JP2007179490A - Information resource retrieval device, information resource retrieval method and information resource retrieval program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To regularly obtain a retrieval result without impairing retrieval efficiency even if metadata assigned to an information resource usable on a Web is imperfect or a retrieval condition is improper. <P>SOLUTION: The device comprises a first score calculation part 21 calculating a first score for information resources with metadata containing a free word matched with an input free word; a second score calculation part 22 calculating a second score for information resources with metadata containing a thesaurus keyword matched with an input thesaurus keyword, and multiplying, for information resources with metadata containing the thesaurus keyword belonging to another node in the same facet, the second score by a weight according to a distance on a tree structure for the information resources with metadata containing a thesaurus keyword belonging to another node in the same facet, and taking the result as the second score; a score addition part 23 obtaining a third score by adding the first and second scores; and a sorting part 24 obtaining a list of information resources sorted in the descending order of the third score as a retrieval result to be outputted. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an information resource search device, an information resource search method, and an information resource search program. More specifically, bibliographic information such as author, title, date of creation, etc. is given as metadata, and in search of information resources and documents that can be browsed via the web browser on the Internet, the hit miss in the search is resolved, Accordingly, the present invention relates to a search technique for improving the availability of information resources and a user interface for inputting search conditions and browsing search results.

  For example, documents such as academic papers that have been converted to HTML or PDF, or text data, and digital contents such as photographs, maps, and rare books can be viewed on the Internet via a web browser. is there. The amount of these information resources is already enormous, and there is a strong demand for the creation of a database, its smooth distribution, and reuse.

  In order to use another person's information resources, it is necessary to use a search engine to arrive at the desired information resource. However, conventional search engines are exclusively targeted for metadata given as bibliographic information to information resources. Then, an information resource including text that matches the input search text is extracted as a search result.

As a metadata description method, for example, the Dublin Core method is standardized (ISO15836). In this Dublin Core system, title (resource title), creator (creator), subject (topics and keywords included in resource contents), description (description of resource contents, summary, table of contents, etc.), date (creation) Date (date or release date), type (the nature or genre of the content of the resource), coverage (range or target of the resource, geographical division, time division, etc.), rights (statement of copyright, industrial property rights, etc.), etc. Are defined as metadata elements to be described. The information resource provider or librarian describes the metadata as, for example, a tag specified by RDF (Resource Description Framework), HTML, etc. according to this description method, either manually or partially automatically, and this metadata is described. Give to information resources and publish on the web.
JP 2000-1212949 A JP 2004-310199 A

  However, the metadata description work to make others use their information resources is cumbersome, and the description of metadata given to existing information resources may lack elements to be described. Or incomplete in terms of missing keywords.

  This is particularly noticeable when a user unfamiliar with information resource search tries to reach the information resource via the search engine, but the input search character string is inappropriate and is described in the meta data. If it does not match the data keyword, no information resource is obtained as a search result (Exact Match type search). Conversely, when a large number of information resources that hit the input search character string are output, it is impossible to determine which is the desired search result, so the obtained search results are scrolled and viewed one by one. And the search efficiency is poor and a lot of labor is required.

  For example, in particular, academic information resources often have keywords defined corresponding to nodes on facets, which are hierarchical directory structures based on thesaurus classification, but are not well-versed. The user does not know the facet system based on the thesaurus classification and cannot input it as an appropriate search keyword.

  Further, in the conventional search engine, when a search condition is input a plurality of times, only those that match the AND condition of the plurality of search conditions are extracted, so that an inappropriate search keyword is input even once. The desired information resource is removed from the search result list, and the opportunity to reach this information resource is missed.

  The present invention has been made in view of the above problems, and its purpose is to provide users who are incomplete or unfamiliar with the bibliographic information given to information resources available on the web. Even if an inappropriate search keyword is entered by, information resource search that can always obtain search results without sacrificing search efficiency and that can be obtained in order from search results that match the search conditions. An apparatus, an information resource search method, and an information resource search program are provided.

  Another object of the present invention is to provide a list of information resources that are relevant to an input search keyword even if the input search keyword does not match the description on the metadata assigned to the information resource. It is in the point that makes it possible to obtain.

  Another object of the present invention is to reliably obtain a more accurate search result each time an additional search keyword is input without causing a search omission when the search keyword input is executed a plurality of times. It is in the point that makes it possible.

  Another object of the present invention is to make it possible to simultaneously specify keywords belonging to a plurality of facets on a thesaurus system as search conditions.

  The principle of the present invention is a free word search for searching a term described in the contents of each information resource in a search of information resources on the web, such as a document or digital content with metadata as bibliographic information, Each information resource is merged with a thesaurus keyword search that searches for keywords assigned to facets that are defined in a hierarchical structure in a tree structure for each concept category in the thesaurus system. The second score for each information resource is calculated each time a thesaurus keyword is input and specified, and the first and second scores are added together in descending order of the combined score. The information resources of the search result are sorted and displayed and output.

  More specifically, as an example of a user interface for a user to search for information resources, a plurality of facets having a tree structure on a thesaurus system can be displayed, and a plurality of arbitrary keywords on an arbitrary facet can be selected and input. (Multi-faceted keyword designation) and the input of a search free word is enabled, and each time a keyword or free word on the thesaurus is input, the score of each information resource is calculated.

  In calculating this score, both the first score and the second score are weighted not only to the information resource that matches the input specified search condition, but also to the related information resource that does not match the search condition. Score is added. This weighting may preferably be performed according to the distance on the facet of the thesaurus tree.

  In the display output of search results, the information resources with the highest score are added and sorted in order. Based on the score of each information resource calculated each time a new keyword or free word is input, the output search result list is also updated and displayed. For example, a key for clearing the score addition may be provided on the search input screen, and the score addition may be repeatedly processed unless the clear key is operated. As a result, search results are always obtained, and after each search result is output, each time a search condition is additionally specified, the search accuracy is improved and more desired information resources are displayed at the top of the list. The

  In the display output of search results, the list of information resources that match the input keyword or free word and the list of information resources that do not match but are related to the search conditions are displayed in separate fields, for example, as separate lists. Or the like. Information resources that are "a bit related to search conditions" because the search conditions specified by the user are not always appropriate search conditions due to incomplete metadata and / or unfamiliarity with the user If the list is displayed together with all or part of the title and bibliographic information and / or keywords of the facets, the user can search for more appropriate search conditions (keywords on free words and / or thesaurus). You can easily make corrections to this search condition.

  According to an aspect of the present invention, metadata provided to an information resource, the metadata including a thesaurus keyword describing a node name in a facet on a thesaurus for each information resource, and the information resource A metadata storage unit for storing metadata including a free word and a pair of importance levels described therein, an input thesaurus keyword for searching for the information resource, and an input unit for inputting an input free word, For an information resource that includes the free word that matches the input free word in the metadata, a first score is calculated based on the degree of importance paired with the matching free word, and the first score is retained A first score calculation unit, the thesaurus keyword that matches the input thesaurus keyword, or the input thesaurus skier A second score is calculated for an information resource that includes in the metadata a thesaurus keyword belonging to a lower node of the matching thesaurus keyword in the facet tree structure to which the node belongs, and the thesaurus keyword belonging to another node in the facet For the information resource including the metadata in the metadata, a second score is obtained by multiplying the second score by a weight β (β ≦ 1) corresponding to the distance on the tree structure, and the second score is retained. 2 score calculation units, and for each information resource, the first score and the second score are combined to obtain a third score, and the third score is sorted in descending order. There is provided an information resource search apparatus comprising an output unit that displays a list of information resources as search results.

  The first score calculation unit holds the first score calculated by the input of the input free word when the input free word is input subsequent to the output of the search result by the output unit. The first score may be added to the first score, and the output unit may update and display the information resources sorted again in descending order of the third score.

  When the input thesaurus keyword is input subsequent to the output of the search result by the output unit, the second score calculation unit holds the second score calculated by inputting the input thesaurus keyword. The second score may be added to the second score, and the output unit may update and display the information resources sorted again in descending order of the third score.

  The input unit may selectively display a plurality of facet tree structures, and simultaneously select nodes on the displayed tree structures.

  The output unit includes an information resource including, in the metadata, the thesaurus keyword that matches the input thesaurus keyword, or a thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs. A first display output column that displays a list sorted in descending order of the third score, and an information resource that includes the thesaurus keywords belonging to other nodes in the facet in the metadata, A second display output field for sorting and displaying a list in order may be provided.

  The output unit may substantially display a list of the number of information resources that can be displayed at one time within the effective display area of the display device.

  The information resource search apparatus further includes a narrowing condition input unit for inputting a narrowing condition thesaurus keyword for narrowing down the search result of the information resource, the thesaurus keyword matching the input narrowing condition thesaurus keyword, or the narrowing condition A filtering processing unit that sets, as information resources to be output, only information resources that include a thesaurus keyword belonging to a node lower than the matching thesaurus keyword in the facet tree structure to which the thesaurus keyword belongs. .

