JP2002014996A - Bookmark system, document proposal method using bookmark and program recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To present a new document along a user's interest area with the resource defined as an object on the Internet. SOLUTION: The bookmark information on each user is unitarily managed by a bookmark server 1. The user can perform a bookmark operation to a user bookmark DB 14 from a client device 2 via a bookmark operation part 11. The server 1 extracts the preference information for each sorting folder by using periodically a user preference extraction part 12 and according to the bookmark information in each user stored in the DB 14. A new document proposal part 13 registers an appropriate new document on the DB 14 as a part of user bookmark from the retrieval result obtained for a directory server on the Internet 0 or a document set 3 given from another outside part in response to the preference information on each user's sorting folder.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for proposing a new document suitable for a user's preference from a vast amount of information published on a network such as the Internet. The present invention relates to a document proposal technique capable of proposing a new document that is extremely suitable for a user's preference without putting a burden on the user.

[0002]

2. Description of the Related Art At present, a large amount of information is disclosed by information servers distributed on a network using a hypermedia mechanism represented by the World Wide Web (WWW). The user can acquire and browse documents held by these information servers from a browsing device or browsing software (browser) on a PC or mobile terminal. The amount of documents that can be browsed on this WWW is enormous, and technologies that support efficient access to the information that users need are indispensable.

It is considered that there are two types of access to a hypermedia document by a general user: regular access to a document browsed on a daily basis, and access for the purpose of discovering and browsing a new document.

As a function to support the former operation, Micros
oft Internet Explorer and Netscape Communicatio
Many browsers, such as ns' Netscape Navigator, have features called "favorites" and "bookmarks." The user can record the position information (URL) of the document that is accessed daily using this function.
As a result, it is possible to access a document to be browsed on a daily basis only by selecting the position information in the menu. This feature will be referred to as a “bookmark” hereafter.
Daily access to the document in the bookmark is referred to as “tour”.

A service that supports the latter operation includes a directory service. This is a service that manages a large number of documents by categorizing and indexing by keywords, and enables a user to search for necessary documents from categories and keywords. Yahoo! (Registered trademark) (http: // w
ww. yahoo. co. jp /), Infosee
k (http://www.infoseek.co.
jp /) are typical examples.

In addition to a new document discovery method using a directory service, a mechanism has been developed that automatically analyzes user preferences and recommends new documents based on the preferences. Examples of such a system include JP-A-10-307845, JP-A-11-15840,
JP-A-2000-48046 is an example.

[0007] The systems described in these publications obtain the preference information of each user based on the user's explicit input, the evaluation of the viewed document, and the operation performed by the user, and based on the information, acquire the preference information. Is recommended to the user. As a method of recommending a document, there are a method of sharing information of a document viewed between users having similar preferences, a method of recommending a document set provided as input from the system that matches the preference information, and the like.

[0008]

The first problem is that the access destination of the user is fixed and obsolete.

[0009] Many users can use bookmarks to streamline their daily access. However, users frequently bookmark new document URLs
Unless is registered, the bookmark information does not increase. For this reason, many general users tend to end up accessing a set of documents closed in a bookmark menu in which not many documents are registered.

[0010] The second problem is that it is difficult to find a new document required by the user.

For a general user, a new document is
Searching with a conventional directory server is very time-consuming. In particular, in the case of performing a search by specifying a keyword, it is troublesome to specify search conditions, and often a large number of search results are output, making it difficult to find necessary documents. Further, even when searching from a group of documents classified into categories, many categories are subdivided, and it takes time to find a document in a user's area of interest from among them.

A third problem is that it is not possible to effectively extract the preference information indicating the user's preference.

It is conceivable to introduce a function of automatically recommending a new document according to the user's preference as introduced in the prior art. However, in such a document automatic recommendation technique, the user's own preference information is input in advance. There is something that needs to be done, which places a heavy burden on the user. Further, even when the preference information is automatically extracted, there is a problem that, in the case of a user whose interest areas are widely distributed, the preference information of all the areas is put together and an appropriate document is not recommended.

A fourth problem is that the insertion position of the bookmark information by the system cannot be specified.

At the start of using a bookmark, the system prepares built-in data including a folder for inserting an advertisement or a new document. The system can automatically insert advertisement bookmarks and proposal documents according to the type of the folder. However, if the user deletes the folder of the built-in data at the time of editing the bookmark, the position where the system automatically inserts the bookmark information may not be specified.

Accordingly, an object of the present invention is to register, in a bookmark, bookmark information indicating a document extremely suitable for a user's preference from among a large number of documents published on a network without burdening the user. In addition to the above, it is possible to reliably prevent the system from being unable to specify the insertion position of the bookmark information.

[0017]

According to the present invention, user preference information is extracted based on a document indicated by bookmark information, and a new document is proposed in a user bookmark based on the preference information. With this feature,
The user receives a proposal for a related new document by registering a document of interest in the bookmark. This function can solve the first problem and the second problem. Also, since only bookmarks that the user is really interested in are registered in the bookmark, the preference of the user can be accurately reflected as compared with the method of proposing a document based on the operation performed by the user.

Further, in the present invention, preference information is extracted for each classified folder (category) in a bookmark, and a new document is proposed for each folder based on the extracted preference information. As a result, a new document can be proposed for each region of interest of the user, and even a user who is interested in various fields can receive a recommendation of a new document for each field. This can solve the third problem.

Further, the present invention realizes a function of restricting a user's editing operation with respect to an advertisement bookmark and a folder automatically inserted by the system. In particular, by prohibiting the deletion and movement of the folder of the built-in data prepared in advance by the system, the insertion position of the advertisement and the proposal document can be secured. Thus, the fourth problem can be solved.

[0020]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention, which comprises a bookmark server 1 and a plurality of client devices 2 mutually connected to the bookmark server 1. First, an outline of the configuration of the present embodiment will be described with reference to FIG.

