  Furthermore, keywords can be grouped according to the number of hits when searching by keyword and the same transition of the number of users who input the keyword. Specifically, the storage unit 1a stores the number of hits of search by keyword and the user identification information of the user who has input the keyword in the storage unit 1b in association with the keyword. Then, the grouping unit 1c groups a plurality of keywords having similar temporal transitions in the number of hits. Alternatively, the grouping unit 1c groups a plurality of keywords with similar temporal transitions of the input number of users. In this way, by grouping in consideration of the temporal transition, it is possible to provide the user with information that has attracted attention.

  By the way, if the content navigation according to the present invention is used on the Internet, a huge amount of information updated every day on the Internet can be used effectively. When the present invention is applied to the Internet, for example, the above function is constructed at the first entrance (portal site server) when the user accesses the Internet. This portal site server mines the user's search behavior pattern. In the portal site server, content navigation based on the search keyword is performed.

  In the following, embodiments of the present invention will be specifically described by taking the case where the present invention is applied to a portal site on the Internet as an example. In the following embodiment, grouping keywords is referred to as clustering, and each generated group is referred to as a cluster.

  FIG. 2 is a diagram showing an example of a system configuration for realizing the embodiment of the present invention. As shown in FIG. 2, a portal site server 100, a plurality of clients 211, 212,..., A search server 220, and a plurality of Web servers 231, 232,.

  The portal site server 100 holds a web search history and provides a content navigation service to the clients 211, 212,... Based on the search history.

  The search server 220 receives a search request from the clients 211, 212,... Via the portal site server 100, and returns a Web page search result. The Web servers 231, 232,... Provide various contents such as Web pages via the Internet 10.

  FIG. 3 is a diagram illustrating a hardware configuration example of the portal site server used in the embodiment of the present invention. The entire portal site server 100 is controlled by a CPU (Central Processing Unit) 101. A random access memory (RAM) 102, a hard disk drive (HDD) 103, a graphic processing device 104, an input interface 105, and a communication interface 106 are connected to the CPU 101 via a bus 107.

  The RAM 102 temporarily stores at least part of an OS (Operating System) program and application programs to be executed by the CPU 101. The RAM 102 stores various data necessary for processing by the CPU 101. The HDD 103 stores an OS and application programs.

  A monitor 11 is connected to the graphic processing device 104. The graphic processing device 104 displays an image on the screen of the monitor 11 in accordance with a command from the CPU 101. A keyboard 12 and a mouse 13 are connected to the input interface 105. The input interface 105 transmits a signal transmitted from the keyboard 12 or the mouse 13 to the CPU 101 via the bus 107.

  The communication interface 106 is connected to the Internet 10. The communication interface 106 transmits and receives data to and from other computers via the Internet 10.

  With the hardware configuration as described above, the processing functions of the present embodiment can be realized. 3 shows an example of the hardware configuration of the portal site server 100. However, the clients 211, 212,..., The search server 220, and the Web servers 231, 232,. Can be realized.

  The function according to the present invention is provided by the portal site server 100. That is, the portal site server 100 creates two types of keyword-based clusters using the search history (search keyword, user identification ID, and search destination URL) recorded in the search site. Then, the portal site server 100 constructs a network that performs navigation according to the user's interest from the created cluster. Hereinafter, the processing function of the portal site server 100 will be described in detail.

  FIG. 4 is a block diagram showing processing functions of the portal site server. The portal site server 100 is provided with a Web page DB 111, a search history DB 112, a basic DB 113, an unnecessary word DB 114, and a cluster DB 115 as databases. Web page DB 111 stores Web page data provided as a portal site. The search history DB 112 stores search histories related to Web searches executed via the portal site server 100. The basic DB 113 stores information indicating the association between information obtained as a result of analyzing the contents of the search history. The unnecessary word DB 114 stores information on terms (unnecessary words) that need not be presented to the user in content navigation. The cluster DB 115 stores information indicating the degree of relevance between keywords based on the search history.

  The portal site server 100 is provided with a portal site content providing unit 120, a search history recording unit 130, a keyword cluster creating unit 140, an unnecessary word DB creating unit 150, and a navigation unit 160 as processing functions. The portal site content providing unit 120 provides the Web page data stored in the Web page DB 111 to the clients 211, 212,. The search history recording unit 130 records, in the search history DB 112, the history of Web search performed by the clients 211, 212,... Using the search server 220. The keyword cluster creation unit 140 constructs a basic DB 113 and a cluster DB 115 based on the search history DB 112. The unnecessary word DB creation unit 150 registers the term designated as an unnecessary word by the administrator in the unnecessary word DB 114. In response to a request from the clients 211, 212,..., The navigation unit 160 extracts keywords related to the specified keyword from the cluster DB 115 and transmits them to the clients 211, 212,.

First, search history accumulation processing in the portal site server 100 will be described.
FIG. 5 is a diagram showing search history accumulation processing. In the example of FIG. 5, a flow from when the user using the client 211 searches for content to select and browse the content of the Web server 231 from the search result is shown.

  The client 211 transmits a search page acquisition request 21 to the portal site server 100 in response to an operation input from the user. The portal site content providing unit 120 of the portal site server 100 transmits the search page data 22 to the client 211 in response to the search page acquisition request 21. The search page data 22 is a structured document described in, for example, HTML (HyperText Markup Language).

  In the client 211, the search page 23 is displayed on the monitor. On the search page 23, a search keyword input part 23a and a search button 23b are displayed. Although omitted in the figure, the search page 23 also displays various other information (news etc.).

  The user inputs one or more keywords to the search keyword input unit 23a and presses the search button 23b. Then, the client 211 transmits a search request including the search keyword to the portal site server 100.

  The search history recording unit 130 of the portal site server 100 transfers the search request 24 to the search server 220. The search server 220 searches for content on the Internet 10 in accordance with the received search request 24. Then, the search server 220 transmits the search result 25 to the portal site server 100.

  The search history recording unit 130 of the portal site server 100 transfers the search result 25 to the client 211. At this time, the search history recording unit 130 processes the search result 25 into Web page data. Control information is embedded in the generated Web page data so that when the user selects a URL in the search result, access to the URL is executed via the portal site server 100. Further, the search history recording unit 130 stores predetermined information (search keyword, number of hits, etc.) in the search result 25 in the search history DB 112 in association with information (for example, a cookie) for uniquely identifying the client 211. .

  Upon receiving the search result 25, the client 211 displays the search result list 26 on the monitor. The search result list 26 includes content identification information 28a, 28b,... (For example, a title or URL (Uniform Resource Locator)) that matches the search keyword. Here, when the user selects content identification information published on the Web server 231, a Web page acquisition request 27 specifying a corresponding URL is output from the client 211.

  The search history recording unit 130 of the portal site server 100 transfers the Web page acquisition request 27 to the Web server 231 and stores the URL information of the accessed Web page in the search history DB 112.

  The web server 231 that has received the web page acquisition request 27 transmits the corresponding web page data 28 to the client 211. On the client 211, the Web page 29 is displayed.