  The information resource search apparatus may further include a third display output field for displaying a list of information resources that do not belong to any node in the facet, sorted in descending order of the third score.

  According to another feature of the present invention, metadata provided to an information resource, the metadata including a thesaurus keyword describing a node name in a facet on a thesaurus for each information resource, and the information A metadata storage unit that stores metadata including a free word described in a resource and a pair of importance thereof, and the free word that matches an input free word for searching for the information resource in the metadata A first score is calculated based on the degree of importance paired with the matching free word, and a first score calculation unit that holds the first score is searched for the information resource to be included. The thesaurus keyword that matches the input thesaurus keyword for the search, or the match on the facet tree structure to which the input thesaurus keyword belongs A second score is calculated for an information resource that includes a thesaurus keyword belonging to a subordinate node of a thesaurus keyword in the metadata, and an information resource that includes a thesaurus keyword that belongs to another node in the facet in the metadata. A second score is obtained by multiplying the second score by a weight β (β ≦ 1) corresponding to the distance on the tree structure, and a second score calculation unit that holds the second score, for each information resource A score summation unit that obtains a third score by summing the first score and the second score, and a search result to be output as a list of information resources sorted in descending order of the third score An information resource search server device characterized by comprising a sorting unit obtained as described above is provided.

  According to another aspect of the present invention, an input free word for searching for information resources is input, a plurality of facet tree structures on the thesaurus are selectively displayed, and the displayed nodes on the tree structure And an input resource for inputting an input thesaurus keyword, and a set of information resources having metadata matching the input free word or the input thesaurus keyword based on the metadata assigned to the information resource. Regarding the first information resource group, a score is calculated and the second information resource group is a set of information resources having metadata matching the thesaurus keyword belonging to another node in the facet to which the input thesaurus keyword belongs , The processing unit for calculating the score, and the above-mentioned sorted in descending order of the score And 1 information resource group, wherein a second resource group information resource retrieval client apparatus characterized by comprising an output section for displaying a screen on distinguishably lists are provided.

    According to another aspect of the present invention, metadata provided to an information resource, the metadata including a thesaurus keyword describing a node name in a facet on the thesaurus for each information resource. On the facet tree structure to which the input thesaurus keyword that matches the input thesaurus keyword or the input thesaurus keyword belongs A score is calculated for the information resource including the thesaurus keyword belonging to the lower node of the matching thesaurus keyword in the metadata, and the information resource including the thesaurus keyword belonging to the other node in the facet is included in the metadata. The tree structure in the score A score is calculated by multiplying the weight β according to the distance above (β ≦ 1), a score calculation unit that holds the score, and an output unit that displays a list of information resources sorted in descending order of the score as a search result. There is provided an information resource search device characterized by comprising.

  According to another feature of the present invention, metadata provided to an information resource, the metadata including a thesaurus keyword describing a node name in a facet on a thesaurus for each information resource, and the information Storing metadata including a free word described in the resource and a pair of importance levels in the metadata storage unit; inputting an input thesaurus keyword and an input free word for searching the information resource; The first score is calculated based on the importance of pairing with the matching free word for the information resource including the free word that matches the input free word in the metadata, and the first score is calculated. And the thesaurus keyword matching the input thesaurus keyword, or the input thesaurus ski A second score is calculated for an information resource that includes in the metadata a thesaurus keyword belonging to a lower node of the matching thesaurus keyword in the facet tree structure to which the node belongs, and the thesaurus belonging to other nodes in the facet For the information resource including the keyword in the metadata, the second score is multiplied by the weight β (β ≦ 1) corresponding to the distance on the tree structure to obtain the second score, and the second score is retained. For each information resource, adding the first score and the second score to obtain a third score, and using the information resources sorted in descending order of the third score as search results An information resource search method including a step of displaying a list.

  According to another feature of the present invention, metadata provided to an information resource, the metadata including a thesaurus keyword describing a node name in a facet on a thesaurus for each information resource, and the information Storing metadata including a free word described in a resource and a pair of importance levels in a metadata storage unit; and selecting the free word that matches an input free word for searching the information resource as the meta data. For the information resource included in the data, a step of calculating a first score based on the degree of importance paired with the matching free word, retaining the first score, and an input for searching for the information resource The thesaurus keyword that matches the thesaurus keyword, or the match on the facet tree structure to which the input thesaurus keyword belongs A second score is calculated for an information resource that includes a thesaurus keyword belonging to a subordinate node of a thesaurus keyword in the metadata, and an information resource that includes a thesaurus keyword that belongs to another node in the facet in the metadata. The second score is multiplied by a weight β (β ≦ 1) corresponding to the distance on the tree structure to obtain a second score, the second score is retained, and for each information resource, the first score Adding a score and the second score to obtain a third score; and obtaining a list of information resources sorted in descending order of the third score as a search result to be output. An information resource search method characterized by the above is provided.

  According to another aspect of the present invention, an input free word for searching for information resources is input, a plurality of facet tree structures on the thesaurus are selectively displayed, and the displayed nodes on the tree structure And selecting an input thesaurus keyword, and based on the metadata assigned to the information resource, a first set of information resources having metadata matching the input free word or the input thesaurus keyword For a second information resource group, which is a set of information resources having metadata corresponding to a thesaurus keyword belonging to another node in the facet to which the input thesaurus keyword belongs, while calculating a score for one information resource group, The step of calculating the score and before sorting in descending order of score A first resource group, and the second resource group information resource retrieval method characterized by including the step of displaying on the screen distinguishably lists is provided.

  According to another aspect of the present invention, there is provided an information resource search program for causing a computer to execute an information resource search process, the program being metadata attached to the information resource to the computer, Metadata stores, for each information resource, metadata including a thesaurus keyword that describes a node name in a facet on a thesaurus and a pair of free words and importance levels described in the information resource. Processing for storing the information resource, processing for inputting an input thesaurus keyword and input free word for searching the information resource, and information resource including the free word that matches the input free word in the metadata. The first score is calculated based on the importance level paired with the free word, and the first score is maintained. And the information resource including in the metadata the thesaurus keyword that matches the input thesaurus keyword, or the thesaurus keyword that belongs to the lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs. 2 for the information resource including the thesaurus keywords belonging to other nodes in the facet in the metadata, the weight β corresponding to the distance on the tree structure (β ≦ 1) ) To obtain a second score, and to retain the second score; and for each information resource, a process for adding the first score and the second score to obtain a third score; , Processing for displaying a list of information resources sorted in descending order of the third score as search results; Information resource retrieval program characterized in that for executing a process comprising is provided.

According to another aspect of the present invention, there is provided an information resource search program for causing a computer to execute an information resource search process, the program being metadata attached to the information resource to the computer, Metadata stores, for each information resource, metadata including a thesaurus keyword that describes a node name in a facet on a thesaurus and a pair of free words and importance levels described in the information resource. Processing to be stored in the department,
For an information resource that includes in the metadata the free word that matches the input free word for searching for the information resource, a first score is calculated based on the importance paired with the matching free word; A process for holding the first score and the thesaurus keyword that matches the input thesaurus keyword for searching the information resource, or a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs A second score is calculated for the information resource including the thesaurus keyword belonging to the metadata in the metadata, and the information score including the thesaurus keyword belonging to another node in the facet is included in the second score as the second score. Weight β (β ≦ 1) according to the distance on the tree structure A second score obtained by multiplying, a process for retaining the second score, a process for obtaining a third score by adding the first score and the second score for each information resource, An information resource search program is provided for executing a process including a process of obtaining a list of information resources sorted in descending order of the third score as a search result to be output.

  According to another aspect of the present invention, there is provided an information resource search program for causing a computer to execute an information resource search process, wherein the program inputs an input free word for searching the information resource to the computer. In addition, by selectively displaying a tree structure of a plurality of facets on the thesaurus and selecting nodes on the displayed tree structure, a process for inputting an input thesaurus keyword, and a meta data assigned to the information resource Based on the data, a score is calculated for the first information resource group that is a set of information resources having metadata that matches the input free word or the input thesaurus keyword, and other scores in the facet to which the input thesaurus keyword belongs are calculated. Information with metadata that matches a thesaurus keyword belonging to For the second information resource group that is a set of resources, a process for calculating a score, the first information resource group sorted in descending order of the score, and the second information resource group are displayed on the display screen. There is provided an information resource search program characterized in that the program includes a process including a process of displaying a list in a distinguishable manner.

  According to the present invention, even when metadata that is bibliographic information assigned to information resources available on the web is incomplete or an inappropriate search keyword is input by an unskilled user. However, it is possible to always obtain the search results without impairing the search efficiency, and to obtain the search results that are more suitable for the search conditions.

  Further, even when the input search keyword does not match the description on the metadata given to the information resource, it is possible to obtain a list of information resources having relevance to the input search keyword.

  In addition, when a search keyword is input a plurality of times, it is possible to reliably obtain a more accurate search result each time an additional search keyword is input without causing a search omission.

