JP2002342364A - Method and system for managing document information - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically collect the information of newly arrived documents of contents required by users to be able to register the information on a document information database and further, to automatically report the information registered on the document information database of each of users to other users collecting similar document information as well for mutually sharing the information. SOLUTION: A profile 120 of document required for every user is made to be registered, and the information of newly arrived documents of high similarity with the profile of a user is collected on an information network 300 for every, registered on the document information database for each of users and shared among the users. Besides, an evaluation value is registered in respect to newly arrived papers acquired by each of users, and concerning newly arrived document information which is not evaluated by a certain user, the evaluation of the relevant user is predicted by collaborative filtering processing. Concerning the newly arrived document information having great evaluation prediction, the newly arrived document information is resported to other corresponding users as well.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document information management method and system.

[0002]

2. Description of the Related Art Conventionally, as a system for collecting the latest document information, a document title or an author name is used as a search key, and a relevant document is extracted from newly issued document information and notified to a user. An alert type service is provided.

As a document information management system, E
ndNote (product name) is known. This software is for organizing the article information that each user wants to refer to, and can organize and manage the database of the document information obtained independently for each user.

[0004]

However, in the conventional document information collection system, since the search key is limited to the title of the paper or document, the author's name, etc., a large amount of unnecessary information is distributed to the user. There is a point. Further, in the conventional document information management system, it is necessary for the user to input data such as a natural sentence to be individually searched for, so that there is a problem that it is troublesome, and the document desired by the user is not necessarily extracted. is there.

In addition, in the conventional document information collection system, it takes time to input the distributed document information to the existing document information management system because no cooperative relationship has been established between the two systems. However, there is a problem in that sharing of document information is difficult because information on who holds what kind of document information is not mutually distributed between different users.

The present invention has been made in view of such a conventional technical problem, and a plurality of users automatically collects new arrival document information of contents required by the user on an information network. It is an object of the present invention to provide a document information management technology that can be registered in its own document information database and can share information registered in each user's document information database with each other.

According to the present invention, a plurality of users can automatically collect new document information of contents required by themselves on an information network and register them in their own document information database. To provide a document information management technology that can automatically notify other users who are collecting similar document information of the registration of a newly arrived document and share the information with each other who has collected similar document information. With the goal.

[0008]

According to the first aspect of the present invention, new documents necessary for each of a plurality of users are collected from a large number of documents that can be downloaded through an information network and registered in a document information database of each user. A document information management method for sharing a document information database for each user between users, wherein a profile of a required document is registered for each of a plurality of users, and a similarity between the profile and each newly arrived document on the information network is registered. Calculate the degree, collect for each user the newly arrived documents with high similarity with the profile of the user, register the collected newly arrived documents in the literature information database for each user,
The document information database for each user is shared between users.

In the document information management method according to the first aspect of the present invention,
By registering the profile of the required document for each of a plurality of users, the similarity between the profile and each newly arrived document on the information network is calculated, and for each user, the similarity with the profile of the user is high. Newly collected documents are collected, the collected new documents are registered in a document information database for each user, and the document information database for each user is shared between users. For this reason, the newly arrived document searched by the profile individually registered by each user can be shared and used by a plurality of users.

According to a second aspect of the present invention, a newly arrived document necessary for each of a plurality of users is collected from a large number of documents that can be downloaded through an information network, registered in each user's document information database, and a document information database for each user is collected. Is a document information management method for sharing information among users, wherein a profile of a required document is registered for each of a plurality of users, and a similarity search between the profile and each newly arrived document on the information network is performed by a predetermined search formula. And collects newly arrived documents having a high degree of similarity with the profile of the user for each user, and, for the collected newly arrived documents, feeds the appropriateness evaluation value of the corresponding user back to the search formula, and Correct the search formula based on the information, update the profile, and collect the new sentence Was registered in the document information database for each user, it is characterized in that the covalent literature information database for each of the user among the users.