  FIG. 6 is a diagram illustrating an exemplary data structure of the search history DB. The search history DB 112 stores search histories 112a, 112b, 112c,... When a Web search is performed via the portal site server 100.

  In each search history 112a, 112b, 112c,..., The search date and time, the ID of the session in which the search was performed, the search keyword, the number of hits, the search result acquisition range (from what number of page information was acquired ), A user ID (user identification information), a URL appearing at the top of the search result, a URL (jump destination) selected by the user in the search result, a search type, and a title of the searched page. .

  For example, in the search history 112a, the search date and time “2003/12/12: 00: 00: 34”, the session ID “1111111111111111111111111111”, the search keyword “Russian folk song Trojan”, the number of hits “hn = 478”, the search result acquisition range “Ri = 10: 21” (10 items from the 21st content), top URL “GU =” ”, cookie“ ck = xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx ”, selected URL“ http://www.xxx.ne. “jp /”, search type “b = NORMAL” The title “t = Troika” of the selected page is set.

  In the search history 112a, the top page information in the search results is not acquired, and therefore the top URL is blank. In the search history 112, since the top page information is acquired, the top URL “GU =“ http://www.yyy.co.jp/ ”is registered.

  In an environment (ISP or the like) in which such a history is accumulated in a large amount every day (1 million cases / day or more), creating a meaningful keyword cluster (so-called keyword mining) is the point of this case. The keyword cluster creation process is executed periodically (for example, once a day at a predetermined time).

  Hereinafter, a process for creating a keyword cluster from the search history will be described in detail. The keyword cluster is a grouping of keywords according to a predetermined viewpoint. In the present embodiment, clusters of different levels (viewpoints) are created by the following two types of operations.

  Level 1 is a clustering of keywords having common contents selected from the search results. In general, keywords included in the cluster of level 1 are often related to variations in notation or different notations. For example, “personal computer”, “personal computer”, “personal computer”, “PC”, etc. constitute a level 1 cluster. A level 1 cluster can be created using the degree of matching of URLs selected by the user in the search results.

  FIG. 7 illustrates level 1 clustering. FIG. 7 shows the relationship between the search keyword and the URL selected by the user as a result of searching with the keyword. For example, this indicates that the user who searched for “Kwd1” has selected “URL1”, “URL2”, and “URL3” from the search results. At this time, it can be assumed that the keywords with high commonality (overlapping degree) of the selected URLs have the same or similar contents. Therefore, keywords having high commonality of the selected URLs are clustered to obtain a level 1 cluster.

Specifically, when a pair of search keyword and URL is considered, the overlapping degree of the selected URL is defined as dup as follows.
dup (Kwd1, Kwd2) = (number of overlapping URLs of Kwd1 and Kwd2) / (total of URLs of Kwd1 and Kwd2)
With respect to all keywords included in the search log, the above-described overlap degree (dup) and the number of simple overlap URLs are calculated, and a keyword group having a certain value or more as a threshold value is defined as a level 1 cluster.

  In the example of FIG. 7, “Kwd1” and “Kwd2” are associated with a common URL. Therefore, “Kwd1” and “Kwd2” constitute a cluster 31. Similarly, “Kwd3” and “Kwd4” are associated with a common URL. Therefore, “Kwd3” and “Kwd4” constitute a cluster 32.

  FIG. 8 is a diagram illustrating the level 1 overlap frequency between two keywords. In the example of FIG. 8, the URLs selected from the search result of “keyword A” are “URL1”, “URL2”, “URL3”, “URL4”, and “URL5”. The URLs selected from the search result of “keyword B” are “URL4”, “URL5”, “URL6”, and “URL7”. Therefore, the overlapping URLs are “URL4” and “URL5”, and the number of overlaps is 2.

  Level 2 is for clustering keywords that are duplicated by the input user. The keywords obtained in the level 2 cluster are generally many at the sibling and related word level. For example, keywords such as “XX camera”, “ΔΔ or”, “□□ electric”, “xx camera”, “convex and convex electric” are grouped as level 2 clusters. Level 2 clusters can be created using keywords retrieved by multiple users.

  FIG. 9 is a diagram illustrating level 2 clustering. FIG. 9 shows the relationship between the user ID and the keyword input by the user of the user ID. For example, this indicates that the user “USR1” has input “Kwd1” and “Kwd2” as search keywords. At this time, it can be assumed that keywords having a high commonality (dup value) of users to be input belong to the same category.

  That is, when a user searches for information by searching, it is common to search by trial and error while inputting various keywords to the target. A group of keywords searched in the same manner by a plurality of users is considered to have some commonality. Therefore, keywords having high commonality among the input users are grouped into a level 2 cluster.

Specifically, the overlapping degree dup can be defined as follows in the same manner as in level 1.
Dup (Kwd1, Kwd2) = (number of common users for Kwd1 and Kwd2) / (total number of users searched for Kwd1 and Kwd2 respectively)
For all the keywords included in the search log, the above-mentioned overlap degree (dup) and the number of simple overlap users are calculated, and a group of keywords whose values are equal to or greater than a certain threshold is defined as a cluster.

  In the example of FIG. 9, “Kwd1” and “Kwd2” are keywords that are commonly searched for by “USR1” and “USR2”. Therefore, “Kwd1” and “Kwd2” are grouped as a level 2 cluster 33. Similarly, “Kwd3” and “Kwd4” are keywords commonly searched for by “USR3” and “USR4”. Therefore, “Kwd3” and “Kwd4” are grouped as a level 2 cluster 34.

  FIG. 10 is a diagram showing the level 2 overlap frequency between two keywords. In the example of FIG. 10, the users selected from the search result of “keyword A” are “USR1”, “USR2”, “USR3”, “USR4”, and “USR5”. The users selected from the search result of “keyword B” are “USR4”, “USR5”, “USR6”, and “USR7”. Therefore, the overlapping users are “USR4” and “USR5”, and the number of overlaps is 2.

Thus, in the embodiment of the present invention, two types of clusters of level 1 and level 2 are generated.
FIG. 11 is a diagram illustrating an example of a cluster to which the keyword “computer” belongs. FIG. 11 shows what other keywords are included in the level 1 cluster 41 and the level 2 cluster 42 to which the keyword “computer” belongs. In this example, keywords such as “personal computer” and “personal computer” are set as the level 1 cluster 41 for “computer”. Further, keywords such as “computer virus” and “security hole” are set as the level 2 cluster 42 for “computer”.

  Thus, a keyword network can be constructed from a search history by regarding a keyword that appears in common between the clusters as a link. A comparison between a general directory and content navigation from the viewpoint of user navigation is as follows.

  FIG. 12 is a diagram illustrating a data structure of the directory method and the content navigation method. 12A shows the data structure of the directory system, and FIG. 12B shows the data structure of the content navigation system.