  Furthermore, keywords belonging to a plurality of facets on the thesaurus system can be simultaneously specified as search conditions.

  Therefore, the search efficiency of information resources on the web on the user side is greatly improved, so that the distribution of information resources on the web is promoted and the data availability is improved.

  Embodiments of the present invention will be described below with reference to the drawings.

<Functional configuration of this embodiment>
FIG. 1 shows a configuration example of an information resource search apparatus according to the present embodiment.

  The information resource search device according to this embodiment includes a data storage unit 3 that is an external storage device that stores information resources that can be used on the web, such as documents, digital contents such as photographs, maps, and rare books. Needless to say, this embodiment does not limit the input means input to the data storage unit 3. Further, the input means may read information resources recorded on an arbitrary external recording medium such as a CD-ROM, DVD, MO, etc., for example, in addition to means for directly accepting content input, and accept them as input. The data storage unit 3 stores metadata 31 that is bibliographic information of each information resource, and the document text of the information resource or the content 32 itself. Alternatively, in light of the number of information resources to be searched, the desired processing speed, the capacity of the data storage unit 3, whether the text or content itself is also subject to free word search, etc. The document text or content itself 32 may be stored in the information resource search apparatus according to the present embodiment and another server apparatus (not shown) connected on the network via a communication line such as the Internet. As a description method of the metadata 31, an arbitrary method may be adopted. For example, the above-described Dublin Core method may be used, and in this case, title (resource title), creator (creator), Subject (the nature or genre of the topic included in the content of the resource), coverage (range or target of the resource, geographical division, time division, etc.), rights (statement of copyright, industrial property rights, etc.) All or some of the element types can be defined as metadata elements. The document body of the information resource or the content itself 32 may be stored as an HTML file, PDF file, text file, CSV file, etc. if it is a document, and other content such as photographs, maps, rare books, etc. It may be stored as image data or the like, and may be compressed by an arbitrary compression method according to the characteristics of the content as necessary.

  The information resource search apparatus according to the present embodiment performs an input unit 1 for inputting a search condition for searching for information resources on the web, and performs a search process according to the present embodiment according to the search condition input to the input unit 1. The search execution unit 2 to be executed, the output unit 4 that outputs the search result obtained by the search execution unit 2 to an arbitrary output device, and the score calculated in the search execution unit 2 are stored in the data storage unit 3 And an automatic metadata generation unit 5 for automatically generating new metadata to be supplemented with the metadata 31 of the information resource to be processed.

  More specifically, the input unit 1 is a search free word input unit 11 for inputting a search free word for searching for a word (hereinafter referred to as “index word”) described in the contents of each information resource. And keywords that are defined in correspondence with nodes on facets that are directories of a tree structure in the thesaurus system (hereinafter referred to as “thesaurus keywords”), and that are added in the scoring process described later. A thesaurus keyword input unit 12 for adding points, a narrowing condition thesaurus keyword input unit 13 for selectively inputting a narrowing condition thesaurus keyword that is a thesaurus keyword given as a narrowing condition to the list of information resources obtained as a search result, and a user Attention keyword to select and input one keyword to be noted Comprising a de input section 14. In the input unit 1, for example, one or more of the search free word input unit 11, the narrow-down condition thesaurus keyword input unit 13, and the attention keyword input unit 14 may be omitted.

  The search execution unit 2 determines whether an index word that matches and is similar to the search free word input from the search free word input unit 11 is previously defined as a part of the metadata 31, or alternatively, the index word is The information resources extracted at the time of the search are searched, a first score is calculated for each of the searched information resources, and the free word search scoring unit 21 that holds the calculated first score, Search for an information resource in which a keyword that matches the point-added thesaurus keyword input from the thesaurus keyword input unit 12 and a keyword that is located at a distance close to the keyword on the facet tree structure is defined as a part of the metadata 31 in advance. For each searched information resource, a second score is calculated and the calculated second score is calculated. The keyword search scoring unit 22 that holds the information, the first score calculated by the free word search scoring unit 21, and the second score calculated by the keyword search scoring unit 22 are added together for each information resource. A score summation unit 23, a sorting unit 24 that sorts information resources in descending order of the summation score (corresponding to the “third score” in the claims) calculated by the score summation unit 23, and a narrow-down condition thesaurus A filtering processing unit 25 that extracts only information resources that meet the narrowing-down conditions input from the keyword input unit 13 and passes them to the sorting unit 24;

  The output unit 3 displays the facets on the thesaurus corresponding to the selected thesaurus keywords, and all the bibliographic information such as subjects, creators, titles, etc. described in the metadata of the information resources output from the sort unit 24 Alternatively, it includes a facet correspondence display unit 41 that displays a list of a part and a link to each information resource or a URL thereof.

  In this specification, a thesaurus is a database that systematically indicates what concepts or words belong to a certain concept, and a facet is a higher rank for each category of concepts in this thesaurus system. A tree structure (directory) in which concept nodes and subordinate concept nodes are linked in a hierarchical manner. FIG. 12 shows an example of this facet. The top level concept “Japan” is linked to its subordinate concepts “Kanto” and “Kansai”, and the “Kanto” node is linked to its subordinate concepts “Tokyo” and “Kanagawa”, and “Tokyo” The “Chuo-ku” node is linked to the lower level of the node, while the “Kansai” node is linked to its subordinate concept “Osaka Prefecture”, and the “Osaka City” node is further subordinate to the “Osaka Prefecture” node. A “Chuo-ku” node is further linked below the “Osaka City” node. Each of these node names shown in FIG. 12 is defined as metadata 31 having an attribute of coverage spatial (resource geographic division) and can be added to an information resource, for example, according to the Dublin Core system. A plurality of facets can be applied to one information resource. For example, in addition to the geographical divisions shown in FIG. 12, each facet tree structure such as a historical division, a resource type (type), a theme (subject), etc. A concept positioned in any of the above may be defined as the metadata 31.

FIG. 2 exemplarily shows the layout of the work area used for calculating the metadata 31 and the score of each information resource stored in the data storage unit 3 shown in FIG. In FIG. 2, for example, a document information resource whose title is “Guidance for Facility X” is a node name (keyword) located on a facet of the thesaurus system, and “history” and “era” for the “theme” facet (F1). For facet (F2), blank (undefined), for “geography” facet (“geographic division” facet) (F3) for “Tokyo Chiyoda-ku” and “Yokohama Naka-ku”, for “institution” facet (F4) “Tokyo Metropolitan Government Office” and “Kanagawa Prefectural Office” are defined as metadata 31, respectively, and in order to efficiently calculate the first score for the search free word input from the search free word input unit 11, the index word (word a, B) and importance within the information resource of the index word _{(S i1,} S _i2) metadata 31 To have been defined. The index word may be assigned to each information resource in advance or may be automatically extracted. Alternatively, the index word for each information resource and the importance of each index word may be obtained when the first score is calculated in the free word search scoring unit 21. In FIG. 2, the work area is a thesaurus keyword input from the input free word storage area 231 input to the search free word input unit 11, the scoring thesaurus keyword input unit 12 and / or the refinement condition thesaurus keyword input unit 13. Storage area 233, each time a free word and a keyword are input, a first score and / or a second score of the information resource is calculated, and a sum storage area 235 obtained by updating the sum of the two is used. Prepare. The thesaurus keyword storage area 233 is provided with cells corresponding to each keyword of facets defined in advance in the metadata 31 (F1 “theme”, F2 “era”, F3 “geography”, F4 “institution”), You may provide the field which shows whether it input as an object for point addition, or as a narrowing-down condition for every cell. The work area may be arranged in an area continuous with the metadata 31, but may alternatively be provided in another temporary storage area such as a RAM or a cache memory.

<Processing flow of search processing according to this embodiment>
3 to 8 show the detailed processing procedure of the information resource search process performed by the search execution unit 2 in this embodiment.

  FIG. 3 is a flowchart showing an outline of search processing and search result display processing performed by the information resource search apparatus according to this embodiment. In FIG. 3, when an input search condition is output from the input unit 1 to the search execution unit 2, the search execution unit 2 determines the type of the input search condition (step S1), and if it is a search free word Then, the character string S of the search free word is input to the free word search scoring unit 21 or the input free word is changed (step S2), and if it is a thesaurus keyword for addition, it is input to the keyword set W used for addition The added thesaurus keyword is added or deleted (step S3), and if it is a refinement condition thesaurus keyword, the thesaurus keyword input to the keyword set N used for the refinement condition is added or deleted (step S8). When the input is not empty (step S4N), the search process according to the present embodiment is executed (step S5), and a list of information resources that are search results obtained by the search process is displayed and output (step S6). The processing from step S1 to step S6 is repeated (step S7N) until the desired information resource is obtained or until a search processing end operation (for example, end of the search engine application) is performed (step S7Y).