In the document information management method according to the second aspect of the present invention,
By registering the profile of the required document for each of a plurality of users, a similar search between those profiles and each newly arrived document on the information network is performed using a predetermined search formula. Newly collected documents having a high degree of similarity to are collected, registered in a document information database for each user, and shared among users. In addition, for the collected newly arrived documents, the appropriateness evaluation value of the user is fed back to the search expression of the similarity calculation, and the search expression is corrected based on the feedback information and the profile is updated, so that each time the search is repeated, Newly arrived documents according to the user's request are collected, registered in a document information database for each user, and shared between users. For this reason, a newly arrived document can be searched by a profile that more accurately expresses the request of each user, registered in the document information database of each user, and shared by a plurality of users.

According to a third aspect of the present invention, in the document information management method according to the first or second aspect, an evaluation value is registered for a newly arrived article obtained by each of the users, and a certain user is evaluated by a collaborative filtering process. It is characterized in that the evaluation prediction of the user is performed on the document information, and for the newly arrived document information having a large evaluation prediction, the relevant user is notified of the newly arrived document information. Users can automatically notify each other of the registration information of newly arrived documents registered in the document information database of other users, thereby avoiding the omission of collection of document information required by each user as much as possible. it can.

The invention according to claim 4 comprises a document information collection system for collecting necessary documents from a large number of documents that can be downloaded through an information network, and a document information sharing system for sharing a document information database of a plurality of users. A document information management system, wherein the document information collection system includes: a profile registration unit for registering a profile of a required document for each of a plurality of users; and a profile registration unit for registering the profile and each newly arrived document on the information network. A similarity search unit that calculates similarity and collects newly arrived documents having a high degree of similarity with the profile of the user for each user, and registers the newly arrived documents collected by the similarity search unit in the document information database of the user. And registration means for performing the registration.

[0014] In the document information management system according to the fourth aspect of the present invention, when the profile of the required document is registered for each of a plurality of users by the profile registration unit of the document information collection system, the similarity search unit stores the profile and the information. The degree of similarity with each of the newly arrived documents on the network is calculated, and newly arrived documents having a high degree of similarity to the profile of the user are collected for each user. The document information is registered in the information database, and the document information sharing system causes the document information in the document information database for each user to be shared among a plurality of users.

Therefore, each user can share and use the newly arrived document information corresponding to the profile of another user.

According to a fifth aspect of the present invention, there is provided a document information collecting system for collecting necessary documents from a large number of documents that can be downloaded through an information network, and a document information sharing system for sharing a document information database of a plurality of users. A document information management system, wherein the document information collection system includes: a profile registration unit for registering a profile of a required document for each of a plurality of users; and a profile registration unit for registering the profile and each newly arrived document on the information network. A similarity search unit that calculates a similarity and collects, for each user, newly arrived documents having a high degree of similarity to the profile of the user, and a new arrival document having a high degree of similarity collected by the similarity search unit. Feedback for feeding the user's suitability evaluation value back to the similarity search means Means, a search formula correcting means for performing a search formula correction of the similarity search means based on feedback information of the feedback means, and updating a profile of the profile registration means with a correction result, and newly arrived documents collected by the similar search means. And a registering means for registering the information in the document information database of the user.

In the document information sharing system according to the fifth aspect of the present invention, the necessary document profile is registered for each of a plurality of users by the profile registration means of the document information collection system, so that the similarity search means can be used for each user. Calculate the degree of similarity between the profile of each of the newly arrived documents on the information network and each user, collect newly arrived documents having a high degree of similarity with the profile of the user for each user, and register them in the document information database of each user by the registering means The document information in the document information database for each user is shared among a plurality of users.

The feedback means feeds back to the similar search means the relevance / evaluation value evaluated by the user with respect to newly arrived documents having a high degree of similarity collected by the similar search means, and the search formula correcting means feeds back the feedback information of the feedback means. Then, the search formula of the similarity search unit is corrected based on the search result, and the profile of each user of the profile registration unit is updated with the correction result.

Thus, the profile can be updated for each user, and as the number of times of use increases, new document information more suited to the needs of each user can be collected and shared among users.

According to a sixth aspect of the present invention, in the document information management system of the fourth or fifth aspect, the document information sharing system cooperates with a user interface for registering evaluation values for newly arrived articles obtained by the users. An evaluation prediction unit configured to perform evaluation prediction of the user with respect to new document information that has not been evaluated by a user by a filtering process; and a unit that notifies the user of the newly arrived document information with respect to the newly arrived document information having a large evaluation prediction. It is characterized by having.