  In the directory method, the relationship between keywords is a tree structure 51, whereas in the method according to the present embodiment, the relationship between keywords is a network structure 53. In the directory system, navigation is in a direction to narrow down the concept, whereas in the content navigation system, navigation is in a direction to expand the concept. That is, in the directory method, the concept is narrowed down by sequentially searching for keywords from the original keyword 52 to a lower structure on the tree structure 51. On the other hand, in the content navigation system, the concept can be expanded by sequentially searching other keywords having relevance on the cluster from the original keyword 54 and searching for the keyword.

  Due to these differences, directory usage scenes are when the target image is clear, whereas content navigation usage scenes are when the target image is not clear. is there.

In order to efficiently perform such clustering of keywords, a basic DB 113, an unnecessary word DB 114, and a cluster DB 115 are created in advance based on the search history. These DBs are created regularly (for example, every day at midnight).

  FIG. 13 is a diagram illustrating a data structure example of the basic DB. The basic DB 113 includes a keyword-specific URL management table 113a, a URL-specific keyword management table 113b, a keyword-specific selection user management table 113c, a user-specific input keyword management table 113d, a URL-specific selection user management table 113e, and a user-specific selection URL management table 113f. Consists of.

  The keyword-specific URL management table 113a includes columns for keywords, URLs, hit counts, and top URLs, and information arranged in the horizontal direction in each column is associated with each other to form one record. The keyword input at the time of search is set in the keyword column. In the URL column, the URL of the Web page selected from the search result by the corresponding keyword is set. The number of Web pages that match the corresponding keyword is set in the hit count column. In the top URL column, the URL of the Web page displayed most preferentially in the search result by the corresponding keyword is set.

  The URL-specific keyword management table 113b includes columns for URL, URL title, keyword, and search type, and information arranged in the horizontal direction of each column is associated with each other to form one record. In the URL column, the URL of the Web page selected from the search results is set. In the URL title column, the title of the Web page indicated by the corresponding URL is set. In the keyword column, the original search keyword when the corresponding URL is selected is set. In the search type column, the type of web page (a type such as advertisement or commerce (electronic commerce)) displayed as a search result is set.

  The keyword-specific selection user management table 113c includes columns for keywords and users, and information arranged in the horizontal direction of each column is associated with each other to form one record. A search keyword input by the user is set in the keyword column. In the user column, identification information of the user who has input the corresponding keyword is set.

  The user input keyword management table 113d is provided with columns of users and keywords, and information arranged in the horizontal direction of each column is associated with each other to form one record. In the user column, identification information of the user who performed the search is set. In the keyword column, a search keyword input by the corresponding user is set.

  The URL-selected user management table 113e has columns of URL and user, and information arranged in the horizontal direction of each column is associated with each other to form one record. In the URL column, the URL selected from the search result is set. In the user column, the identification information of the user who selected the corresponding URL is set.

  The user-selected URL management table 113f is provided with columns of users and URLs, and information arranged in the horizontal direction of each column is associated with each other to form one record. In the user column, identification information of the user who performed the search is set. In the URL column, a URL selected by the corresponding user from the search result is set.

  FIG. 14 is a diagram illustrating a data structure example of the unnecessary word DB. The unnecessary word DB 114 is provided with columns of unnecessary words, derivation words, number of repetitions, level 1 overlap number, level 1 dup value, level 2 overlap number, and level 2 dup value. Information arranged in the horizontal direction of the column is associated with each other to form one record.

  In the unnecessary word column, a keyword selected as an unnecessary word is set. In the derivation source word column, keywords that are compared as criteria for determining unnecessary words are set. In the number of repetitions column, a numerical value indicating how many unnecessary word detection processes have been performed when an unnecessary word is detected is set.

  In the level 1 overlap number column, the number of overlaps between the unnecessary words and the derivation source words when configuring the level 1 cluster is set. In the level 1 dup value column, a dup value between an unnecessary word and a derivation source word for configuring a level 1 cluster is set.

  In the level 2 overlap number column, the number of overlaps between unnecessary words and derivation source words when configuring a level 2 cluster is set. In the level 2 dup value column, a dup value between an unnecessary word and a derivation source word for configuring a level 2 cluster is set.

  FIG. 15 is a diagram illustrating an example of a data structure of the cluster DB. The cluster DB 115 is provided with columns of target words, number of users, number of hits, and corresponding words, and information arranged in the horizontal direction in each column is associated with each other to form one record.

  In the target word column, a keyword serving as a reference when searching for clusters is set. The total number of users who input the target word as a search keyword is set in the number of users column. In the hit count column, the hit count when the target word is searched as a search keyword is set.

  Information on other keywords (corresponding words) to be compared with the target word is set in the corresponding word column. The corresponding word column is further subdivided into notation, overlap number, dup value, user number, and hit number columns.

  In the column of notation, a character string for notation of the corresponding word is set. The overlap number column is divided into two upper and lower levels, the upper level is set to the level 1 overlap number between the target word and the corresponding word, and the lower level is the level between the target word and the corresponding word. An overlap number of 2 is set. The dup value column is divided into two upper and lower levels. The upper level contains level 1 dup values between the target word and the corresponding word, and the lower level indicates the level between the target word and the corresponding word. A dup value of 2 is set. In the number of users column, the number of users who input the corresponding word as a search keyword is set. In the number of hits, the number of pieces of information hit by a search when a corresponding word is input as a search keyword is set.

Next, a method for creating each database will be described. First, a procedure for creating the basic DB 113 will be described.
FIG. 16 is a flowchart showing a basic DB creation procedure. In the following, the process illustrated in FIG. 16 will be described in order of step number.

  [Step S1] The keyword cluster creation unit 140 refers to the search history DB 112, collects a plurality of different search histories such as search keywords and jump destination URLs as a single search history using a unique key such as a session ID, and in time order. Sort.

  [Step S2] The keyword cluster creation unit 140 reads the search history for each record and normalizes the keywords. In normalization, full-width alphanumeric symbols are changed to half-width uppercase letters, half-width katakana to full-width uppercase letters, full-width and half-width spaces at the beginning and end of lines are deleted, and multiple consecutive spaces contained in keywords are 1 Processing such as replacement with one half-width space is performed.

  [Step S3] The keyword cluster creation unit 140 creates a hash array for each combination of data, and stores the value in the basic DB 113. Specifically, a hash array indicating the URL corresponding to the keyword, a hash array indicating the user corresponding to the keyword, a hash array indicating the keyword corresponding to the user, a hash array indicating the URL corresponding to the user, and a keyword corresponding to the URL And a hash array indicating the user corresponding to the URL are created.

  [Step S4] The keyword cluster creation unit 140 determines whether all records have been processed. If all records have been processed, the process ends. If there is an unprocessed record, the process proceeds to step S2.

  FIG. 17 is a diagram illustrating an example of a hash array in which URLs corresponding to keywords are defined. As shown in FIG. 17, the hash array 61 associates each keyword with a list of URLs selected from the keyword search result, the number of hits (Hit #), and the top URL (TopURL).

  FIG. 18 is a diagram illustrating an example of a hash array in which users corresponding to URLs are defined. As shown in FIG. 18, the hash array 62 associates each URL with a list of users who have selected that URL.