  FIG. 4 is a flowchart showing the procedure of the search process executed in step S5 of FIG. In FIG. 4, when the narrowing condition is input, whether the information resource (for example, document, hereinafter referred to as “document”) i to be processed is suitable for the input narrowing condition or not in the filtering processing unit 25. Judgment is made (step S51). Details of the narrowing-down condition processing in step S51 will be described later with reference to FIGS. When the document i matches the input narrowing condition (step S52Y), it is determined whether either a free word or a thesaurus keyword for addition is input, and when either is input (step S53Y). The score of the document i is calculated (step S54), and the document i is added to the search result list (step S55). Details of the scoring process in step S54 will be described later with reference to FIG. In step S53, when neither a free word nor a thesaurus keyword for adding points is input (step S53N), the process proceeds to step S55. The processing from step S51 to step S55 is repeated until all the documents are processed (step S56).

  FIG. 5 is a flowchart showing details of the narrowing-down condition processing in step S51 of FIG. In FIG. 5, it is determined whether or not a narrowing condition thesaurus keyword is input (step S511). If a narrowing condition thesaurus keyword is input (step S511Y), the document i matches the narrowing condition of facet f. (Step S512), and if it matches (step S513Y), the processing from step S512 to step S513 is repeated until it is determined whether all facets meet the narrowing conditions (step S514), and the document It is assumed that i matches the input narrowing condition and is a search result candidate (step S515). If no narrowing condition keyword is input in step S511 (step S511N), the process proceeds to step S515. If the document i does not conform to the narrowing condition of facet f in step S513 (step S513N), it is determined that the document i does not conform to the input narrowing condition and is deleted from the search result candidates (step S516).

FIG. 6 is a flowchart showing details of the determination process of whether or not the document i in step S512 of FIG. 5 meets the narrowing condition of the facet f. In FIG. 6, it is determined whether or not there is a narrowing condition keyword N belonging to facet f (step S5121). If there is any (step S5121Y), a keyword belonging to facet f from narrowing condition keyword N is determined. extracting a set X (step S5122), whether the document i is not related to the keyword _{X j} is determined, and if not relevant (step S5123Y), for other keywords belonging to keyword set X, document i is Whether or not there is a relationship is repeatedly determined (step S5124), and if it is not related to any keyword (step 5124Y), it is determined that the document i does not conform to the narrowing condition of facet f (step S5125). In step S5121, if there is no narrowing condition keyword N belonging to facet f (step S5121N), it is determined that document i conforms to the narrowing condition of facet f (step S5126). If it is determined in step S5123 that the document i is related to the keyword Xj (step S5123N), the process proceeds to step S5126.

FIG. 7 is a flowchart showing details of the scoring process for document i in step S54 of FIG. In FIG. 7, it is determined whether the input search condition is a free word or a thesaurus keyword for adding points. If a free word is input (step S451Y), the document i of the free word character string S is searched. Score a _i is calculated (step S542). Details of the free word scoring process (first score calculation process) in step S542 will be described later with reference to FIG. In step S541, when a search free word is not input (step S541N), the process proceeds to step S546. In step S543, if the point addition Thesaurus keyword is input (step S543Y), it added point Thesaurus keyword _{W j} calculates a score _{b ij} of document i by (step S544), for all of the input point addition Thesaurus Keyword The process of step S544 is repeated until the score is calculated (step S545). When scores are calculated for all the added thesaurus keywords for input (step S545Y), the process proceeds to step 546. In step S543, when the point addition thesaurus keyword is not input (step S543N), the process proceeds to step S546. In step S546, the first score calculated in step S542 and the second score calculated in step S544 are added together, and the score of the document i is calculated as c _i = a _i + Σb _ij .

<Details of first score calculation process based on search free word and pre-registration process of index word and importance>
The first score calculation process executed by the free word search scoring unit 21 quantifies the degree of association between the search free word input to the search free word input unit 11 and the information resource to be searched, and uses it as a score. . The search free word to be input is designated by a list of free keywords expressed in a natural language such as Japanese or English, or a natural sentence including punctuation marks.

FIG. 8 is a flowchart showing details of the scoring process (first score calculation process) using free words in step S542 in FIG. In FIG. 8, the search term is cut out from the character string of the search free word to create the word set Q (step S5421), the first score ai of the document i is initialized to 0 (step S5422), and the first Score a _i + = Q _k weight S _ij in document i (step S5423). The process of step S5423 is repeated until the weights for all after the search are added (step S5424). In step S5421, it is not necessary to create the word set Q every time a document is processed.

  Although any method can be used to calculate the importance of the index word as the metadata 31 that is a basic numerical value for the first score calculation, for example, the score is calculated by a vector space model (VSM). And a technique for calculating a score using a relevance-based superposition model (RS model) can be implemented in the free word search scoring unit 21 according to the present embodiment.

FIG. 9 is a flowchart showing a processing procedure for calculating the index word and importance in the information resource for calculating the first score and registering it in the metadata 31 when the vector space model is used. In FIG. 9, a word set T is created by cutting out words to be index words from the text of document i or metadata 31 (step S92). The importance of the word Tk belonging to the word set T is calculated by the tfidf (term frequency inverse document frequency) method with the number of applications of the word Tk in the document i and the number of documents including the word Tk in all documents as arguments. (Step S93), the processing of Step S93 is repeated until the importance level for all index words is calculated (Step S94), and the processing of Step S92 to Step S94 is repeated until the calculation of the importance level for all documents is completed. (Step S95). From the words of which importance is calculated, one or a plurality of words having the highest importance is scored in the metadata 31 as shown in FIG. 2 together with the corresponding importance as index words. Register as a free word and the degree of importance paired with it. The tfidf method is a method for calculating the importance of a word in consideration of how often the word appears in the document and how often the word does not appear in other documents. The importance D _{j, i} of the word ti (i = a,..., n) in dj is expressed by the following equation (1).

Here, tc _{j, i} is the number of occurrences of the word ti in the document dj, tf _{j, i} = tc _{j, i} / max (tc _{j, i} ), and df _i is the number of occurrences of the word t _{i in} all the documents. , DN is the total number of documents.

FIG. 10 is a flowchart illustrating a processing procedure for calculating the index word and importance in the information resource for calculating the first score and registering in the metadata 31 when the RS model is used. The RS model is a document search of a vector space model, which classifies documents dj based on the accessibility such as including the same keyword to create a document cluster C _k (k = A,..., N) representative vector r _k representing features of document clusters _{Ck (k = a, ...,} N) generates a further modeling technique for correcting the document vector D _j by using the surrogate table vector r _k. For example, Patent Document 2 (Japanese Unexamined Patent Application Publication No. 2004-310199) discloses a document classification method using this RS model. Here, the document cluster is a collection of documents having a semantic group by keywords, and has the same topic.

FIG. 11A is a schematic diagram for specifically explaining the RS model. In FIG. 11 (a), 2 single keyword A and B are documents d1, ..., shows a case present in d5, the document _{d j} containing the keyword A in the document cluster _{C A,} documents containing keywords B d _j belongs to the document cluster C _B , and the document d _j including both the keywords A and B belongs to both the document clusters C _A and C _B. Form that is in the RS model, we are able to perform a non-exclusive document classification, a situation where one of the document d _j is over a plurality of keyword (topic), that belongs to more than one document cluster C _k It can be expressed with. Then, for example, root-mean-square (RMS) is also calculated for each of the document vectors D _j of the documents d _j included in the created document clusters C _A and C _B , respectively. Generate representative vectors r _a and r _b for document clusters C _A and C _B. Here, the representative vector r _k of the document cluster C _k is expressed by the following equation (2). Incidentally, the representative vector r _k not represent the characteristics of the document cluster C _k, a feature vector in the same space as the document vector D _j, with the same number of dimensions and the document vector D _j.

Here, r _{k, i} is the i-th element of the representative vector r _k , | C _k | is the number of documents included in the document cluster C _k , D _{j, i} is the i-th element of the document vector D _j of the document d _j is there.

Next, as shown in FIG. 11B, this representative vector r _k is used to correct the document vector D _j of each document d _j , which is the same for all document clusters C _k to which the document d _j belongs. and RMS representative vector r _k, is compared with the document vector D _j for each element, if the former intended to replace as new elements of the document vector D _j, is represented by the following equation (3).

Here, S _{j, i} is a correction vector, C (d _j ) is a document cluster to which the document d _j belongs, and | C (d _j ) | is the number of document clusters to which the document d _j belongs.

By using the RS model, the value of the document vector D ′ _j is calculated in consideration of not only the feature quantity originally provided in the document d _j but also the feature quantity of the document cluster C _k having the same keyword. Can do.

Returning to FIG. 10, the importance (document vector D _j ) of all the index words extracted for all documents is calculated using, for example, the tfidf method shown in FIG. 9 (step S102). Next, as shown in FIG. 11A, the documents are classified by topic (document cluster) (step S102). Here, one document may belong to a plurality of topics (clusters). The importance (representative vector) of the index word is calculated for all topics (clusters) (step S104). For example, the above RMS may be used for calculating the representative vector. The document vector is corrected based on the calculated representative vector (step S105). Based on the corrected document vector D ′ _j , one or a plurality of words from the words having the highest importance among the already calculated importance corrected are used as index words together with the importance as shown in FIG. As shown in FIG. 2, the free word for scoring and the importance are registered in the metadata 31.