The user operates a display unit in the client device 2.
Bookmark user interface (UI) displayed on 21
And the input unit 22 such as a mouse and a keyboard, the client device 2 in the bookmark information for each client device registered in the bookmark server 1.
Can access the bookmark information corresponding to. That is, the user can access the bookmark information by using the operation request unit configured by the bookmark UI displayed on the display unit 21 and the input unit 22.

The bookmark operation unit 11 in the bookmark server 1 performs operations such as registration, update, deletion, and display of new bookmark information in response to a user request. The bookmark information registered by each client device is stored in the user bookmark DB 14 while being distinguished for each registration source. The user preference extracting unit 12 extracts the preference information of each user based on a document published on the network indicated by the bookmark information registered in the user bookmark DB 14, and stores the information in the user preference information DB 15. The new document proposing unit 13 extracts the preference information for each user from the user preference information DB 15, and proposes a new document that is not registered in a bookmark that matches the extracted preference information to the user. As a method of acquiring a new document used in the proposal, there are a method of acquiring a new document via the Internet 0 and a method of selecting from a given document set 3.
Recording medium K1 connected to bookmark server 1
Is a disk, a semiconductor memory, or another recording medium in which a program for causing a computer to function as the bookmark server 1 is recorded. This program is read by a computer, and controls the operation of the computer, thereby realizing a bookmark operation unit 11, a user preference extraction unit 12, and a new document proposal unit 13 on the computer.

In the following, detailed operation of each part constituting this embodiment will be described.

First, the operation of the bookmark operation unit 11 will be described with reference to FIGS. Bookmark operation section 11
Then, the following operation can be performed on the bookmark DB 14 in response to a request from the client device 2.

-Browsing: Browsing bookmark information-Adding: Registering a URL pointing to a new document-Updating: Changing registration URL, folder title, comment-Deleting: deleting registration URL, folder-Moving: Display order of registration URL , Change of storage folder ・ Category folder creation: Create folder

The folders and URLs registered and browsed by the bookmark operation unit 11 are collectively called entries. This entry is divided into a system entry and a user entry.
Built-in bookmarks prepared in advance in the user bookmark DB 14 by the system (bookmark server 1), newly proposed documents based on user preferences, and advertisements are registered as system entries having respective types. Further, the bookmark information added by the user is registered as a user entry.

FIG. 2 is a detailed block diagram of the bookmark operation unit 11. The access management unit 111 manages operations on bookmark information from the user and the system.
If access is an entry registration operation by the system,
The system entry registration unit 114 registers the given entry as a system entry in the bookmark of the designated user. In the case of an entry registration operation by a user, the user entry registration unit 113 registers the given entry as a user entry in a bookmark of the user. For operations other than the registration operation, such as bookmark browsing, deletion, and movement, the bookmark browsing / editing unit 112 performs the corresponding operation on the user bookmark DB 14. At this time, if the operation target is a system entry, the editing operation according to the type of the entry becomes invalid. The type of the entry can be determined based on an entry flag described later.

FIG. 3 shows an example of bookmark information of a certain user in the user bookmark DB 14. Each line indicates one piece of bookmark information.
1, a bookmark URL 152, a title 153, a comment 154, a non-display flag 155, and a pointer. In this example, the entry where the value of the entry flag 151 is 0 is a user entry,
One or more entries indicate that they are system entries. In addition, the value of the entry flag 151 is different depending on the difference in the operation to be prohibited for the corresponding entry. In the present embodiment, 1 indicates the built-in bookmark information, and only movement is prohibited. Indicates a built-in folder or an advertisement bookmark, and delete, move, and update operations are prohibited. In addition, the case of 3 indicates a proposed document folder, and in addition to prohibiting delete, move, and update operations, the operation of registering a user bookmark in the folder is prohibited, and if 0, all operations are permitted. ing. In the bookmark URL 152, a URL pointing to a document and a folder name of a folder accommodating the URL are registered. Also, the non-display flag 155
Indicates whether to display at the time of browsing, and the pointer indicates the hierarchical relationship of the folders and is contained in each folder.
Indicates a URL.

FIG. 4 shows an example of a browsing user interface (browsing UI) returned from the bookmark server 1 in response to a browsing request from the client device 2. The viewing UI 200 is
It is composed of a folder hierarchy view 210 and a document view 220. The folder hierarchy view 210 displays the hierarchical structure of bookmarks. Each character string indicates a folder, and the folder indicates a category for classifying bookmarks. Folders with an icon at the beginning of the character string have subfolders below
In step 211, a character string without an icon at the beginning is a terminal folder 214 having no subfolders below. The degeneration icon 212 indicates that a subfolder exists under that folder, but the display is degenerated. The expansion icon 213 indicates that a subfolder exists under that folder and is currently expanded and displayed. For the folder of interest, the background color is emphasized as the current folder 215. URL pointing to the document
Are stored in a folder. In other words, the tree structure is such that a leaf portion is a URL pointing to a document, and a node portion is a folder.

The proposed document folder 216 under the top-level folder “sports car” contains a suggested document that matches the user's preference found based on the document pointed to by the URL in “sports car” and its subfolders. Is stored. In the proposal document folder 216, a preference document corresponding to the preference for “sports car” is proposed. The proposal documents are periodically recalculated by the system and each time the proposal documents folder 216
Is replaced with a new one. In the example of FIG. 4, other top-level folders such as “News / Topics” are not shown because they are displayed in a reduced manner. However, a proposal document folder is provided under each of the other top-level folders. I have.

In the document view 220, a document entry 221 indicated by a URL registered in the current folder 215 selected in the folder hierarchy view 210 is displayed.