  The hash array generated in this way indicates the basic DB 113 shown in FIG. That is, FIG. 13 shows the basic DB 113 in a tabular form in an easy-to-understand manner, but in the actual portal site server 100, the basic DB 113 is managed by a hash array.

  Next, a procedure for creating the unnecessary word DB 114 will be described. The search history of the portal site server 100 includes many keywords that do not need to be disclosed as a service to general users. Generally, it is impossible to delete all unnecessary words included in the search history, but it is possible to delete them efficiently by creating the unnecessary word DB 114 using the above-described keyword cluster creation method. it can. The unnecessary word DB 114 is created according to the following steps.

FIG. 19 is the first half of a flowchart showing a procedure for creating an unnecessary word DB. In the following, the process illustrated in FIG. 19 will be described in order of step number.
[Step S11] The unnecessary word DB creation unit 150 retrieves a search history within a predetermined period from the search history DB 112. For example, a search history for a period from the previous execution of the cluster DB creation process to the present is taken out.

  [Step S12] The unnecessary word DB creation unit 150 performs cleaning on keywords and links in the extracted search history. In the cleaning, for example, a full-width alphanumeric symbol of a keyword is converted to a half-width alphanumeric symbol, and an alphanumeric lowercase character is converted to an uppercase character. Also, the session ID and user ID associated with the URL are deleted.

  [Step S13] The unnecessary word DB creation unit 150 extracts predetermined information from each extracted search history, and stores the result in the basic DB 113. Specifically, the unnecessary word DB creation unit 150 aggregates (accumulates using each as a key) from each search history for each keyword, user, and destination URL. Then, the unnecessary word DB creation unit 150 registers the aggregation result in the basic DB 113 for registration.

  [Step S14] The unnecessary word DB creation unit 150 sorts the keywords in descending order of the number of unique users. Here, the number of unique users indicates the number of users who input each keyword. When obtaining the number of unique users, if the same user inputs the same keyword a plurality of times, they are collectively converted to one user.

  Specifically, the number of users registered in association with each keyword in the keyword-specific selected user management table 113c (after eliminating duplicate user IDs) is counted, and the keywords are rearranged in descending order. In addition, the number of unique users of each keyword may be grouped in units of time, day, week, and month, and sorted within the unit. At this time, those with a small number of unique users (such as 1 and 2 that do not exceed a predetermined threshold) may be deleted from the list.

  [Step S15] The unnecessary word DB creation unit 150 displays keywords on the screen in descending order of the number of unique users, and allows the user to select one or more keywords to be examined as to whether or not they should be unnecessary words. The unnecessary word DB creation unit 150 creates a seed list of unnecessary words including one or more keywords selected by the user. For example, the user selects keywords that always appear at the top when they are grouped by hour, day, week, or month. The seed list is stored in the RAM 102, for example.

  [Step S16] The unnecessary word DB creation unit 150 extracts an unprocessed keyword from the seed list, and obtains a user and URL corresponding to the keyword with reference to the basic DB 113. Specifically, the unnecessary word DB creation unit 150 refers to the keyword-specific selected user management table 113c and acquires a user ID corresponding to the extracted keyword. Further, the unnecessary word DB creation unit 150 refers to the keyword-specific URL management table 113a and obtains a URL corresponding to the extracted keyword.

  [Step S17] The unnecessary word DB creation unit 150 obtains a keyword corresponding to the user and URL obtained in Step S16, and further obtains the URL and user of the keyword. Specifically, the unnecessary word DB creation unit 150 refers to the user-specific input keyword management table 113d and obtains a keyword corresponding to each user obtained in step S16. Then, the unnecessary word DB creation unit 150 refers to the keyword-specific selected user management table 113c and the keyword-specific URL management table 113a to obtain a URL and a user corresponding to the obtained keyword.

FIG. 20 is the second half of the flowchart showing the procedure for creating the unnecessary word DB. In the following, the process illustrated in FIG. 20 will be described in order of step number.
[Step S18] The unnecessary word DB creation unit 150 calculates the number of URL overlaps and the dup function between the keyword obtained in Step S17 and the keyword extracted from the seed list. This is a process for determining whether or not there is a level 1 cluster relationship.

  [Step S19] The unnecessary word DB creation unit 150 calculates the number of overlapping users and the dup function between the keyword obtained in step S17 and the keyword extracted from the seed list. This is a process for determining whether or not there is a level 2 cluster relationship.

[Step S20] The unnecessary word DB creation unit 150 adds, to the unnecessary word DB 114, keywords whose overlapping number or dup function value is equal to or greater than a predetermined threshold.
[Step S21] The unnecessary word DB creation unit 150 displays the keyword added in step S20 on the screen, and accepts selection input of an inappropriate keyword as an unnecessary word from the user. When a keyword is selected by the user, the keyword is deleted from the unnecessary word DB 114.

  [Step S22] The unnecessary word DB creation unit 150 determines whether or not the unnecessary word registration process has been repeated a predetermined number of times. If it is repeated a predetermined number of times, the process is terminated. If it has not been repeated a predetermined number of times, the process proceeds to step S16 using the keywords newly added as unnecessary words in step S20 (except for those deleted in step S21) as a seed list.

  In this way, the unnecessary word DB 114 can be created. At this time, the unnecessary word extraction omission can be reduced by repeating the determination of the unnecessary word using the newly set unnecessary word as a seed list.

Next, a procedure for creating the cluster DB 115 will be described.
FIG. 21 is the first half of a flowchart showing the procedure of the cluster DB creation process. In the following, the process illustrated in FIG. 21 will be described in order of step number.

  [Step S31] The keyword cluster creation unit 140 retrieves a search history within a predetermined period from the search history DB 112. For example, a search history for a period from the previous execution of the cluster DB creation process to the present is taken out.

  [Step S32] The keyword cluster creation unit 140 cleans the keywords and links in the extracted search history. In the cleaning, for example, a full-width alphanumeric symbol of a keyword is converted to a half-width alphanumeric symbol, and an alphanumeric lowercase character is converted to an uppercase character. Also, the session ID and user ID associated with the URL are deleted.

  [Step S33] The keyword cluster creation unit 140 extracts predetermined information from each extracted search history, and stores the result in the basic DB 113. Specifically, the keyword cluster creation unit 140 aggregates the keywords, users, and jump destination URLs from each search history (aggregates using each as a key). Then, the keyword cluster creation unit 140 registers the aggregation results in the basic DB 113.

  [Step S34] The keyword cluster creation unit 140 deletes, from the basic DB 113, keywords whose number of unique users is equal to or less than a predetermined threshold and keywords included in the unnecessary word DB 114.

  [Step S35] The keyword cluster creation unit 140 sorts the keywords in descending order of unique users. Specifically, the keyword cluster creation unit 140 counts the number of users registered after being associated with each keyword in the keyword-specific selected user management table 113c (after eliminating duplication of user IDs), and in ascending order of the number. Sort keywords.