  In the index word importance calculation processing of FIG. 9 or FIG. 10, the document to be processed needs to be a natural sentence or a keyword list, and the metadata 31 previously given as bibliographic information to the information resource. A part or the whole may be a processing target, or alternatively, a document body may be a processing target.

<Details of second score calculation process based on scoring thesaurus keywords according to this embodiment>
The second score calculation process executed by the keyword search scoring unit 22 includes a scoring thesaurus keyword selected and input to the scoring thesaurus keyword input unit 12 and a scoring thesaurus defined in advance for the information resource to be searched. The distance between the keywords on the facet tree structure of the thesaurus is calculated, and the degree of coincidence or relevance weighted according to the calculated distance is digitized to obtain a score. The added thesaurus keywords to be inputted can be designated by selecting a desired node on a facet tree structure of a thesaurus that can be selectively displayed on a display device, for example.

  The second score calculation process by the keyword search scoring unit is executed according to the following procedure.

First, let D _{0 be} the result set of the first score calculation process (process according to the procedure of FIG. 8). When the search free word is not specified and the result set is empty, D ₀ is set to the entire set of information resources to which the keyword specified by the attention condition input from the attention keyword input unit 14 is assigned.

Α is preferably set to a value corresponding to the characteristics of the facet (lexical table) in the thesaurus system, for example, 2.

  -It is preferable to describe the limit on the number of hops per facet in the configuration file when data is input.

  -Even if point addition processing is performed, the source of the set of resources (resources that are constituent elements) does not increase or decrease.

For example, in the facet structure shown in FIG. 12, when “Tokyo” is input as a thesaurus keyword for scoring, as understood from FIG. 13, “Tokyo” and “Tokyo” A score α is added to a document to which the scoring thesaurus keyword “Chuo-ku” is assigned as the metadata 31, and the scoring thesaurus keywords “Kanto”, “Kanagawa”, and “Yokohama” are added to the metadata 31. The score is added to the document given as “Japan”, “Kansai”, “Osaka Prefecture”, “Osaka City”, and “Chuo Ward”, which is positioned under this “Osaka City”. A score αβ ² is added to a document to which a thesaurus keyword is assigned as metadata 31.

  Alternatively, in the second score addition process in the keyword search scoring unit 22, for each facet, it is possible to set the score calculation method by the distance on the facet tree structure of the thesaurus and / or whether or not to add points. May be.

<Details of score summation processing according to this embodiment>
FIG. 13 shows each document in the case where “Tokyo” is selected and inputted as a scoring thesaurus keyword and words “A” and “C” are inputted as search free words on the facet structure shown in FIG. The score calculated by the score summation unit 23 is shown. The document 1 includes a geography classification “Japan” and an age classification “21st century” as scoring thesaurus keywords, and “A” (importance S ₁₁ ) and “C” (importance S ₁₃ ) as scoring free words. Are defined as metadata 31, the combined score of the document 1 calculated by the score adding unit 23 is αβ ² + S ₁₁ + S ₁₃ . In the document 2, the geography division “Tokyo” and the era division “20th century” are defined as the scoring thesaurus keywords, and “B” (importance S ₂₂ ) is defined as the metadata 31 as the scoring free words. The score of the document 2 calculated by the score summation unit 23 is α. Document 3 includes a geography classification “Chuo-ku (located below Tokyo)” and a period classification “19th century” as scoring thesaurus keywords, and “A” (importance S ₃₁ ) as a scoring free word. Assuming that “B” (importance S ₃₂ ) is defined as the metadata 31, the combined score of the document 3 calculated by the score adding unit 23 is α + S ₃₁ . The document 4 includes a geography classification “Kanto” and a period classification “20th century” as scoring thesaurus keywords, and “B” (importance S ₄₂ ) and “C” (importance S ₄₃ ) as free words for scoring. Are defined as metadata 31, the combined score of the document 4 calculated by the score adding unit 23 is αβ + S ₄₃ . The document 5 defines the geographic classification “Yokohama City” and the period classification “19th century” as scoring thesaurus keywords, and the metadata 31 as the scoring free word “C” (importance S ₅₃ ). The score of the document 5 calculated by the score summation unit 23 is αβ + S ₅₃ . The document 6 includes the geography classification “Chuo-ku (located below Osaka City)” and the period classification “21st century” as scoring thesaurus keywords, and “A” (importance S ₆₁ ) as the scoring free word. Are defined as metadata 31, the combined score of the document 1 calculated by the score adding unit 23 is αβ ² + S ₆₁ .

<Details of Filtering Processing by Refinement Condition Thesaurus Keyword According to this Embodiment>
In the filtering process executed by the filtering processing unit 25, scoring thesaurus keywords that match the filtering condition thesaurus keyword selected and input to the filtering condition thesaurus keyword input unit 13 among the documents for which the combined score has been calculated are metadata. Only the document defined as 31 is extracted and output to the sorting unit 24. The narrow-down condition thesaurus keywords to be input can be specified by selecting a desired node on the facet tree structure of the thesaurus that can be selectively displayed on the display device, for example.

  The filtering process by the filtering processor 25 is executed in the following procedure.

A set of resources to which a keyword k _i specified as a narrowing condition is assigned is K _i . Refine result set D ₂ is defined by the following equation.

D ₂ = D ₁ ∩ (K ₀ ∪K ₁ ...) ∩ (K _a ∪ ...) ∩ (K _b ∪ ...) ...
Here, the parenthesized set is a keyword resource set belonging to the same facet.

  ・ The score of information resources does not change even if the conditions are met.

  In FIG. 14, “Tokyo” is selected and inputted as a thesaurus keyword for scoring on the facet structure shown in FIG. 12, words “A” and “C” are inputted as search free words, and “ For each document when “20th century” and “21st century” are selected and input, the document extracted by the filtering processing unit 25 is shown. The filtering processing unit 25 regards an OR condition (logical sum) when a plurality of narrowing conditions within the same facet are input, and assumes an AND condition (logical product) when a plurality of narrowing conditions between different facets are input. Execute the filtering process. 14, among the documents 1 to 6 for which the combined score is calculated in FIG. 13, the documents 3 and 5 in which the age classification “19th century” is defined as the metadata 31 as the scoring thesaurus keyword Is cleared and is not output to the sorting unit 24. Note that no score is added to the documents 1, 2, 4, and 6 that conform to the narrowing-down conditions.

In FIG. 15, the words “A” and “C” are input as search free words on the facet structure shown in FIG. 12, and “Tokyo,” “Osaka,” “20th century” and “ For each document when “21st century” is selected and input, the document extracted by the filtering processing unit 25 is shown. The filtering processing unit 25 executes the filtering process by regarding that the subordinate concept of the thesaurus keyword specified as the narrowing-down condition is also applicable to the narrowing-down condition. 15, among the documents 1 to 6 for which the combined score is calculated in FIG. 13, the document 2 in which the geographic classification “Tokyo” and the period classification “20th century” are defined as the metadata 31 as the scoring thesaurus keywords. , And as a scoring thesaurus keyword, the document 6 in which the geographic classification “Chuo-ku (located under Osaka Prefecture)” and the period classification “21st century” is defined as the metadata 31 is extracted as a document that matches the filtering condition. The other documents 1, 3, 4, and 5 are not extracted because a scoring thesaurus keyword that matches the narrow-down condition thesaurus keyword is not defined. The total score of the document 2 is 0 because the scoring free words that match the input search free words “A” and “C” are not defined. The total score of the document 6 is the importance S ₆₁ for the word “A” defined as the free word for scoring.

  Alternatively, in the score summation process in the score summation unit 23, if an input thesaurus keyword is input subsequent to the output of the search result, the score of the newly input keyword is added to the currently stored summation score. Alternatively, it may be possible to set whether the total score (or a part thereof) currently held is replaced. In this case, a rule is preferably defined in relation to the facet. For example, when a plurality of thesaurus keywords are selected in the same facet, the summed score currently held is replaced, and a new keyword is replaced with a different facet. If selected, it may be configured to be added to the currently held total score.

<User interface configuration example in this embodiment>
16 to 23 show exemplary configurations of user interfaces provided by the input unit 1 and the output unit 4 of the information resource search apparatus according to the present embodiment.

  FIG. 16 shows an example of an initial search screen displayed and output by the information resource search apparatus according to this embodiment.