Each folder in the folder hierarchy view 210 is
The system includes a system folder 217, which is a system entry prepared in advance by the system, and a user folder 218, which is a user entry created by the user. The user folder 218 can be freely moved and deleted during the bookmark editing operation, but the system folder 217 cannot be moved and deleted. In the system folder 217, users can freely
Although the URL can be registered, the user cannot register the URL only in the proposal document folder 216.

Each document entry in the current folder 215 of the document view 220 includes a system document area 223 for displaying a system entry which is a document prepared by the system in advance, and a system entry for displaying an advertisement periodically inserted by the system. An advertisement document area 222 and a user document area 224, which is a user entry for displaying a document registered by the user, are provided. The entry in the user document area 224 can be freely edited and operated by the user, such as moving and deleting, but the moving operation is prohibited in the system document area 223, and the document in the advertising document area 222 is moved and deleted. , Update editing operation is restricted.

Normally, the system folder cannot be deleted, but can be hidden on the user interface. FIG. 5 shows an edit user interface (edit U
I) Shows 250 examples. The hidden system folder 262 of the editing UI 250 contains the registration form 251, the update form 252,
By a delete button 254 of the edit panel 270 provided with a move button 253, a delete button 254, and a redisplay button 255,
It is an example of a hidden system folder. The hidden system folder 262 is not displayed in the viewing UI 200. However, in the editing UI 250, it is displayed together with the icon indicating that it is not displayed, and the redisplay button 255 of the editing panel 270 is displayed.
To return to the normal system folder. When the folder 262 is hidden by the delete button 254 as described above, the bookmark operation unit 11 sets the bookmark information non-display flag 155 of the folder 262 registered in the user bookmark DB 14 to “valid”. This indicates that the folder 262 is a hidden folder. Then, when browsing of a bookmark is requested by the client device 2, the folder 262 in which the non-display flag 155 is “valid” is not displayed in the folder hierarchy view 210. In addition, when an instruction to return to the normal system folder is given by the redisplay button 255, the non-display flag 155 is set to "invalid".

Next, the operation of the user preference extracting section 12 will be described with reference to FIGS. The extraction of the user preference is performed in the background irrespective of the user's browsing operation on the bookmark, and the preference information is extracted from the bookmarks of all the users in the user bookmark DB 14 at regular intervals.

FIG. 6 is a detailed block diagram of the user preference extracting section 12. The document collection unit 1201 has a function of acquiring text data of all registered documents of all users registered in the user bookmark DB 14 from the Internet 0 and storing the text data in the document archive 1204. This acquisition operation is performed first in the user bookmark DB 14
URL of all registered documents of all users registered in
Is listed. At that time, make sure that the same URL is not listed twice. Next, using software called a WWW crawler, the text data of the document corresponding to the listed URL is
The text data is downloaded from the WWW server described in L, and stored in the document archive 1204 as a pair with the URL.

The important keyword extracting unit 1202 analyzes the text data of the document stored in the document archive 1204 and extracts an important keyword. At this time, the URL of the document and the set of important keywords are combined and stored in the important keyword DB 1205. Subsequently, the feature keyword analysis unit 1203 obtains a bookmark registration document URL of each user from the user bookmark DB 14 and obtains an important keyword associated with the URL from the important keyword DB 1205. An important keyword for each category of the user bookmark is analyzed and registered in the user preference information DB 15 as the preference information of the user.

FIG. 7 is an example of a document stored in the document archive 1204, and FIG. 8 is a flowchart showing a process for extracting important keywords from the document. First,
The HTML text 500 is extracted from the document archive 1204 (FIG. 8, step S1001). Words are extracted from the beginning of the HTML text 500 (step S1002). The respective appearance scores are calculated (step S1003).

Here, the calculation method of the appearance score will be described by taking the words "soccer" and "ball" as examples. In the text, more occurrence scores are set for words used for titles, headings, highlighting, and the like, and for words that appear repeatedly, the appearance scores at each occurrence position in the text are added. Here, <title>
10 points, 5 points in <h1>, 4 points in <h2>, and 1 point in plaintext. The appearance score of the word "soccer"
<Title> sentence 501, <h1> sentence 502,503 and plaintext 504,507
Therefore, 10 points × 1 time + 5 points × 2 times + 1 point × 2 times = 22 points. In addition, the appearance score of “ball” is expressed in the <h2> sentence 50.
6 and plaintext 505,507, so 4 points x 1 time + 1 point x 2 times = 6 points.

Steps S1002 to S1003 are repeated until the appearance scores of all the words in the text have been calculated (step S1004). When the calculation of the appearance scores of all the words in the text is completed, the keyword having the higher appearance score is extracted (step S1005). Text location information (UR
L) and this set of keywords as important keywords DB1205
Is registered (step S1006). FIG. 9 shows an example of a set of a URL and a keyword group registered at this time. The important keyword group 522 and the URL 521 appearing in the text of the URL 521 are registered as a set in the important keyword DB 1205, and the URL 523 and the important keyword group 524 are similarly registered as a set. Steps S1001 to S1006 are repeated until all the important keywords of the text registered in the document archive 1204 have been registered (step S1007).

FIG. 10 is a list of bookmark URLs registered in a bookmark folder of a certain user's category "sports". FIG. 11 shows a characteristic keyword analysis unit 12 based on the important keywords of each text obtained above.
13 is a flowchart of a process of extracting user preference information in 03.
Attention is focused on one folder in the bookmark information of a certain user registered in the user bookmark DB 14 (FIG. 11, step S2001). Now, for example, if attention is paid to the folder 550 shown in FIG.
Based on the bookmark URLs 551, 553, and 555 in 0, the important keyword groups 552, 554, and 556 of the URLs 551, 553, and 555 are extracted from the important keyword DB 1205 (step S2002). Subsequently, the positive scores for each important keyword in the folder 550 are totaled. In the example of FIG. 10, “soccer”
It appears in the important keyword group 552,554 of URL 551,553,
The sum of each score "40" is a positive score of 557
(Step S2003). At this time, by looking at the registration date and time of the bookmark URL, the score of the important keyword of the new URL, which is higher, and the score of the important keyword of the old URL, which is lower, may be totaled. Step S2002 for important keywords of all documents in the folder
Steps S2003 to S2003 are repeated (step S2004).