[Step S36] The keyword cluster creation unit 140 extracts keywords from the top of the sorted list, and sets the extracted keywords as target words. Then, the keyword cluster creation unit 140 obtains a user and URL corresponding to the target word. Specifically, the keyword cluster creation unit 140 refers to the keyword-specific selected user management table 113c and acquires a user ID corresponding to the extracted keyword. Further, the keyword cluster creation unit 140 refers to the keyword-specific URL management table 113a to obtain a URL corresponding to the extracted keyword.

[Step S37] The keyword cluster creation unit 140 obtains a keyword (corresponding word) corresponding to the user and URL obtained in Step S36, and further obtains the URL and user of the keyword. Specifically, the keyword cluster creation unit 140 refers to the user-specific input keyword management table 113d and obtains a keyword corresponding to each user obtained in step S36. Then, the keyword cluster creation unit 140 refers to the keyword-specific selected user management table 113c and the keyword-specific URL management table 113a to obtain a URL and a user corresponding to the obtained keyword.

FIG. 22 is the second half of the flowchart showing the procedure of the cluster DB creation process. In the following, the process illustrated in FIG. 22 will be described in order of step number.
[Step S38] The keyword cluster creation unit 140 calculates the number of URL overlaps and dup functions between the corresponding word obtained in Step S37 and the target word. This is a process for determining whether or not there is a level 1 cluster relationship. Further, the keyword cluster creation unit 140 calculates the number of overlapping users and the dup function between the corresponding word obtained in step S37 and the target word. This is a process for determining whether or not there is a level 2 cluster relationship.

  [Step S39] The keyword cluster creation unit 140 registers, in the cluster DB 115, a corresponding word in which either the number of overlaps obtained in step S38 or the value of the dup function is equal to or greater than a predetermined threshold is associated with the target word. At this time, the number of overlaps of level 1 and level 2 of the corresponding word and the value of the dup function are also registered.

  [Step S40] The keyword cluster creation unit 140 registers the URL and user information of the registered corresponding word in the basic DB 113 as target word information. Thereafter, the keyword cluster creation unit 140 deletes the registered corresponding word information from the basic DB 113. This process is called network degeneration.

  [Step S41] The keyword cluster creation unit 140 performs unnecessary word DB creation processing (see FIGS. 19 and 20) for the updated basic DB 113, and updates the unnecessary word DB 114.

  [Step S42] The keyword cluster creation unit 140 determines whether or not the update processing of the cluster DB 115 has been executed a predetermined number of times. If the predetermined number of times of execution is entered, the predetermined process ends. If not, the process proceeds to step S31. Thereby, the same processing is repeated a predetermined number of times.

  In addition, when a corresponding word is added to the cluster DB 115, information indicating how many times the process is performed can be additionally written. Further, the date of addition of the corresponding word can also be registered in the cluster DB 115. By registering the date information, it is possible to grasp the change of the cluster keyword over time.

  In this way, the cluster DB 115 can be created. At this time, as shown in step S40, the network is simultaneously degenerated. In the degeneracy of a network, an operation for grouping keyword groups that are particularly similar in notation fluctuation and concept is performed. By expressing a group of keywords that are particularly close to each other and representing them as one representative keyword, the number of keywords can be reduced and the scale of the navigation network can be reduced. As a result, when presenting keywords included in the cluster to the user, it is possible to efficiently present the minimum necessary keywords.

  Thus, by alternately extracting the set of keywords to be clustered and the reduction processing of those keywords, it is possible to reduce the situation where conceptually related keywords are leaked from the cluster.

  FIG. 23 is a diagram illustrating changes in clusters when clustering and keyword degeneration are repeated. For example, it is assumed that three corresponding words are included in the first level 1 cluster 63 when “Company A” is the target word. At this time, the user registered in the corresponding word is added to the user's item corresponding to the target word in the basic DB 113. Similarly, the URL registered in the corresponding word is added to the URL item corresponding to the target word.

  Thereafter, when the level 1 cluster creation process is executed again, a cluster 64 including other keywords expanded conceptually is generated. By repeating such processing an arbitrary number of times, it is possible to form a cluster in which keywords that are conceptually similar are collected.

  As described above, by creating the cluster DB 115, a content navigation function can be provided to a user who accesses the portal site server 100 and searches for content on the Internet. Content navigation is performed by the navigation unit 160.

  In the present embodiment, it is assumed that the URL of the top page of content navigation is prepared in advance. When the user operates the client 211 to access the URL, navigation top screen data is created by the navigation unit 160. In the present embodiment, the navigation unit 160 refers to the keyword-specific selected user management table 113c of the basic DB 113 and creates a list of keywords having a large number of users input as search keywords. Then, navigation top screen data including the keyword list is created. The navigation top screen data is transferred to the client 211, and the navigation top screen is displayed on the client.

  FIG. 24 is a diagram illustrating an example of the navigation top screen. The navigation top screen 71 displays a keyword input unit 71a, a jump button 71b, a plurality of keyword candidates 71c, the number of unique users 71d for each keyword candidate 71c, and a standard directory link object 71e.

  The keyword input unit 71a is a text box for inputting an arbitrary keyword as a representative keyword. The jump button 71b is a button for issuing a content navigation execution instruction. When the jump button 71b is pressed, a content navigation request using the keyword input to the keyword input unit 71a as a representative keyword is transmitted to the portal site server 100.

  The keyword candidate 71c is a keyword candidate to be selected as a representative keyword. In this example, keywords that are frequently input as search keywords by a user within a predetermined period in the past are displayed as keyword candidates 71c. When any one of the keyword candidates 71c is selected by the user, a content navigation request using the keyword as a representative keyword is transmitted to the portal site server 100. The number of unique users of each keyword candidate 71c is the number of users who input the keyword as a search keyword.

  The standard directory link object 71e is an object in which a link for transition to the standard directory screen is defined. When the standard directory link object 71e is selected, a request to display the standard directory screen is issued to the portal site server 100.

  Here, it is assumed that one of the keyword candidates 71c is selected by the user. Then, in the portal site server 100 that has received the content navigation request, the navigation unit 160 performs content navigation processing and transmits the processing result to the client 211. Specifically, the navigation unit 160 searches the cluster DB 115 for a record in which a keyword designated as a representative keyword is registered as a target word. Then, the navigation unit 160 extracts a predetermined number of keywords from the corresponding word of the detected record in descending order of the level 1 dup value (or number of overlaps), and similarly the level 2 dup value (or number of overlaps). ), A predetermined number of keywords are extracted in descending order. The navigation unit 160 creates navigation screen data including these keywords.

  Further, the navigation unit 160 refers to the keyword-specific URL management table 113a of the basic DB 113, and extracts a URL that has been selected from a search result based on the keyword specified as the representative keyword. Furthermore, the navigation unit 160 refers to the URL-specific selected user management table 113e, counts the number of users who have selected each extracted URL, and sorts the URLs according to the number of users. Then, the navigation unit 160 adds the sorted URL and related information of the URL to the navigation screen data.