  In FIG. 16, a search character string input field 164 displayed on the upper left is an example of the configuration of the search free word input unit 11, and a facet keyword selection input field 161 displayed on the left middle is a thesaurus keyword input for adding points. This is a configuration example of the unit 12, the narrow-down condition thesaurus keyword input unit 13, and the attention keyword input unit 14. In the initial search screen of FIG. 16, the facet “theme” on the thesaurus system is selected as the keyword of interest when omitted. Tabs 165 are displayed for each of a plurality of facets “theme”, “period division”, “geographic division”, “institution name”, and “resource type”, and a desired tab is selected and keywords belonging to the facet are selected. By selecting and inputting on the displayed hierarchical structure, it is possible to simultaneously instruct the input of a thesaurus keyword for addition or the narrow-down condition thesaurus keyword across a plurality of facets. In FIG. 16, the right column is an example of the configuration of the facet correspondence display unit 41, and the list 162 displayed in the upper right column shows a list of information resources related to the attention keyword “theme” itself, and the corresponding document exists. Therefore, the list 163 displayed as an empty list and displayed in the lower right column displays a list of, for example, three information resources belonging to “humanities”, which is a subordinate concept of the keyword of interest “theme”. Similarly, information resources belonging to “social science” are continuously displayed in a list. In each list, for example, information resources are preferably sorted and output in descending order of the total score, and information resources having a new update date are preferentially displayed among information resources having the same score.

  FIG. 17 shows an example of a screen in the case where “Hokkaido” 171 is selected and input as a scoring thesaurus keyword, considering that “I want to know the state of Hokkaido in the past” in FIG. In the “current score / squeezing condition” column 175, it is displayed that facets “geography”, “Japan”, and “Hokkaido” are designated as the thesaurus keywords for score addition. The list 172 displayed in the upper right column displays a list of information resources in which the scoring thesaurus keyword “Hokkaido” and its subordinate concepts are defined as scoring keywords. The list 173 displayed in the middle right section does not include the scoring thesaurus keyword “Hokkaido” and its subordinate concepts in the defined scoring keywords, but belongs to “geographic division” which is the same facet as “Hokkaido”, for example. Information resources whose keywords are defined as scoring keywords are listed in the “Related” column as relevant although they do not meet the search conditions. The information resources listed in the “Related” column have keywords that belong to the “Geographic division” that is the same facet as “Hokkaido”. A second score obtained by multiplying the weight according to the distance is calculated. Information resources that do not belong to any of the list 172 and the list 173 are classified as “unclassified” and are displayed in a list 174 displayed in the lower right.

  The number of information resources displayed in each list may be any number, but preferably, the list of all the numbers displayed in the lists 172, 173, and 174 does not require a large amount of screen scrolling. The number is within a range substantially within the display area of one screen.

  In FIG. 18, in order to know the old state of Hokkaido in FIG. 17, we wanted to select the next theme. However, it is difficult to determine which keyword is appropriate from the displayed facet list of the tree structure. An example of a screen when the facet itself of “theme” 181 is designated as a keyword is shown. In FIG. 18, a document for each subordinate concept keyword belonging to the facet “theme” includes a list of information resources 183 in which the keyword “human science” and scoring keywords belonging to the subordinate concept are defined, the keyword “social science”, and the A list 184 of information resources in which scoring keywords belonging to the subordinate concepts are defined, and a list 185 of information resources in which the keyword “art” and scoring keywords belonging to the subordinate concepts are defined are displayed.

  FIG. 19 shows the bibliographic information of information resources displayed in the “Humanities” list, for example, the title in FIG. 18 and thinks that “It looks interesting because you can see a picture”, and the thesaurus keyword displayed in the upper right column. An example of a screen when the “humanities” 182 is clicked is shown. The clicked thesaurus keyword “Humanities” is input as a narrow-down condition thesaurus keyword, and information resources in which the keyword “Humanities” and scoring keywords belonging to its subordinate concepts are defined are displayed in a list.

  FIG. 20 shows the bibliographic information of information resources displayed in the “history” list 192 in FIG. 19, for example, the title, and it is displayed in the middle right, thinking that “the photo seems to be in history”. An example of a screen when the thesaurus keyword “history” 191 is clicked is shown. The clicked thesaurus keyword “history” is input as a narrow-down condition thesaurus keyword, and only information resources in which the keyword “history” and scoring keywords belonging to its subordinate concepts are defined are listed in the list 202. . In FIG. 20, the “related” list 203 includes information resources in which keywords that belong to the same facet “theme” as “history” are defined as scoring keywords, but do not match the filtering condition “history”. In addition, the “uncategorized” list 204 displays information resources that are not classified in either the list 202 or the list 203. In the “relevant” list 203 arranged in the same screen as the list 202, information resources that do not match the filtering condition but are related are displayed. For example, the information resource (scoring) displayed in the second item in the list 203 is displayed. "Hokkaido", "Geography", and "Edo period" are defined as thesaurus keywords. In the “Current score / restriction conditions” display column, “Theme”, “Humanities” and “History” on the first facet are refined, and “Geography”, “Japan” and “Hokkaido” are added on the second facet. Indicates the state specified as the thesaurus keyword for each.

  In FIG. 20, “history” is input as a refinement condition thesaurus keyword. However, in the lower part of the display screen of the information resource list 202 that conforms to the refinement condition thesaurus keyword, the relevant information resource that does not match is associated. Sorted in descending order of degree (similarity) and output to the “relevant” list 203. For this reason, even when there are no search results that match the narrow-down condition thesaurus keyword, a search error does not occur, and related information resources can always be grasped.

  FIG. 21 considers “I want to investigate environmental pollution” on the initial screen shown in FIG. 16, and selects “Environmentology” in the facet “theme” 211 in the facet selection input field in the upper left column as the scoring thesaurus keyword. An example of a screen when input is shown. The list 214 in the upper right column displays a list of information resources in which the keyword “environmental science” and scoring keywords belonging to its subordinate concepts are defined, together with the display 213 of the input category “theme> natural science> environmental science”. In the “relevant” list 215 in the middle right, there are information resources in which keywords that do not conform to the additional condition “environmental science” but that belong to the same facet “theme” as “environmentology” are defined as scoring keywords A list is displayed.

  FIG. 22 shows an example of a screen when the search free word “environmental pollution” is entered in the search character string column 221 in FIG. 21, considering that “the range is too wide just by specifying“ Environment ”on the thesaurus”. Indicates. The list 222 in the upper right column shows, as understood in comparison with the list 214 in FIG. 21, the score of the information resource in which the scoring free word “environmental pollution” is defined and information resources similar thereto. A score of 1 is added, and information resources that have been re-sorted in descending order of the combined score are displayed in a list. In general, keywords on the thesaurus are effective for specifying a relatively rough search condition, and free words are effective for specifying a specific search condition. Therefore, the search efficiency is improved by using both of the condition specifications as needed.

  FIG. 23 thinks that “I want to investigate environmental pollution other than Japan” in FIG. 22, and additionally inputs “overseas” in facet “geographic division” 231 as a thesaurus keyword for adding in the facet selection input field in the upper left column. An example of the screen in the case of having been done. It is difficult to specify a condition that leads to “overseas” by adding to a search character string for free word search. In this case, selection input from the facet selection input field is effective. The list 233 in the upper right column is an information resource in which the keyword “overseas” is defined and any keyword that belongs to the facet “geographic division” to which “overseas” belongs, as understood from the list 222 in FIG. Is added to the score of the defined information resource, and the information resources rearranged in descending order of the total score are displayed in a list. As understood from the list 233, information resources related to environmental pollution in China and Asia that are more suitable for the search purpose are displayed in a list. In addition, “Filter by resource type” is displayed at the top right of FIG. 23, and it is difficult to search by free word search, for example, “paper”, “researcher homepage”, “research data”, etc. Search conditions for a certain “resource type” facet can be specified directly.

  Alternatively, the node on the facet of the thesaurus system that is selected when the search condition is input and the node that becomes the reference when the output unit 4 performs display output may be independently specified. For example, when inputting the search conditions, if the “psychology” node of the “theme” facet is selected and input as an input thesaurus keyword, the search result is “psychology” and its You may categorize and display in a subordinate list, related list, or unclassified list, or you can select and input another node in the same “theme” facet independently, and the above classification based on this other node. You may display, or you may classify and display on the basis of nodes in different facets, such as by age, by location, by resource type.

<Hardware Configuration of Information Resource Retrieval Device According to this Embodiment>
FIG. 24 is a block diagram exemplarily showing a hardware configuration of the information resource search apparatus according to this embodiment. In the information resource search apparatus which is the computer apparatus 110 shown in FIG. 24, the CPU 111 controls the entire system by using the RAM 115 as a work memory for primary storage in accordance with programs stored in the ROM 114 and / or the hard disk drive 116. . Furthermore, the CPU 111 executes information resource search processing according to the present embodiment based on a program stored in the hard disk drive 116 in accordance with a user instruction input via the mouse 112 a or the keyboard 112. The display interface 113 is connected to a display such as a CRT or LCD, and displays an input standby screen for information resource search processing executed by the CPU 111, processing progress and search results, the contents of the information resource selected from the list, and the like. . The removable media drive 117 is mainly used when writing a file from the removable medium to the hard disk drive 116 or writing a file read from the hard disk drive 116 to the removable medium. Removable media include floppy disk (FD), CD-ROM, CD-R, CD-R / W, DVD-ROM, DVD-R, DVD-R / W, DVD-RAM and MO, memory card, CF Cards, smart media, SD cards, memory sticks, etc. can be used.