Thereafter, all the folders other than the folder 550 of interest are regarded as one folder, and the above-described step S2002
The same operation as in steps S2005 to S200 is performed.
7). The sum of the appearance scores calculated for each keyword calculated here is referred to as a negative score. The calculation at this time is the same as the calculation method of the positive score 557.

A feature score for each keyword is calculated from the positive score and the negative score calculated above (step S2008). The feature score of each keyword calculated here is Sp for the positive score of a certain word W, Sn for the negative score, Np for the total number of occurrences of the word in the folder of interest, and Np for the word in other folders. When the number of appearances is Nn and the number of types of appearing words is Nw, it is calculated by the following equation (1). Feature score = ｛(Sp + 0.5) ÷ (Np + Nw × 0.5)｝ (Sn + 0.5) ÷ (Nn + Nw × 0.5)｝… (1)

Further, instead of using the equation (1), the feature score may be calculated by the following equation (2). Feature score = (Sp ÷ Np) ÷ (Sn ÷ Nn) ... (2)

According to equation (1) or equation (2), the folder (category) of interest has a high importance,
In other folders (categories), the feature score of a keyword that is not so important can be set high.

Some of the important keywords in the folder of interest that have a high feature score are extracted as the feature keywords of the folder (step S200).
9). The processing of steps S2001 to S2009 is repeated for all folders (step S2010). The characteristic keyword for each folder is registered as user preference information in the user preference information DB 15 (step S2011). FIG. 12 shows an example of the user preference information recorded at this time. The user preference information includes a category 601 and a characteristic keyword 602 in a folder holding each category information. The characteristic keyword 602 indicates the user's preference within the category 601.

Next, the detailed operation of the new document proposing section 13 having the configuration shown in FIG. 13 will be described with reference to FIGS.
This will be described with reference to FIGS. User information management unit 13
01 acquires the preference information of each user from the user preference information DB15. Based on this, the search proposal unit 1302 and the document filtering proposal unit 1303 select a proposal document.

The search suggestion unit 1302 includes a user information management unit 1301
A search is performed on a directory server that manages resources on the Internet by combining the preference keywords included in the preference information received from the Internet. The document filtering suggestion unit 1303 selects, from the externally provided document set 3, a document set including many preference keywords included in each user's preference information received from the user information management unit 1301 as a candidate document candidate for the user. Return to management unit 1301.

The user information management unit 1301 includes a search suggestion unit 1302
Alternatively, from among the suggested document candidates returned from the document filtering suggestion unit 1303, select an unregistered proposed document in the bookmark of the corresponding user, register the URL in an appropriate proposed folder of the user bookmark, and update the user bookmark DB14. I do. Thereafter, when the user browses the bookmark, a new document appears in the proposal folder.

FIG. 12 is an example of user preference information, and FIG. 14 is a flowchart showing the processing in the search suggestion unit 1302. FIG.
The characteristic keyword in the folder “computer” is extracted from the user preference information (step S3001 in FIG. 14). An order is assigned to the characteristic keywords in the order of “MP3”, “download”, “subnote” (step S3002). The ordering of the characteristic keywords at this time may be in the order of preference scores, or it may be effective to order the randomized numbers so that the proposed document is not fixed.

Next, the keyword “MP3” is extracted from the head of the characteristic keyword permutation, and the directory server is searched (step S3003). If there are too many search results, the following keyword “download” is added and a search is performed on the directory server using the search expression “MP3 AND download” connected by AND (step S3004). Add keywords to the search formula by AND until the search result is below the threshold. As a result of the addition, if the corresponding search result is lost, the result of the immediately preceding search expression is used as the proposal document.

When the search result falls below the threshold value, the URL of the document of the search result is passed to the user information management unit 1301 as a proposal document of the folder (steps S3005 and S3).
006). Steps S3001 to S3006 are repeated for all folders to be proposed by the user (step S3007).

FIG. 15 is a flowchart showing the operation of the document filtering proposal unit 1303. It is assumed that two documents pointed to by two URLs 521 and 523 shown in FIG. 9 are a document set 3 provided from outside.

The characteristic keywords of each folder are extracted from the user preference information from the user information management unit 1301 (FIG. 1).
5, step S3101). The document of the URL 521 shown in FIG. 9 is extracted from the document set 3 given from the outside (step S3102). The sum of the appearance scores of the characteristic keywords “soccer” and “ball” of the folder “sports” in the document is “28” because “soccer” is “22” and “ball” is “6” (step S3103). ). Steps S3102 to S3103 are repeated for all documents in document set 3 (step S3104).

When comparing the two documents in FIG. 9,
The characteristic keyword of the folder "sports" is "soccer",
The total appearance score in the document of the URL 521 of "ball" is "28", and the appearance score in the document of the URL 523 is "11" of only the "ball", so the URL 5 with a high appearance score
21 documents are passed to the user information management unit 1301 as proposal documents in the folder “Sports” (step
S3105). Steps S3101 to S3105 are repeated for all folders to be proposed by the user (step S3106)
.

At this time, the document set 3 includes:
Document lists created manually by service providers, advertisement information, bookmark document lists of similar taste users obtained by collaborative filtering, etc.
All or some of these may be combined.

[Another Embodiment of the Invention] As another embodiment of the bookmark system of the present invention, a configuration as shown in FIG. 16 can be considered. The feature of this embodiment is that the client device 2000 includes a user bookmark DB 14 for storing bookmark information of a user and a bookmark operation unit 11 for performing the bookmark operation.