The generated navigation screen is transmitted to the client 211 and displayed on the client 211.
FIG. 25 is a diagram illustrating an example of a navigation screen. On the navigation screen 72, a representative keyword 72a, a level 1 cluster keyword 72b, a level 2 cluster keyword 72c, and content information 72d are displayed.

  Next to the representative keyword 72a, the number of unique users who input the representative keyword as a search keyword is displayed. In the level 1 cluster keyword 72b, a keyword having a level 1 cluster relationship with the representative keyword is displayed. In the level 2 cluster keyword 72c, a keyword having a level 2 cluster relationship with the representative keyword is displayed.

  The content information 72d displays information related to content that is frequently selected from the search results when the representative keyword is used as the search keyword. The content information 72d includes a content access number 72e, a content title 72f, and a corresponding word 72g. The content access number 72e is the number of times that the displayed content is selected as an access target when the displayed content is displayed as a search result. The content title 72f is a content title. The corresponding word 72g is a keyword that can detect the content as a search result.

  Note that the navigation screen can be displayed so that the type of content can be identified. In this case, the navigation unit 160 refers to the search type column of the URL-based keyword management table 113b when creating the navigation screen data, and sets different display attributes for each search type. For example, the display color can be different for each search type.

  FIG. 26 is a diagram illustrating an example of a navigation screen in which type identification is possible. In the navigation screen 73, the content title 73a of the service content and the access number 73b of the advertisement content are displayed in a color different from that of the other content (in FIG. 26, the highlighted display portion is indicated by a broken line). . Service content is content provided on a website that provides services such as electronic commerce. The advertising content is advertising content such as corporate product promotion. By displaying the content type in different colors, the user can easily identify the target content.

  As described above, based on the search history, which content the user has selected from the search results is managed in the database, and keywords are clustered based on the database. Then, the representative keyword designated in the content navigation and other keywords clustered are presented to the user. As a result, the keyword related to the keyword designated by the user can be presented by appropriately reflecting the actual user's preference and the like.

Further, content arbitrarily designated by the user can be highlighted (for example, highlighted).
FIG. 27 is a diagram illustrating an example of a navigation screen in which arbitrary content is highlighted. The navigation screen 74 shown in FIG. 27 is provided with highlight designation parts 74a and 74b. In the highlight designating parts 74a and 74b, there is a character string input part indicating the content to be highlighted. The content including the character string input to the input unit in the URL or title is highlighted by highlighting.

  When the standard directory link object 71 e on the navigation top screen 71 is selected, the standard directory screen data is created by the navigation unit 160 and transmitted to the client 211. The standard directory screen data includes content that is regularly and frequently accessed.

  FIG. 28 is a diagram illustrating an example of a standard directory screen. The standard directory screen 75 displays content that is constantly accessed frequently. Next to the title of each content, the number of accessed unique users is shown.

  As described above, if the content navigation shown in the embodiment of the present invention is performed, the user can acquire the trendy keyword at that time by selecting the representative keyword. Then, the user can browse popular content widely. Also, it is possible to access all related contents without worrying about keyword notation.

  For example, the keywords collected for a certain period (hours, days, weeks, months, etc.), the number of unique users selected by the keyword, the degree of attention (increase from the average number of users), the number of keyword hits, the size of the cluster, etc. Sort by using the index and display the top keywords as the top of the directory search. The user can see what is happening now by seeing this.

  Note that such content navigation is user navigation different from directory search. That is, in content navigation, changes in preferences of users who use the network and changes in contents can be monitored, and appropriate navigation can be performed in accordance with changes in user interests.

  In addition, companies that conduct electronic commerce use the results of their site's SEO (SearchEngineOptimization: a technique for optimizing site keywords and composition so that users can search their site accurately) and SEM (SearchEngineMarketing: search keyword advertising). Can also be used to maximize the profits of their own site.

By the way, this embodiment can be applied as follows.
[Create content cluster]
In the above example, a cluster between keywords is created. However, contents related to a clustered keyword can be clustered. Specifically, a content cluster can be created using the cluster DB 115 and the basic DB 113. A group of keyword contents constituting a cluster is considered a cluster.

  FIG. 29 is a diagram illustrating an example of a content cluster. As shown in the figure, for each keyword included in the cluster 81 between keywords (either level 1 or level 2 or both), the URL associated on the basic DB 113 (from the search result by the corresponding keyword, the user The URL selected by (1) is extracted. Then, the content cluster 82 is configured by the extracted URL.

  In this way, by configuring the content cluster 82, when content is selected by the user, a list of other content belonging to the same cluster can be displayed on the screen. Thereby, the user can easily access other contents similar to the selected contents.

[Expand cluster by combining clusters]
The cluster for the representative keyword can be expanded step by step based on other keywords included in the cluster. For example, it is assumed that B, C, and D are extracted as clusters of the representative keyword A. This relationship is expressed as (A: B, C, D). At this time, if (B: C, E) is given to the representative keyword B, the cluster of A can be expanded to (A: (B: C, E), D).

  FIG. 30 is a diagram illustrating cluster expansion processing. As shown in FIG. 30, a cluster 93 is created by combining the cluster 91 and the cluster 92. In this example, the target word “C car” of the cluster 92 is one of the corresponding words of the cluster 91. Therefore, the corresponding word of the cluster 92 is associated as a substructure of the corresponding word “C car” of the cluster 91. At that time, the keywords that are originally set as the corresponding words of the cluster 91 are excluded from the objects of association with the subordinate structure of the corresponding word “C car”.

  Note that which cluster is lower can be determined by, for example, the number of search users or the number of hits in the search results. In the example of FIG. 30, the cluster 91 has a larger number of retrievals than the cluster 92, so that the cluster 93 is generated by arranging the cluster 92 under the cluster 91.

  By expanding the cluster in this way, it is possible to present related keywords in a wider range when performing content navigation of the representative keyword selected by the user.

[Retrieve cluster of interest]
Consider the temporal change in the number of search keyword hits and the number of unique users included in the basic DB 113 and the degree of attention by the user at that time. For example, if the number of users increases, the degree of attention is considered high. Also, if the number of hits increases, the degree of attention is considered high.

FIG. 31 is a diagram illustrating the degree of attention according to changes in the number of users and the number of hits. In this figure, the degree of attention is evaluated in five stages. The larger the value, the higher the degree of attention.
Thus, the attention level for each keyword is set in advance. When displaying a keyword belonging to the same cluster as the representative keyword, a keyword with a high degree of attention is preferentially displayed. Thereby, the keyword which shows the thing which attracts attention recently by many users among the keywords relevant to a representative keyword can be known easily.

[Cluster creation method that considers search terms with similar search patterns as the same cluster]
When plotting the increase / decrease in the number of searches for a search keyword with the vertical axis representing the number of searches per hour and the horizontal axis representing time, the keywords that increase or decrease at the same time are included in the same cluster. . Further, instead of the number of searches, a change in the attention level of the keyword may be seen. The keyword attention level is defined by the following expression, for example.