In the document information management system according to the sixth aspect of the present invention, if an evaluation value is registered for a newly arrived article obtained by each user using the user interface of the document information sharing system, the evaluation prediction means can execute the collaborative filtering process. A certain user makes an evaluation prediction of the new arrival document information that has not been evaluated, and notifies the corresponding user of the newly arrived document information of the newly arrived document information whose evaluation prediction is large.

Therefore, if the document information collection system collects new document information for each user according to the profile and registers the new document information in their own document information database, other users who need similar new document information can be used. Notification that similar new arrival document information has been registered in another user's document information database, and the shared document information can be shared, and the newly arrived document information collected by each user can be effectively used among many users. .

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 shows a document information management system according to one embodiment of the present invention. The document information management system according to the present embodiment performs automatic collection and management of dissertation information, and includes a dissertation information distribution system 100 and a dissertation information sharing platform 200.

The dissertation information distribution system 100 receives new dissertation information downloaded from a pre-registered reference information distribution service site via the Internet 300, or accesses a specific dissertation distribution site by itself via the Internet 300 and registers there. A similarity search module 110 that downloads the newly arrived article information, determines whether the article needs to be collected by the similarity search process, and performs filtering to obtain the necessary article if necessary. A profile registration unit 120 in which a profile used for a search formula is registered is provided.

The dissertation information sharing platform 200
User PC 21 connected by AN and used by each user
., And these user PCs 210
-1, 210-2,..., The management server 22 that manages the connection
0 is provided.

Then, the newly-arrived paper extracted by the similarity search module 110 is presented to the user concerned, and the user is made to input a propriety evaluation for proper / unsuitable feedback.
The system includes a user interface 211 for inputting an instruction to discard newly arrived papers, and a paper database 212 for registering paper information for each user.

Next, a description will be given of a document information management method using the above-structured document information management system.

The profile registration unit 120 registers a profile composed of document data for each user.
The profile registration unit 120 registers one or a plurality of pieces of document information (text data) of a paper in an academic field similar to the content of a paper required most by each user, or expresses the content of the paper required by the user himself. When document information (text data) is input, word analysis and vectorization are performed and registered as profile data.

Then, as shown in the flowchart of FIG. 2, the similarity search module 110
For the new article information document data downloaded from 0, the similarity to the profile for each user registered in the profile registration unit 120 is calculated, and for new article information having a similarity higher than a predetermined threshold, Inform the corresponding user PC among the PCs 210-1, 210-2,... That the new paper has arrived (step S1).
1). Note that this similarity calculation is based on the conventional technique of Japanese Patent Application No. 2000-69477, which is the invention of the present inventors, as shown in FIG.
The calculation method described using is used. The newly arrived paper information describes items as shown in FIG. Article ID, Exhibit (magazine name), Issue number, Author, Title, Pointer to the article text (this may be a URL on the Internet or the article full-text storage area of the large-capacity document storage device on the system 100 side. May be)
And it is an abstract (text data).

The user who has arrived at the newly arrived paper displays the newly arrived paper information using the user interface 211 or prints out the outline to confirm the outline. By accessing the storage location of the text, the data of the text of the paper is downloaded to its own database 212 and read. If the paper has the contents required by the user, feedback is given that the paper is suitable, and if not, feedback is provided that the paper is not suitable (steps S12 to S17).

The similarity search module 110, upon receiving feedback from the user on the propriety of the user, performs a process of expanding a search formula for similarity search, and the profile registration unit 12.
0 is updated (step S1).
8).

This search formula expansion process employs, for example, the search formula processing method based on the word contribution described in the prior art of Japanese Patent Application No. 2000-69477. If it is necessary to increase the filtering accuracy, a non-conforming profile is registered in the profile registration unit 120 together with the conforming profile described in the embodiment of the invention of the same application and FIGS. 3, 4, and 6 for each user. If the newly arrived paper has a high degree of similarity with a certain user's matching profile, the similarity with the non-matching profile is also calculated. By not determining, there is adopted a filtering method in which only papers having a high similarity to the conforming profile and not having a high similarity to the non-conforming profile are newly arrived papers requested by the user. in this case,
The update of the profile is performed according to the method shown in the flowcharts of FIGS.