In the present embodiment, the user preference extraction unit 12 of the recommendation server 1000 accesses the bookmark DB 14 in each client device 2000 via a network, thereby converting the user preference information into the user preference information in the recommendation server 1000.
Store in DB15. In addition, based on the user preference information stored here, the URL of the document proposed by the new document proposal unit 13 is stored in the bookmark DB 14 in the client device 2000 of the user via the network.

This configuration is considered to be effective when the user is determined to be a specific user with respect to the client device 2000 and the client device 2000 has a processing capability enough to manage the bookmark information of the user. Since it is not necessary to manage the huge amount of user bookmark DB 14 on the recommendation server 1000 side, the processing capacity and storage capacity of the server can be relatively small. Also, the bookmark operation unit 11 and the bookmark DB
Since 14 is located in the client device 2000, the response speed for registration, update, deletion, etc. to the bookmark is increased. still,
The recording media K2 and K3 provided in the recommendation server 1000 and the client device 2000 are disks, semiconductor memories, and other recording media.
0, a program for functioning as the client device 2000 is recorded. The program recorded on the recording medium K2 is read by a computer, and controls the operation of the computer, thereby realizing the user preference extracting unit 12 and the new document suggesting unit 13 on the computer. Further, the program recorded on the recording medium K3 is read by the computer, and controls the operation of the computer, thereby realizing the bookmark operation unit 11 on the computer.

As another embodiment of the bookmark system of the present invention, a configuration as shown in FIG. 17 can be considered. The feature of this embodiment is that the recommendation server 1100 includes only the new document proposal unit 13 and the client device 2100 includes the bookmark operation unit 11, the user preference extraction unit 12, the user bookmark DB 14, and the user preference information DB 15.

In this embodiment, each user operates the user bookmark DB 14 and extracts preference information from the bookmark in the client device 2100. The recommendation server 1100 periodically refers to the user preference information DB 15 stored in the client device 2100 via a network, and obtains the UR of the document proposed by the new document proposal unit 13.
L is stored in the user bookmark DB 14 in the client device 2100 of the user.

In this configuration, the user is determined to be a specific user with respect to the client device 2100, and has a relatively high processing capability for managing bookmark information of the user and extracting user preference information from the bookmark information. It is considered effective in the case. Recommendation server
In 1100, only the process of proposing a new document to each user needs to be performed, so that the load on the server is much smaller than in other embodiments. The recording media K4 and K5 provided in the recommendation server 1100 and the client device 2100 are a disk, a semiconductor memory, and other recording media. Programs for causing a computer to function as the recommendation server 1100 and the client device 2100 are recorded. Have been. The program recorded on the recording medium K4 is read by the computer, and controls the operation of the computer, thereby realizing the new document proposal section 13 on the computer. The program recorded on the recording medium K5 is read by the computer, and the operation of the computer is controlled, thereby realizing the bookmark operating unit 11 and the user preference extracting unit 12 on the computer.

Further, regarding the proposal of a new document,
In the example of FIG. 4, the proposal document folder 216 exists only in the top-level folder. However, not only the top-level folder as shown in FIG. 18 but also a registered document under a subfolder such as a proposal document folder 317 in the folder “soccer”. , A proposal document folder may be generated.

In a system having a proposed document folder in the above-mentioned subfolder, a proposed document folder may be generated only for the terminal or intermediate subfolder without generating a proposed document for a top-level folder.

FIG. 6 shows a method of extracting user preference information.
As a method for calculating important keywords in the document collection unit 1201 and the important keyword extraction unit 1202 of the user preference extraction unit 12, instead of extracting important keywords from a single text, a text set having a semantic Important keywords may be extracted as a unit.
This procedure will be described with reference to the flowchart of FIG.

First, the document archive 1204 shown in FIG.
Then, a text set having a semantic unit is extracted from (step S1101). The document collection unit 1201 registers the text registered in the user bookmark as the representative text, and registers the text associated therewith with a certain range of hyperlinks in the document archive 1204 as a text set. FIG. 20 is a schematic diagram showing an example of a text set. A text group consisting of the HTML text 702 linked by a hyperlink 703 within two columns from the representative text 701 is registered as a text set 704 in the document archive 1204.

The text set 704 is regarded as one large text, and words appearing from the beginning are extracted in order (step S1102). An appearance score is calculated for each word that has appeared (step S1103). The calculation method at this time is the same as step S1003 in FIG. Steps S1102 to S1103 are repeated until the scores of all the words in the text set 704 are calculated (step S1104).

When the calculation of the scores of all the words in the text set 704 is completed, the keyword having the higher appearance score is extracted (step S1105). A set of the position information (URL) of the representative HTML text 701 of the text set 704 and this keyword group is registered in the important keyword DB (step S1106). All text sets registered in document archive 1204
Steps S1101 to S1106 are repeated until the 704 important keywords are registered (step S1107).

Further, the calculation of the characteristic keyword in the characteristic keyword analysis unit 1203 can be simplified. FIG.
This procedure will be described. First, user bookmark DB14
The folder is extracted from the bookmark information of the user (step S2101). The important keywords of the documents registered in the folder are extracted from the important keyword DB 1205 (step S2102). Calculate the score for each important keyword. The score of the important keyword that first appears in the folder is the score in the important keyword DB1205,
If the same keyword appears, the score is added (step S2103). Steps S2102 to S2 for important keywords of all documents in the folder of interest
Step 103 is repeated (step S2104). Among the important keywords included in the folder of interest, some of the keywords having a high score are extracted as characteristic keywords of the folder (step S2105). The processing of steps S2101 to S2105 is repeated for all folders (step S2106). A user preference information database that summarizes the characteristic keywords of each folder and stores the user preference information
Registered in step 15 (step S2107).