  A printer such as a laser beam printer or an ink jet printer is connected to the printer interface 118. The network interface 119 is an interface for connecting a computer device to a network.

  Note that the input device in the information resource search device according to each of the above embodiments is not limited to the mouse 112a or the keyboard 112, and any pointing device such as a trackball, a trackpad, or a tablet can be used as appropriate. When the portable information terminal is used as the information resource search device according to each of the above embodiments, the input unit may be configured with a button, a mode dial, or the like.

  In addition, the hardware configuration of the information resource search apparatus according to each of the embodiments shown in FIG. 24 is merely an example, and it is needless to say that any other hardware configuration can be used.

  In particular, all or part of the information resource search process according to each of the above embodiments may be realized by the computer terminal device 110 or a portable information terminal device such as a PDA. The Internet or any well-known local network connected via a wired communication line such as Bluetooth (registered trademark) wireless or the Internet (TCP / IP), public telephone network (PSTN), integrated service digital network (ISDN), etc. The information resource search process may be realized by a network system including an area network (LAN) or a wide area network (WAN). For example, the client apparatus includes an input unit 1 and an output unit 4 for information resource search request and browsing display, and a search condition input via the input unit 1 is set as a search execution unit 2, a data storage unit 3, and metadata. The server apparatus that implements the automatic generation unit 5 transmits the search result to the client apparatus having transmitted the search condition or other specified identifier, and causes the client apparatus to output the search result. May be.

    As described above, according to the present embodiment, the first score and the second score are added together, and the information resources are sorted by the obtained combined score (third score). The search results are displayed in the upper order in order, and the first score and / or the second score that is always maintained is updated by adding a new score each time a plurality of subsequent search condition inputs are performed, and updated. Based on the score, the search result list is updated and displayed. Therefore, even if the metadata that is bibliographic information assigned to information resources available on the web is incomplete or an inappropriate search keyword is input by an unskilled user, search is possible. A search result can always be obtained without impairing efficiency, and even when a large amount of information resources is obtained as a search result, it is possible to obtain a search result that is more suitable for the search condition.

  In addition, even if the input search keyword does not match the description on the metadata attached to the information resource, the input search keyword is related based on the distance on the facet tree structure of the thesaurus. A list of information resources is always displayed. For this reason, it is possible to reliably obtain related information resources that do not match the input search conditions, and based on the related information resources, it is possible to correct the initially input search conditions to more appropriate search conditions. .

  In addition, when search condition input is executed a plurality of times, each time the search condition is input, the score that is always held is updated by adding a new score, and based on the updated score, Refresh the list. For this reason, it is possible to reliably obtain a more accurate search result each time an additional search keyword is input without causing a search omission.

  Furthermore, a plurality of facet tree structures can be selectively displayed, and a plurality of keywords across a plurality of facets can be designated simultaneously by selecting and inputting an arbitrary node on the displayed facet tree structure. This makes it possible to simultaneously specify keywords belonging to multiple facets on the thesaurus as search criteria, eliminating the need to create new facets for specifying keywords across multiple facets, and creating a facet directory structure. It can be simplified. This improves the degree of matching in classification work and provides classification criteria that are easy for the user to understand. On the other hand, multi-faceted search conditions can be set, and the characteristics of concepts and information resources Because it is possible to describe with a combination of various characteristics, it is possible to flexibly cope with a new concept that was not assumed when the thesaurus was constructed.

  Therefore, the search efficiency of information resources on the web on the user side is greatly improved, so that the distribution of information resources on the web is promoted and the data availability is improved.

  The scope of the present invention is not limited to the illustrated and described exemplary embodiments, but includes all embodiments that provide the same effects as those intended by the present invention. For example, the present invention is not limited to information resources such as academic information resources available on the web, but can be easily applied to, for example, a personal file management system. Further, the scope of the present invention is not limited to the combination of features of the invention defined by claim 1 but can be defined by any desired combination of specific features among all the disclosed features. .

It is a block diagram which shows an example of a function structure of the information resource search apparatus which concerns on embodiment of this invention. It is a schematic diagram which shows an example of the data structure and layout of the metadata 31 of FIG. It is a flowchart which shows an example of the schematic process sequence of the information resource search process which the information resource search apparatus which concerns on embodiment of this invention performs. It is a flowchart which shows an example of the detailed process sequence of the search process in FIG.3 S5. FIG. 5 is a flowchart illustrating an example of a detailed processing procedure of determination processing for determining whether or not a document conforms to an input narrowing condition in step S <b> 51 of FIG. 4. FIG. 6 is a flowchart illustrating an example of a detailed processing procedure of determination processing for determining whether or not a document conforms to a facet narrowing condition in step S <b> 512 of FIG. 5. It is a flowchart which shows an example of the detailed process sequence of the scoring process in step S54 of FIG. It is a flowchart which shows an example of the detailed process sequence of the score calculation process of the document by the search free word in step S542 of FIG. It is a flowchart which shows an example of the detailed process sequence of the free word for scoring, and a registration process in the case of calculating the free word for scoring and its weight by a vector space model. It is a flowchart which shows an example of the detailed process sequence of the free word for scoring, and a registration process in the case of calculating the free word for scoring and its weight by RS model. It is a schematic diagram explaining the calculation and correction process of the document vector by RS model. It is a schematic diagram which shows an example of the facet on a thesaurus. FIG. 13 is a diagram for explaining an example of score calculation in the case where “Tokyo” is a keyword for addition in FIG. 12 and “A” and “C” are search free words. FIG. 12 is a diagram for explaining an example of score calculation when “Tokyo” is used as a keyword for addition, “A” and “C” are search free words, and “20th century” and “21st century” are narrowed down keywords. It is. FIG. 13 is a diagram for explaining an example of score calculation when “Tokyo”, “Osaka Prefecture”, “20th century” and “21st century” are used as narrowing keywords and “A” and “C” are used as search free words. . It is a figure which shows an example of the search initial screen displayed and output on the display of the computer which performs the information resource search program which concerns on embodiment of this invention. It is a figure which shows an example of the search result screen displayed and output on the display of the computer which performs the information resource search program which concerns on embodiment of this invention. It is a figure which shows an example of the search result screen displayed and output on the display of the computer which performs the information resource search program which concerns on embodiment of this invention. It is a figure which shows an example of the search result screen displayed and output on the display of the computer which performs the information resource search program which concerns on embodiment of this invention. It is a figure which shows an example of the search result screen displayed and output on the display of the computer which performs the information resource search program which concerns on embodiment of this invention. It is a figure which shows an example of the search result screen displayed and output on the display of the computer which performs the information resource search program which concerns on embodiment of this invention. It is a figure which shows an example of the search result screen displayed and output on the display of the computer which performs the information resource search program which concerns on embodiment of this invention. It is a figure which shows an example of the search result screen displayed and output on the display of the computer which performs the information resource search program which concerns on embodiment of this invention. It is a figure which shows an example of the hardware constitutions of the information resource search apparatus which concerns on each embodiment of this invention.

Explanation of symbols

Input part 1
Search execution part 2
Data storage unit 3
Output unit 4
Automatic metadata generation unit 5
Search free word input part 11
Adding point thesaurus keyword input part 12
Refinement condition thesaurus keyword input part 13
Attention keyword input part 14
Free word search scoring part 21
Keyword search scoring unit 22
Score totaling section 23
Sorting section 24
Filtering section 25
Metadata 31
Body / Content 32
Faceted display 41

Claims (17)