The paper information sharing platform 200 connects the user PCs 210-1, 210-2,.
N. Management server 220 in this dissertation information sharing platform 200
Manages data as shown in FIG. 4 for each user for all the article information stored in these user PCs 210-1, 210-2,. This is an article ID, a flag indicating whether the article has been browsed, and an evaluation value (that is, a value obtained by, for example, evaluating the article whether or not it is a content required by the user in five steps from 1 to 5). . The management data format is, for example, as shown in FIG. In FIG. 5, the row indicates the document ID, and the column indicates the registered user index. For example, the user a browses the document d1 and indicates that the evaluation value 1 is set. Further, it indicates that the users a and b do not browse the document d3, and thus the evaluation value is none (-).

As described above, each user holds the article information in the article database 212 of the user PC, but when a new article information is notified to a certain user, it is necessary for other users. The server 200 determines whether or not the paper is high, and distributes the same newly arrived paper information as recommended paper information to the user who is estimated to be highly necessary. This is based on the processing in the flowchart of FIG.

As shown in the table of FIG. 5, article information of article IDs d1 to dn is registered in the management server 220. The user a evaluates the article information related to the article viewed by the user a. It is assumed that a value has been registered (step S21).

The management server 220 performs an evaluation prediction on a paper di not evaluated by the user a among the papers d1 to dn, which has not been evaluated, using a collaborative filtering method (steps S22 to S24). If the predicted evaluation value eai is a predetermined threshold value (for example, eai> 3 in the case of a five-level evaluation), recommendation information is distributed as a non-evaluated paper necessary for the user a (step S1). S26). Then, the processing of steps S22 to S26 is repeated for all newly arrived papers (step S28). The information of the recommended paper is performed in the same format as the paper data of FIG.

In the collaborative filtering method, focusing on users a and d in the table of FIG. 5, in principle, they both evaluate papers d2 and d6 with high evaluation, and paper d1
It is inferred that interest is similar because of only low evaluations of. Therefore, the evaluation given by the user d to the unrated paper d4 is the same as that of the user a, and it is assumed that the evaluation will be 3 or 4. More specifically, in predicting an evaluation value of a specific user using collaborative filtering, a correlation coefficient between users is obtained, and an evaluation value for an unrated paper is predicted using the correlation coefficient.

If the user who has received the notification of the new paper information is interested in the new paper accepted by another user, the user accesses the database of the paper storage destination connected via the network and stores the corresponding new paper data. It can be browsed and printed out if necessary, or downloaded and stored in the article database of its own PC.

As described above, according to the document information management system of the present embodiment, the profile registration unit 120 of the dissertation information distribution system 100 allows each of a large number of users to register a profile for collecting new dissertations having their own contents of interest.
, A new article having a content similar to the profile can be acquired from a large number of articles published on the Internet 300 and stored in the article database 212. In addition, the contents of the acquired newly-arrived papers are examined, and if the papers are required by the user, conformance information is displayed.
By feeding back from 1 to the dissertation information distribution system 100, the profile of each user in the profile registration unit 120 is updated to more appropriate content as the use is continued, and highly accurate document filtering becomes possible. .

In addition to sharing the article information in the article database 212, which is collected and held by a large number of users, with other users, when a new article is accepted by a certain user, the contents of the same tendency are displayed. Registration information of new articles is automatically distributed by selecting other users who are interested in dissertation information, so many scholars, especially universities and large research institutions, collect many dissertations individually. Institutions doing research can reduce the omission of important papers, and also prevent the waste of obtaining papers that are not necessary for some researchers.

In order to share a large number of papers, the information of similar users who collect papers of similar contents is communicated among users, so that papers newly obtained by a unique method can be used by other users. It is effective to use it. Thus, the dissertation information sharing platform 200 calculates the similarity between the document information of the registered dissertations in the dissertation database 212 of each user, determines a user having a high similarity as a similar user, and notifies each user.