[0071]

A first effect is that the user can access various unknown documents by a new document proposed as appropriate.

The second effect is that new registration, recalculation of taste,
By continuing new registration, the use of past resources and the discovery of new documents based on the user's preferences at that time can be efficiently and repeatedly performed without biasing to the same resource.

The third effect is that since a new document is presented for each folder (category) to be organized by the user, even a user who is interested in various fields can receive a proposal of a new document in an organized form. It is possible.

A fourth effect is that by limiting the user's editing operation on a bookmark automatically inserted by the system, the insertion position of a newly proposed document is secured and the user is prevented from deleting an advertisement bookmark. It is possible.

[Brief description of the drawings]

FIG. 1 is a block diagram of one embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration example of a bookmark operation unit 11.

FIG. 3 is a diagram illustrating an example of bookmark information.

FIG. 4 is a diagram illustrating an example of a browsing user interface.

FIG. 5 is a diagram illustrating an example of an editing user interface.

FIG. 6 is a block diagram illustrating a configuration example of a user preference extraction unit 12.

FIG. 7 is a diagram illustrating an example of an HTML text.

FIG. 8 is a flowchart illustrating a processing example of an important keyword extraction unit 1202;

FIG. 9 is a diagram showing an example of a document URL and a keyword group appearing in a text.

FIG. 10 is a diagram for explaining a positive score calculation method.

FIG. 11 is a flowchart showing a processing example of a feature keyword analysis unit 1203;

FIG. 12 is a diagram illustrating an example of user preference information.

FIG. 13 is a block diagram illustrating a configuration example of a new document proposal unit 13.

FIG. 14 is a flowchart showing a processing example of a search suggestion unit 1302;

FIG. 15 is a flowchart showing a processing example of a document filtering proposal unit 1303;

FIG. 16 is a block diagram of another embodiment of the present invention.

FIG. 17 is a block diagram of another embodiment of the present invention.

FIG. 18 is a diagram showing another example of the browsing user interface.

FIG. 19 is a diagram illustrating another processing example of the important keyword extraction unit 1202.

FIG. 20 is a diagram illustrating an example of a text set.

FIG. 21 is a flowchart showing another processing example of the characteristic keyword analysis unit 1203;

[Explanation of symbols]

 0 Internet 1 Bookmark server 2, 2000, 2100 Client device 3 Document set 11 Bookmark operation unit 12 User preference extraction unit 13 New document proposal unit 14 User bookmark DB 15 User preference information DB 21 Display Part 22 Input part 111 Access management part 112 Bookmark browsing / editing part 113 User entry registration part 114 System entry registration part 151 Entry flag 152… Bookmark URL 153… Title 154… Comment 155… Hidden flag 200, 300… Bookmark UI 210,310… Folder hierarchy view 211,311… Folder 212,312… Degenerate icon 213,313… Expand icon 214,314… End folder 215,315… Current folder 217… System folder 218… User folder 216,316,317… Proposed document folder 220,320… Document view 221… Document Entry 222, 321… Advertising document area 223… System document area 224… User document area 250… Edit UI 251… Registration form 252… Update form 253… Move button 254… Delete button 255… Redisplay button 260… Folder hierarchy view 261 … Folder entry 262… Hidden system folder 270… Edit panel 280… Document view 500… HTML text 501-507… HTML text 521,523… Bookmark URL 522,524… Important keyword group 550… Bookmark URL in category sports 551,553,555… Bookmark URL 552,554,556… Keyword 557… Positive score of word in category sports 601… Category name 602… Characteristic keyword in category 701… Representative HTML text 702… HTML text 703… Hyperlink 704… Text set 1000, 1100… Recommended server 1201 ... throb Instrument collection unit 1202 ... important keyword extraction unit 1203 ... feature keyword analysis unit 1204 ... document archive 1205 ... important keyword DB 1301 ... the user information management unit 1302 ... Search Tips section 1303 ... document filtering proposed section K1~K5 ... recording medium

Claims (29)