This is an evaluation formula for obtaining the attention degree AT _t (w _i ) when the number of users of the search word w _i at time t is represented by UU _t (w _i ) and the correction value thereof is represented by C _t .
FIG. 32 is a diagram comparing a daily transition state between the number of searches and the degree of attention. In this example, a “post office” graph 94 and a “New Year's card” graph 95 are shown. In the graphs 94 and 95, the vertical axis is a numerical value (the number of users or the degree of attention), and the horizontal axis is time. In the graphs 94 and 95, the change in the number of search users and the change in the value of attention are plotted for each hour in “New Year's card” and “Post office”, respectively. Changes in the value of the number of search users are indicated by broken lines 94a and 95a, and changes in the value of the attention level are indicated by broken lines 94b and 95b.

  This example is a record of the lottery date of the winning number of a New Year postcard with New Year's cards. As can be seen from FIG. 32, the number of search users, the rising time of the attention level, and the peak time are almost synchronized in the two keywords. In this way, keywords having the same change in either or both of the number of search users and the degree of attention are entered in the cluster. As a result, an effective cluster can be created only from the input search keyword information.

[Realization by content navigation program]
The above processing functions can be realized by a server computer of a client server system. In that case, a server program describing the processing contents of the functions that the portal site server 100 should have is provided. The server computer executes the server program in response to a request from the client computer. As a result, the processing function is realized on the server computer, and the processing result is provided to the client computer.

  The server program describing the processing contents can be recorded on a recording medium readable by the server computer. Examples of the recording medium readable by the server computer include a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory. Examples of the magnetic recording device include a hard disk drive (HDD), a flexible disk (FD), and a magnetic tape. Examples of the optical disc include a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only Memory), and a CD-R (Recordable) / RW (ReWritable). Magneto-optical recording media include MO (Magneto-Optical disk).

When distributing the server program, for example, portable recording media such as a DVD and a CD-ROM in which the server program is recorded are sold.
The server computer that executes the server program stores, for example, the server program recorded on the portable recording medium in its own storage device. Then, the server computer reads the server program from its own storage device and executes processing according to the server program. The server computer can also read the server program directly from the portable recording medium and execute processing according to the server program.

(Supplementary note 1) In a content navigation program for supporting content search,
Computer
A storage unit that associates the keyword for search with the identification information of the selected content and stores it in a storage unit each time a user performs a search based on the keyword and selects any content from the search results;
A grouping unit that determines a relevance between the keywords stored in the storage unit based on a correspondence relationship between the keyword and the selected content, and groups the plurality of related keywords;
When an arbitrary representative keyword is selected, related keyword output means for outputting the other keyword belonging to the same group as the selected representative keyword,
A content navigation program characterized by functioning as

(Additional remark 2) The said grouping means groups the said keywords with which the said related content is common, The content navigation program characterized by the above-mentioned.
(Supplementary Note 3) When the two keywords are compared, the grouping unit groups the two keywords when the number of the commonly associated contents is equal to or greater than a predetermined value. The content navigation program according to attachment 2.

  (Additional remark 4) When the said grouping means compares two said keywords, in the said content linked | related with at least one of the said two keywords, the ratio for which the said content linked | related is shared The content navigation program according to appendix 2, wherein the two keywords are grouped when the value is equal to or greater than a predetermined value.

(Supplementary Note 5) The storage means further stores user identification information of a user who has input the keyword in the storage means in association with the keyword,
The content navigation program characterized in that the grouping means groups the keywords with which the associated user identification information is common.

  (Appendix 6) The grouping means groups two keywords when the two keywords are compared, and the number of commonly associated user identification information is a predetermined value or more. The content navigation program according to appendix 5.

  (Additional remark 7) When the said grouping means compares two said keywords, the said user identification information linked | related commonly among the said user identification information linked | related with at least one of the said two keywords 6. The content navigation program according to appendix 5, wherein the two keywords are grouped when the ratio occupied by is greater than or equal to a predetermined value.

  (Additional remark 8) The said grouping means makes one said keyword contained in the said memory | storage means a target word, makes the said other keyword relevant to the said target word a corresponding word, the said target word, the said corresponding word, The content navigation program according to claim 1, wherein the other keywords related to the corresponding word are further included in the same group.

(Additional remark 9) The said computer is further functioned as an unnecessary word storage means to store the said keyword selected as unnecessary from a user as an unnecessary word in an unnecessary word storage means,
The content navigation program according to claim 1, wherein the related keyword output unit outputs the keyword excluding the unnecessary word stored in the unnecessary word storage unit.

  (Additional remark 10) The said unnecessary word storage means determines the relationship between the said unnecessary word selected by the user, and the said other keyword based on the correspondence of the said keyword and the said selected content, The content navigation program according to appendix 9, wherein another keyword related to the unnecessary word is stored in the unnecessary word storage unit as a new unnecessary word.

  (Additional remark 11) The said grouping means made the said one keyword contained in the said memory | storage means into an object word, made the said other keyword relevant to the said object word into a corresponding word, and was linked | related with each said corresponding word The content is registered in the storage unit in association with the target word and the corresponding word is deleted from the storage unit, and then the relevance between the keywords stored in the storage unit is determined, The content navigation program according to appendix 1, wherein the keywords are grouped.

  (Additional remark 12) In the said grouping means, when two groups produced | generated by grouping have the said common keyword, the said other group under the said one group is tree-structured via the said common keyword. 2. A content navigation program according to appendix 1, wherein a new group connected to is generated.

(Supplementary Note 13) The storage means further stores the number of search hits by the keyword in the storage means in association with the keyword,
The content navigation program according to appendix 1, wherein the grouping unit groups a plurality of the keywords having similar temporal transitions in the number of hits.

(Supplementary Note 14) The storage means further stores user identification information of a user who has input the keyword in the storage means in association with the keyword,
The content navigation program according to appendix 1, wherein the grouping means groups a plurality of the keywords with similar temporal transitions of the number of input users.

(Supplementary Note 15) In a content navigation method for supporting a search for content by a computer,
Whenever the user performs a search based on the keyword and selects any content from the search results, the storage means associates the keyword for search with the identification information of the selected content and stores it in the storage means. ,
A grouping unit determines a relationship between the keywords stored in the storage unit based on a correspondence relationship between the keyword and the selected content, and groups the plurality of related keywords;
The related keyword output means outputs another keyword belonging to the same group as the selected representative keyword when an arbitrary representative keyword is selected.
A content navigation method characterized by the above.