That is, as shown in the flowchart of FIG. 7, when a similar user is searched for a certain user a, the user a and another user group U = {u1, u2,.
The similarity between the document information of the group of papers registered in the paper database 212 is calculated with respect to um @ (steps S31 to S33).

If the similarity is larger than the predetermined threshold, the user is determined to be a similar user, and the user is notified to the user a as a similar user to the user a (steps S34, S35).

Thereafter, the processing of steps S33 to S35 is repeated for all the users of the user group U, and the user a
Then, all the similar users to are extracted and notified to the user a (steps S36 and S37).

As a result, it is possible for the users of the document information management system to know which other users are collecting articles similar to themselves, and for each user to obtain information on articles independently collected by similar users. , And the omission of collection of paper information required by each user can be avoided.

Although the above embodiment has been described with reference to a document information management system that collects dissertation information, it can be used in any technical field for collecting document information.
For example, it can be widely used for managing new book information and managing new news.

The present invention also includes a computer program for realizing the system functions of the above-described embodiments, and a recording medium on which the computer program is recorded.

[0048]

As described above, according to the document information management method of the first aspect of the present invention, by registering the required document profiles for each of a plurality of users, those profiles and the information on the information network are registered. Calculates the degree of similarity with each newly arrived document, collects newly arrived documents having a high degree of similarity with the profile of the user for each user, registers the collected newly arrived documents in a document information database for each user, and retrieves document information for each user. Since the database is shared among the users, the newly arrived documents searched by the profiles individually registered by the users can be shared and used by a plurality of users.

According to the document information management method of the second aspect of the present invention, by registering the profile of the required document for each of a plurality of users, similarity search between the profile and each newly arrived document on the information network is performed. Is performed using a predetermined search formula, new documents having a high similarity to the profile of the user are collected for each user, registered in a document information database for each user, shared between users, and With respect to the document, the appropriateness evaluation value of the user is fed back to the search expression of the similarity calculation, and the search expression is modified and the profile is updated based on the feedback information. New documents can be collected, registered in the document information database for each user, and shared between users. , And registered in the user individual document information database and search for new documents by profile more accurately represent the requirements of each user can be utilized by shared among multiple users.

According to the document information management method of the third aspect of the present invention, evaluation values are registered for newly arrived papers acquired by each user, and the collaborative filtering process allows a certain user to obtain information on the newly arrived literature that has not been evaluated. Evaluation prediction is performed, and for newly arrived document information having a large evaluation prediction, the newly arrived document information is notified to the corresponding user. Therefore, among users having similar contents of required document information, the document information database of another user is automatically generated. The registered information of the newly arrived documents registered in each of the users can be notified to each other, and omission of collection of document information required by each user can be avoided as much as possible.

According to the document information management system of the present invention, each user can share and use newly arrived document information corresponding to the profile of another user.

According to the document information management system of the present invention, it is possible to update the profile for each user, and collect new document information according to each user's needs as the number of times of use is increased. Can be shared and used.

According to the document information management system of the present invention, similar information can be obtained if the document information collection system collects newly arrived document information for each user in accordance with the profile and registers it in their own document information database. It notifies other users who need new arrival document information that similar new arrival document information has been registered in another user's document information database, and allows them to share and use the new arrival document information collected by each user. Can be effectively used among many users.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a functional configuration of a dissertation information management system according to one embodiment of the present invention.

FIG. 2 is a flowchart of a process executed by the dissertation information distribution processing system in the embodiment.

FIG. 3 is a data structure diagram of new article information in the embodiment.

FIG. 4 is an explanatory diagram of dissertation evaluation data in the embodiment.

FIG. 5 is an explanatory diagram showing a user-specific paper evaluation table in the embodiment.

FIG. 6 is a flowchart of new article information recommendation processing using a collaborative filtering method executed by the article information sharing platform in the embodiment.