[Claims] 1. A user preference extraction unit for extracting preference information indicating user preference based on a document indicated by bookmark information registered in a bookmark, and a user preference information extracted by the user preference information extracted by the user preference extraction unit. Search for documents that match the user's preferences,
A new document proposal unit for registering bookmark information indicating the found document in the bookmark. 2. The bookmark system according to claim 1, wherein the user preference extracting unit collects a document pointed to by the bookmark information, and each of the documents in the document collected by the document collecting unit. Of the words, those whose appearance scores determined by their appearance frequency and appearance position are higher are extracted as important keywords,
A bookmark extracting unit that outputs the extracted important keyword as preference information indicating the user's preference. 3. The bookmark system according to claim 1, wherein the user preference extracting unit collects a document pointed to by the bookmark information, and each of the documents in the document collected by the document collecting unit. An important keyword extracting unit that extracts, as important keywords, an appearance score determined by its appearance frequency and appearance position as an important keyword, among the important keywords extracted by the important keyword extraction unit, A bookmark system, comprising: a feature keyword analysis unit that outputs a product having a higher product of an appearance score and an appearance frequency as preference information indicating the user's preference. 4. The bookmark system according to claim 1, wherein the bookmark has a proposal document folder, and the new document proposal section stores a URL of a document to be proposed to a user in the proposal document folder. A bookmark system comprising: 5. The bookmark system according to claim 1, wherein the user preference extracting unit selects one of the categories based on documents of a plurality of categories indicated by bookmark information registered in the bookmark. The new document suggestion unit matches the user's preference based on the preference category's preference information extracted by the user preference extraction unit. A bookmark system having a configuration in which a document in the category of interest is searched for, and bookmark information indicating the found document is registered in the bookmark. 6. The bookmark system according to claim 5, wherein the bookmark includes a category folder for each category and a category folder for each category folder contained in the category folder.
A URL, and a suggested document folder provided under each of the classified folders, wherein the user preference extracting unit selects a category of interest in each of the categories based on a document pointed to by the URL in each of the classified folders. The new document suggestion unit searches for a document that matches the user's preference based on the preference information of the attention category extracted by the user preference extraction unit. A bookmark system for storing a URL pointing to the found document in a proposal document folder under a classification folder corresponding to the noted category. 7. The bookmark system according to claim 5, wherein the user preference extracting unit collects documents in a plurality of categories indicated by the bookmark information registered in the bookmark. For each document collected by the document collection unit, an important keyword extraction unit for extracting, as an important keyword in the document, a word having a higher appearance score determined by an appearance frequency and an appearance position among words in the document. And calculating the positive score of each important keyword by adding the appearance scores of the important keywords of each document in one attention category among the plurality of categories to each other, thereby obtaining a positive score for each important keyword. Weight of each document in By adding the appearance scores of the keywords to the same important keywords, a negative score for each important keyword is obtained, and a characteristic keyword of the attention category is obtained based on the positive score and the negative score for each important keyword. A bookmark system, comprising: a feature keyword analysis unit that outputs a keyword as preference information indicating the preference of the user in the category of interest. 8. The bookmark system according to claim 2, 3 or 7, wherein the important keyword extracting unit converts an important keyword of the document collected by the document collecting unit into a hyperlink within a certain range from the document. A bookmark system configured to perform extraction based on an appearance score determined by an appearance frequency and an appearance position of each word in a document set including a plurality of document groups associated with each other. 9. The bookmark system according to claim 1, wherein the preference information extracted by the user preference extraction unit includes a plurality of keywords, and the new document suggestion unit determines a search result in advance. Until the number of items becomes equal to or less than one, the user selects one unselected keyword from among the plurality of keywords constituting the preference information, adds the selected keyword to the search formula, and searches the search engine on the Internet. A bookmark suggestion unit for registering, in the bookmark, bookmark information indicating a searched document when a search result is equal to or less than the predetermined number of times. . 10. The bookmark system according to claim 1, wherein the preference information extracted by the user preference extraction unit includes a plurality of keywords, and the new document proposal unit is prepared in advance. In each of the documents, a document to be proposed to the user is determined from the plurality of documents based on the sum of the appearance scores of the respective keywords constituting the preference information, and the bookmark information indicating the determined document is stored in the document. A bookmark system comprising a document filtering proposal unit for registering a bookmark. 11. The bookmark system according to claim 1, wherein the bookmark information registered by the user is registered in the bookmark as data having no restriction on the editing operation by the user, and the built-in data and advertisement registered by the system are registered. , The proposed document, and the proposed document folder are registered in the bookmark as data having a predetermined restriction on the editing operation by the user for each type. A bookmark system comprising a bookmark operation unit that performs only an authorized editing operation. 12. The bookmark system according to claim 11, wherein the bookmark includes a category folder for each category, a URL contained in each category folder, a proposed document folder under each category folder, and A URL contained in each proposed document folder, and information indicating a hierarchical structure of the folder, the bookmark operation unit, when there is a browsing request from the user, based on the information in the bookmark, to the bookmark A bookmark system comprising: a display unit that displays a hierarchical structure of existing folders on a display unit, and a document pointed to by a URL contained in a folder selected by a user on the display unit. 13. The bookmark system according to claim 12, wherein, when a delete operation is performed on bookmark information for which a delete operation is not permitted at the time of editing, the bookmark operation unit performs the delete operation. The information is displayed on the display unit in a display format indicating that it is non-display data, and information indicating that the bookmark information on which the deletion operation has been performed is non-display data is stored. A bookmark system, wherein display data is not displayed on the display unit. 14. A bookmark system comprising a bookmark server and a plurality of client devices interconnected with the bookmark server, wherein the bookmark server is a user in which bookmark information of each of the client devices is registered. A bookmark DB, a bookmark operation unit that operates the bookmark information corresponding to the requesting client device registered in the user bookmark DB according to the request from each client device, and is registered in the user bookmark DB. Based on the document indicated by the bookmark information, a user preference extraction unit that extracts preference information representing the preference of each user of each of the client devices, based on the preference information of each user extracted by the user preference extraction unit, Each of the above An operation request for requesting the bookmark server to operate bookmark information corresponding to its own client device with respect to the bookmark server. A bookmark system comprising a unit. 15. A bookmark system comprising a plurality of client devices and a recommendation server interconnected with each of the client devices, wherein each of the client devices has registered therein bookmark information of its own client device. A user bookmark DB, and a bookmark operation unit that operates bookmark information registered in the user bookmark DB in accordance with a request of the user of the own client device, wherein the recommendation server stores the user bookmark DB in the user bookmark DB of each of the client devices. A user preference extraction unit for extracting preference information of each user of the client device based on a document indicated by the registered bookmark information; and a user preference information extracted by the user preference information extracted by the user preference extraction unit. , Each of the above users And a new document proposal section for proposing a document that meets a user's preference. 