Metadata assigned to an information resource, the metadata including, for each information resource, a thesaurus keyword that describes a node name in a facet on a thesaurus, a free word described in the information resource, and its A metadata storage unit for storing metadata including importance pairs;
An input unit for inputting an input thesaurus keyword and an input free word for searching the information resource;
For an information resource that includes the free word that matches the input free word in the metadata, a first score is calculated based on the degree of importance paired with the matching free word, and the first score is retained A first score calculation unit
A second score is obtained for an information resource that includes in the metadata the thesaurus keyword that matches the input thesaurus keyword or the thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs. And calculating and multiplying the second score by a weight β (β ≦ 1) corresponding to the distance on the tree structure for the information resource including the thesaurus keyword belonging to the other node in the facet in the metadata. A second score calculator that holds the second score as a second score;
For each information resource, a score summation unit that sums the first score and the second score to obtain a third score;
An information resource search apparatus comprising: an output unit that displays a list of information resources sorted in descending order of the third score. The first score calculation unit holds the first score calculated by the input of the input free word when the input free word is input subsequent to the output of the search result by the output unit. Add to the first score to get the first score,
The information output device according to claim 1, wherein the output unit updates and displays the information resources sorted again in descending order of the third score. When the input thesaurus keyword is input subsequent to the output of the search result by the output unit, the second score calculation unit holds the second score calculated by inputting the input thesaurus keyword. Add to the second score to get the second score,
3. The information resource search apparatus according to claim 1, wherein the output unit updates and displays information resources sorted again in descending order of the third score. 4. The input unit according to claim 1, wherein the input unit selectively displays a plurality of facet tree structures, and the nodes on the displayed tree structures can be simultaneously selected. 5. Information resource retrieval device. The output unit includes an information resource including, in the metadata, the thesaurus keyword that matches the input thesaurus keyword, or a thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs. A first display output column for sorting and displaying a list in descending order of the third score;
And a second display output column for sorting and displaying information resources including thesaurus keywords belonging to other nodes in the facet in the metadata in the descending order of the third score. The information resource search device according to any one of claims 1 to 4. 6. The information resource search device according to claim 1, wherein the output unit substantially displays a list of information resources that can be displayed at a time within an effective display area of the display device. The information resource retrieval apparatus further includes:
A narrowing condition input unit for inputting a narrowing condition thesaurus keyword for narrowing down the search result of the information resource;
Only the information resources that include in the metadata the thesaurus keywords that match the input narrowing condition thesaurus keywords, or the thesaurus keywords that belong to the lower nodes of the matching thesaurus keywords on the facet tree structure to which the narrowing condition thesaurus keywords belong, The information resource search device according to claim 1, further comprising a filtering processing unit that sets the information resource to be output. The information resource retrieval apparatus further includes:
6. The information according to claim 5, further comprising: a third display output field for displaying a list of information resources that do not belong to any node in the facet, sorted in descending order of the third score. Resource search device. Metadata assigned to an information resource, the metadata including, for each information resource, a thesaurus keyword that describes a node name in a facet on a thesaurus, a free word described in the information resource, and its A metadata storage unit for storing metadata including importance pairs;
For an information resource that includes in the metadata the free word that matches the input free word for searching for the information resource, a first score is calculated based on the importance paired with the matching free word; A first score calculation unit for holding the first score;
An information resource that includes in the metadata the thesaurus keyword that matches the input thesaurus keyword for searching for the information resource, or the thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs For information resources that include the thesaurus keywords belonging to the other nodes in the facet in the metadata, the second score is weighted according to the tree structure distance β ( a second score that is multiplied by β ≦ 1) and holds the second score;
For each information resource, a score summation unit that sums the first score and the second score to obtain a third score;
An information resource search server device comprising: a sorting unit that obtains a list of information resources sorted in descending order of the third score as a search result to be output. An input free word for searching an information resource is input, a plurality of facet tree structures on the thesaurus are selectively displayed, and nodes on the displayed tree structure are selected, whereby an input thesaurus keyword is selected. An input section to input,
Based on the metadata assigned to the information resource, a score is calculated for the first information resource group that is a set of information resources having metadata matching the input free word or the input thesaurus keyword, and the input A processing unit that calculates a score for a second information resource group that is a set of information resources having metadata matching the thesaurus keyword belonging to another node in the facet to which the thesaurus keyword belongs;
An information resource search comprising: an output unit that displays a list of the first information resource group and the second information resource group that are sorted in descending order of score so as to be distinguishable on a display screen. Client device. A metadata storage unit that stores metadata including a thesaurus keyword that describes a node name in a facet on a thesaurus for each information resource;
An input unit for inputting an input thesaurus keyword for searching for the information resource;
A score is calculated for an information resource that includes in the metadata the thesaurus keyword that matches the input thesaurus keyword or the thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs. , For information resources that include thesaurus keywords belonging to other nodes in the facet in the metadata, multiply the score by a weight β (β ≦ 1) corresponding to the distance on the tree structure, and obtain the score. A score calculator to hold;
An information resource search apparatus comprising: an output unit that displays a list of information resources sorted in descending order of the score. Metadata assigned to an information resource, the metadata including, for each information resource, a thesaurus keyword that describes a node name in a facet on a thesaurus, a free word described in the information resource, and its Storing metadata including importance pairs in a metadata storage unit;
Inputting an input thesaurus keyword and an input free word for searching the information resource;
For an information resource that includes the free word that matches the input free word in the metadata, a first score is calculated based on the degree of importance paired with the matching free word, and the first score is retained And steps to
A second score is obtained for an information resource that includes in the metadata the thesaurus keyword that matches the input thesaurus keyword or the thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs. And calculating and multiplying the second score by a weight β (β ≦ 1) corresponding to the distance on the tree structure for the information resource including the thesaurus keyword belonging to the other node in the facet in the metadata. A second score, and retaining the second score;
For each information resource, adding the first score and the second score to obtain a third score;
And displaying a list of information resources sorted in descending order of the third score as a search result. Metadata assigned to an information resource, the metadata including, for each information resource, a thesaurus keyword that describes a node name in a facet on a thesaurus, a free word described in the information resource, and its Storing metadata including importance pairs in a metadata storage unit;
For an information resource that includes in the metadata the free word that matches the input free word for searching for the information resource, a first score is calculated based on the importance paired with the matching free word; Holding the first score;
An information resource that includes in the metadata the thesaurus keyword that matches the input thesaurus keyword for searching for the information resource, or the thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs For information resources that include the thesaurus keywords belonging to the other nodes in the facet in the metadata, the second score is weighted according to the tree structure distance β ( multiplying β ≦ 1) to obtain a second score, and holding the second score;
For each information resource, adding the first score and the second score to obtain a third score;
Obtaining a list of information resources sorted in descending order of the third score as a search result to be output. An input free word for searching an information resource is input, a plurality of facet tree structures on the thesaurus are selectively displayed, and nodes on the displayed tree structure are selected, whereby an input thesaurus keyword is selected. Step to enter,
Based on the metadata assigned to the information resource, a score is calculated for the first information resource group that is a set of information resources having metadata matching the input free word or the input thesaurus keyword, and the input Calculating a score for a second information resource group that is a set of information resources having metadata that matches the thesaurus keyword belonging to another node in the facet to which the thesaurus keyword belongs;
A method of searching for an information resource, comprising: displaying a list of the first information resource group and the second information resource group that are sorted in descending order of score so as to be distinguishable on a display screen. An information resource search program for causing a computer to execute an information resource search process, the program causing the computer to
Metadata assigned to an information resource, the metadata including, for each information resource, a thesaurus keyword that describes a node name in a facet on a thesaurus, a free word described in the information resource, and its Processing to store metadata including the importance pair in the metadata storage unit;
A process of inputting an input thesaurus keyword and an input free word for searching the information resource;
For an information resource that includes the free word that matches the input free word in the metadata, a first score is calculated based on the importance that is paired with the matching free word, and the first score is retained Processing to
A second score is obtained for an information resource that includes in the metadata the thesaurus keyword that matches the input thesaurus keyword or the thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs. And calculating and multiplying the second score by a weight β (β ≦ 1) corresponding to the distance on the tree structure for the information resource including the thesaurus keyword belonging to other nodes in the facet in the metadata. A process of setting the second score and holding the second score;
For each information resource, a process of adding the first score and the second score to obtain a third score;
An information resource search program for executing a process including a process of displaying a list of information resources sorted in descending order of the third score as a search result. An information resource search program for causing a computer to execute an information resource search process, the program causing the computer to
Metadata assigned to an information resource, the metadata including, for each information resource, a thesaurus keyword that describes a node name in a facet on a thesaurus, a free word described in the information resource, and its Processing to store metadata including the importance pair in the metadata storage unit;
For an information resource that includes in the metadata the free word that matches the input free word for searching for the information resource, a first score is calculated based on the importance paired with the matching free word; Processing to hold the first score;
An information resource that includes in the metadata the thesaurus keyword that matches the input thesaurus keyword for searching for the information resource, or the thesaurus keyword that belongs to a lower node of the matching thesaurus keyword on the facet tree structure to which the input thesaurus keyword belongs For information resources that include the thesaurus keywords belonging to the other nodes in the facet in the metadata, the second score is weighted according to the tree structure distance β ( a process of multiplying β ≦ 1) to obtain a second score, and holding the second score;
For each information resource, a process of adding the first score and the second score to obtain a third score;
An information resource search program for executing a process including a process of obtaining a list of information resources sorted in descending order of the third score as a search result to be output. An information resource search program for causing a computer to execute an information resource search process, the program causing the computer to
An input free word for searching an information resource is input, a plurality of facet tree structures on the thesaurus are selectively displayed, and nodes on the displayed tree structure are selected, whereby an input thesaurus keyword is selected. Process to enter,
Based on the metadata assigned to the information resource, a score is calculated for the first information resource group that is a set of information resources having metadata matching the input free word or the input thesaurus keyword, and the input A process of calculating a score for a second information resource group that is a set of information resources having metadata matching the thesaurus keyword belonging to another node in the facet to which the thesaurus keyword belongs;
It is for executing a process including a process of displaying the first information resource group and the second information resource group, which are sorted in descending order of scores, in a list so as to be distinguishable on a display screen. An information resource retrieval program characterized by