FIG. 7 is a flowchart of a similar user search / notification process executed by the dissertation information sharing platform in the embodiment.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 dissertation information distribution system 110 similarity search module 120 profile registration unit 200 logical information sharing platform 210-1, 210-2, user PC 211 user interface 212 dissertation database 220 management server 300 internet

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Keiichiro Hoashi 2-1-1-15 Ohara, Kamifukuoka City, Saitama Prefecture KDDI Research Institute, Inc. (72) Kazuo Hashimoto 2-1, Ohara, Kamifukuoka City, Saitama Prefecture No. 15 Inside KDDI Research Institute (72) Inventor Atsushi Ito 2-3-2 Nishishinjuku, Shinjuku-ku, Tokyo Inside KDD Corporation (72) Inventor Yasuya Harada Totsuka, Shinjuku-ku, Tokyo 104-cho, Waseda University (72) The inventor Noriaki Kusumoto 1-104, Totsukacho, Shinjuku-ku, Tokyo No. 1 Inside K.com (72) Inventor Takuya Sato 4-3-1 Yotsuya 4-chome, Shinjuku-ku, Tokyo Co., Ltd. Cast in the F-term (reference) 5B075 KK02 ND20 PP26 PQ05 PR06 PR08 UU06 UU08

Claims (6)

[Claims] 1. Collecting new documents required for each of a plurality of users from a large number of documents that can be downloaded through an information network, registering them in each user's document information database, and sharing a user-specific document information database among users. A document information management method for registering a required document profile for each of a plurality of users, calculating a similarity between the profile and each newly arrived document on the information network, and Collecting newly arrived documents having a high degree of similarity to the profile, registering the newly arrived documents in a document information database for each user, and sharing the document information database for each user among users. Management method. 2. Collecting new documents necessary for each of a plurality of users from a large number of documents that can be downloaded through an information network, registering them in each user's document information database, and sharing a user-specific document information database among users. A document information management method for registering a required document profile for each of a plurality of users, performing a similarity search between the profile and each newly arrived document on the information network using a predetermined search formula, For each of the users, new documents having a high degree of similarity to the profile of the user are collected, and for the collected new documents, the appropriateness evaluation value of the user is fed back to the search formula, and the search is performed based on the feedback information. Correct the formula and update the profile, and collect the newly arrived documents for each user. Document information management method, characterized in that registered in the document information database, to share a document information database for each of the user among the users. 3. The user is caused to register an evaluation value for a newly arrived article obtained by each of the users, and a collaborative filtering process is performed to estimate the evaluation of the user with respect to the newly-arrived document information that has not been evaluated by a certain user. 3. The document information management method according to claim 1, wherein the new user information is notified to a corresponding user. 4. A document information management system configured by a document information collection system that collects necessary documents from a large number of documents that can be downloaded via an information network, and a document information sharing system that shares a document information database of each of a plurality of users. A document registration system for registering a profile of a required document for each of a plurality of users; and calculating a similarity between the profile and each newly arrived document on the information network. A similarity search unit that collects, for each user, newly arrived documents having a high degree of similarity with the profile of the user; and a registration unit that registers the newly arrived documents collected by the similarity search unit in the document information database of the user. Provided document information management system. 5. A document information management system comprising a document information collection system for collecting necessary documents from a large number of documents that can be downloaded through an information network, and a document information sharing system for sharing a document information database of each of a plurality of users. A document registration system for registering a profile of a required document for each of a plurality of users; and calculating a similarity between the profile and each newly arrived document on the information network. A similarity search unit that collects, for each user, newly arrived documents having a high degree of similarity to the profile of the user; and a propriety evaluation value of the user with respect to the newly arrived documents having a high degree of similarity collected by the similarity search unit. Feedback means for feeding back to the similarity search means; A search formula correction unit that corrects a search formula of the similarity search unit based on feedback information of the feedback unit, and updates a profile of the profile registration unit with a correction result; A document information management system comprising: a registration unit for registering in the document information database. 6. The document information sharing system includes: a user interface for registering an evaluation value for a newly-arrived article obtained by each of the users; and a user-evaluation of the newly-arrived document information for which a user has not been evaluated by a collaborative filtering process. 6. The document information management system according to claim 4, further comprising: an evaluation prediction unit for performing prediction; and a unit for notifying a user of the newly arrived document information with respect to the newly arrived document information having a large evaluation prediction. system.