16. A bookmark system comprising a plurality of client devices and a recommendation server interconnected with each of the client devices, wherein each of the client devices has registered therein bookmark information of its own client device. A user bookmark DB, a bookmark operation unit that operates the bookmark information registered in the user bookmark DB according to a request to the user of the own client device, and a document indicated by the bookmark information registered in the user bookmark DB. A user preference extraction unit that extracts the preference information of the user of the own client device based on the preference information extracted by the user preference extraction unit of the client device based on the preference information of the client device. According to user preference Bookmark system, characterized in that it was equipped with a new document proposing unit to propose the regulations document. 17. A user preference extracting step of extracting preference information indicating the user's preference based on a document indicated by bookmark information registered in a bookmark, and a user preference information extracted in the user preference extraction step. A new document suggestion step of searching for a document that matches the user's preference and registering bookmark information indicating the found document in the bookmark. 18. The document proposal method using a bookmark according to claim 17, wherein the user preference extracting step includes: a document collection step of collecting a document indicated by the bookmark information; and a document collection step of collecting the document indicated by the bookmark information. An important keyword that extracts, as important keywords, those having a higher appearance score determined by the appearance frequency and appearance position of each word in the document, and outputs the extracted important keywords as preference information representing the user's preference. A document proposal method using a bookmark, comprising: an extraction step. 19. The document proposal method using a bookmark according to claim 17, wherein the user preference extracting step includes: a document collection step of collecting a document indicated by the bookmark information; and a document collection step of collecting the document indicated by the bookmark information. An important keyword extraction step of extracting, as important keywords, an appearance score determined by the appearance frequency and appearance position of each word in the document as an important keyword; and, of the important keywords extracted in the important keyword extraction step, A feature keyword analyzing step of outputting a product having a higher product of the appearance score and the appearance frequency of the important keyword as preference information indicating the user's preference. 20. The method according to claim 17, wherein the bookmark has a suggested document folder, and the new document suggesting step includes a step of setting a URL of a document to be proposed to a user as the suggested document. A document proposal method using bookmarks, which is stored in a folder. 21. The document proposal method using a bookmark according to claim 17, wherein, in the user preference extracting step, based on a plurality of categories of documents indicated by bookmark information registered in the bookmark. Extracting preference information indicating the user's preference for the attention category in each category; and matching the user's preference based on the preference information of the attention category extracted by the user preference extraction step in the new document proposal step. A document proposal method using a bookmark, wherein a document in the category of interest is searched for, and bookmark information indicating the found document is registered in the bookmark. 22. The document proposal method using bookmarks according to claim 21, wherein the bookmarks are classified into each of the classified folders for each category and each of the classified folders contained in each of the classified folders.
A URL and a proposed document folder provided under each of the classification folders;
Based on the document pointed to by L, extract preference information indicating a user's preference for the attention category in each of the categories, and in the new document proposal step, the preference of the attention category extracted in the user preference extraction step. A document using a bookmark, wherein a document matching the user's preference is searched based on the information, and a URL pointing to the searched document is stored in a proposed document folder under a classification folder corresponding to the noted category. Proposed method. 23. The document proposal method using a bookmark according to claim 21, wherein the user preference extracting step collects documents of a plurality of categories indicated by the bookmark information registered in the bookmark. A collection step, and for each document collected in the document collection step, among words in the document, a word having a higher appearance score determined by an appearance frequency and an appearance position is extracted as an important keyword in the document. An important keyword extraction step of: adding the same important keywords to the appearance scores of the important keywords of each document in one category of interest among the plurality of categories to obtain a positive score for each important keyword; By adding the appearance scores of the important keywords of each document in the category other than the attention category to the same important keywords, a negative score for each important keyword is obtained, and based on the positive score and the negative score for each important keyword, A characteristic keyword analyzing step of obtaining a characteristic keyword of a category and outputting the obtained characteristic keyword as preference information indicating the user's preference in the noted category. Method. 24. The document proposal method using a bookmark according to claim 17, wherein the bookmark information registered by the user is registered in the bookmark as data having no restriction on the editing operation by the user, and the bookmark information is registered by the system. The bookmark information of each type of built-in data, advertisement, proposed document, and proposed document folder is registered in the bookmark as data having a predetermined restriction on the editing operation by the user for each type, and becomes a target of the editing operation. A document proposal method using a bookmark, comprising a bookmark operation step of performing only an editing operation permitted for bookmark information. 25. A user preference extracting step for extracting preference information representing the user's preference based on a document indicated by bookmark information registered in the bookmark, and a user preference extracted in the user preference extraction step. A program recording medium for recording a program for performing a new document proposal step of searching for a document that matches the user's preference based on information, and registering bookmark information indicating the searched document in the bookmark. 26. The program recording medium according to claim 25, wherein: in the computer, in the user preference extracting step, a document collection step of collecting a document indicated by the bookmark information; Of the words in the collected documents, those having a higher appearance score determined by their appearance frequency and appearance position are extracted as important keywords, and the extracted important keywords are output as preference information representing the user's preferences. A program recording medium on which a program for performing an important keyword extracting step is recorded. 27. The program recording medium according to claim 25, wherein in the user preference extracting step, the computer collects a document indicated by the bookmark information; An important keyword extraction step of extracting, as an important keyword, an appearance score determined by the appearance frequency and appearance position of each word in the collected documents as an important keyword; and an important keyword extracted in the important keyword extraction step. A characteristic keyword analyzing step of outputting, as preference information indicating the preference of the user, a product having a higher product of the appearance score and the appearance frequency of the important keyword. 28. The program recording medium according to claim 25, wherein in the user preference extracting step, the user preference is set based on a plurality of categories of documents indicated by bookmark information registered in the bookmark. In the new document proposal step, based on the preference information of the attention category extracted by the user preference extraction step, the user is made to perform a process of extracting preference information indicating a user's preference for the attention category in each category. A program recording medium for recording a program for performing a process of searching for a document in the noted category that matches the preference of the user and registering the bookmark information indicating the searched document in the bookmark. 29. The program recording medium according to claim 25, wherein the computer registers the bookmark information registered by the user in the bookmark as data having no restriction on the editing operation by the user. The bookmark information of each type of built-in data, advertisement, proposed document, and proposed document folder is registered in the bookmark as data having a predetermined restriction on the editing operation by the user for each type, and becomes a target of the editing operation. A program recording medium in which a program for performing a bookmark operation step of performing only an editing operation permitted for bookmark information is recorded.