(Supplementary Note 16) In a content navigation apparatus for supporting content search,
A storage unit that associates and stores in the storage unit the keyword for search and the identification information of the selected content each time the user performs a search based on the keyword and selects any content from the search results;
Grouping means for determining a relevance between the keywords stored in the storage means based on a correspondence relationship between the keyword and the selected content, and grouping a plurality of related keywords;
When an arbitrary representative keyword is selected, related keyword output means for outputting the other keyword belonging to the same group as the selected representative keyword;
A content navigation apparatus comprising:

(Supplementary Note 17) In a computer-readable recording medium on which a content navigation program for supporting content search is recorded,
Computer
A storage unit that associates the keyword for search with the identification information of the selected content and stores it in a storage unit each time a user performs a search based on the keyword and selects any content from the search results;
A grouping unit that determines a relevance between the keywords stored in the storage unit based on a correspondence relationship between the keyword and the selected content, and groups the plurality of related keywords;
When an arbitrary representative keyword is selected, related keyword output means for outputting the other keyword belonging to the same group as the selected representative keyword,
A computer-readable recording medium on which a content navigation program is recorded.

It is a conceptual diagram of the invention applied to embodiment. It is a figure which shows the system configuration example for implement | achieving embodiment of this invention. It is a figure which shows the hardware structural example of the portal site server used for embodiment of this invention. It is a block diagram which shows the processing function of a portal site server. It is a figure which shows a search log | history accumulation process. It is a figure which shows the example of a data structure of search log | history DB. It is a figure which shows level 1 clustering. It is a figure which shows the overlap frequency of the level 1 between two keywords. It is a figure which shows the clustering of level 2. FIG. It is a figure which shows the overlap frequency of the level 2 between two keywords. It is a figure which shows the example of the cluster to which the keyword of a computer belongs. It is a figure which shows the data structure of a directory system and a content navigation system. 12A shows the data structure of the directory system, and FIG. 12B shows the data structure of the content navigation system. It is a figure which shows the data structure example of basic DB. It is a figure which shows the data structure example of unnecessary word DB. It is a figure which shows the data structure example of cluster DB. It is a flowchart which shows the preparation procedure of basic DB. It is a figure which shows the example of the hash array in which URL corresponding to a keyword was defined. It is a figure which shows the example of the hash array in which the user corresponding to URL was defined. It is the first half of the flowchart which shows the preparation procedure of unnecessary word DB. It is the latter half of the flowchart which shows the preparation procedure of unnecessary word DB. It is the first half of the flowchart which shows the procedure of cluster DB creation processing. It is the latter half of the flowchart which shows the procedure of cluster DB creation processing. It is a figure which shows the change of a cluster at the time of repeating clustering and keyword degeneracy. It is a figure which shows the example of a navigation top screen. It is a figure which shows the example of a navigation screen. It is a figure which shows the example of the navigation screen which can identify a type. It is a figure which shows the example of the navigation screen which highlighted the arbitrary content. It is a figure which shows the example of a classic directory screen. It is a figure which shows the example of a content cluster. It is a figure which shows the expansion process of a cluster. It is a figure which shows the attention degree according to the change of the number of users and the number of hits. It is the figure which compared the daily transition situation of the number of searches and attention degree.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Content navigation apparatus 1a Storage means 1b Storage means 1c Grouping means 1d Related keyword output means 2,5 Client 3 Search server 4 Content server 6a Keyword 6b Search result 7a Content acquisition request 7b Content 8a Representative keyword 8b Related keyword

Claims (5)

In a content navigation program for supporting content search,
Computer
Each time a user performs a search based on a keyword and selects any content from the search results, the identification information of the selected content and the identification information of the user who has input the keyword are input for the keyword for search. Storage means for associating and storing in the storage means,
With reference to the storage means, a keyword having a large number after deduplication of identification information of the associated user is selected as a target word, and for each non-selected keyword other than the target word stored in the storage means, the target words and the number of identification information of content associated with common between the non-selected keyword, divided by the sum of the identification information of the content associated with each of the said subject word and the non-selected keyword, the division result numerical A non-selected keyword having a predetermined value equal to or greater than a predetermined value as a corresponding word, the target word and the corresponding word are grouped into one group, content identification information associated with the corresponding word is associated with the target word, and stored in the storage unit regard the subject word stored and each non-grouped keyword other than the corresponding word, associate in common between the subject word and the non-grouped keywords The number of identification information obtained content, divided by the sum of the identification information of the content associated with each of the said subject word and non grouping keyword, division result of numerical said non grouping keywords than a predetermined value Grouping means included in one group,
Related keyword output means for outputting a keyword other than the representative keyword belonging to the same group as the input representative keyword when an arbitrary representative keyword is input by a user's operation input;
A content navigation program characterized by functioning as   2. The content navigation program according to claim 1, wherein the grouping unit deletes the corresponding word from the storage unit. In a content navigation method for supporting search of content by a computer,
  The computer is
  Each time a user performs a search based on a keyword and selects any content from the search results, the identification information of the selected content and the identification information of the user who has input the keyword are input for the keyword for search. Is stored in the storage means in association with
  With reference to the storage means, a keyword having a large number after deduplication of identification information of the associated user is selected as a target word, and for each non-selected keyword other than the target word stored in the storage means, the target The number of content identification information commonly associated with a word and a non-selected keyword is divided by the sum of the content identification information associated with each of the target word and the non-selected keyword, and the numerical value of the division result A non-selected keyword having a predetermined value equal to or greater than a predetermined value as a corresponding word, grouping the target word and the corresponding word into a group, associating content identification information associated with the corresponding word with the target word, For each ungrouped keyword other than the stored target word and the corresponding word, the target word and the ungrouped keyword are associated in common. The number of pieces of identification information of the obtained content is divided by the sum of the pieces of content identification information associated with the target word and the ungrouped keyword, and an ungrouped keyword whose division result is a predetermined value or more In one group,
  When an arbitrary representative keyword is input by a user operation input, a keyword other than the representative keyword belonging to the same group as the input representative keyword is output.
  A content navigation method characterized by the above. 4. The content navigation method according to claim 3, wherein, when grouping, the corresponding word is deleted from the storage means. In a content navigation apparatus for supporting content search,
  Each time a user performs a search based on a keyword and selects any content from the search results, the identification information of the selected content and the identification information of the user who has input the keyword are input for the keyword for search. Storage means for associating and storing in the storage means,
  With reference to the storage means, a keyword having a large number after deduplication of identification information of the associated user is selected as a target word, and for each non-selected keyword other than the target word stored in the storage means, the target Divide the number of content identification information commonly associated between the word and the non-selected keyword by the sum of the content identification information associated with the target word and the non-selected keyword, A non-selected keyword having a predetermined value equal to or greater than a predetermined value as a corresponding word, the target word and the corresponding word are grouped into one group, content identification information associated with the corresponding word is associated with the target word, and stored in the storage unit For each ungrouped keyword other than the stored target word and the corresponding word, the target word and the ungrouped keyword are associated in common. The number of pieces of content identification information obtained is divided by the sum of the content identification information associated with each of the target word and the ungrouped keyword, and an ungrouped keyword whose division result is equal to or greater than a predetermined value is obtained. Grouping means to be included in one group;
  Related keyword output means for outputting a keyword other than the representative keyword belonging to the same group as the input representative keyword when an arbitrary representative keyword is input by a user's operation input;
  A content navigation apparatus comprising:

Images