JP4084647B2 - Information search system, information search method, and information search program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a system for handling data to which metadata is attached (called a systematization system), and relates to a technique for giving metadata, a document and database search method, a search system, or a computer program for search. .
[0002]
[Prior art]
There are the following technologies related to the technology related to the addition of metadata, or the search method and search system of documents and databases, or the computer program for search (see, for example, Patent Document 1 and Patent Document 2).
[0003]
In Patent Document 1, “basic management information for collecting information of a database system connected to a network is registered in advance in an information resource management dictionary that manages metadata of the entire database system. "Technology that collects metadata from individual database system metadata and individual information resource tools at the timing specified by the basic information for collection, and reflects the collected metadata in the information resource management dictionary" Has been.
[0004]
Patent Document 2 states that “a keyword shaper refers to synonyms / unnecessary words and generates document word pairs and document metadata from document data. A keyword relationship extractor refers to a hierarchical relationship, Hierarchical relations, equivalence relations, and associative relations are extracted, and the directory file generator generates a directory file from the top keywords, hierarchical relations, equivalence relations, associative relations, and document metadata. "Technology for accessing directory files".
[0005]
[Patent Document 1]
JP-A-10-91503 (for example, paragraph numbers 0010 and 0014, FIG. 1 and the like)
[Patent Document 2]
JP 2000-10996 A (for example, 0039, 0136, FIG. 1, FIG. 2, etc.)
[0006]
[Problems to be solved by the invention]
  A system that is easier to use in search processing,MethodAnd programsIs desired. For example, the present invention organizes information with conceptual information (hereinafter referred to as metadata), uses the metadata for searching,Improve the operability of adding metadata to documents as information categorization work for search processing, and improve the operability of search processing in search methods using metadata labels, etc., or Information search system, information search method and information search for improving search accuracyThe purpose is to provide a program.
[0007]
[Means for Solving the Problems]
  To solve the problems in the present inventionFirstMeans ofThe first similarity information indicating the similarity between all metadata values stored in the database and the input information, and the existing information stored in the database that is similar to the input information The metadata value suitable for the input information is determined based on the second similarity information indicating the similarity between the metadata value and the input information.
  Furthermore, the second means for solving the problems in the present invention is as follows:Based on the search information for searching the search target such as general documents and databases, a plurality of metadata as described in the following embodiment is extracted and displayed as a label on the screen, and a specific label is selected from these labels. By specifying, the search result is output based on the specified label. In doing so, it is easier to understand the relationship with the metadata labels that are related to each other.OctopusIt is.
[0008]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments according to the present invention will be described. FIG. 1 shows the configuration of the information search system. This system includes a plurality of operation devices 102 and a search processing device (for example, configured by a server) 101 that executes a search based on a search request from these operation devices 102. The operation device 102 is a device for a user to perform a search operation, and each operation device 102 has a keyboard and mouse as an input device, a display as an output device, and printing means. A client as a user uses this search processing apparatus 101 to register a document. The metadata that should be given to the document is automatically given by the system. The metadata itself is registered by the system administrator at the time of system construction, but can be manually registered even after the system construction with the permission of the system administrator.
[0009]
Next, a procedure for inputting metadata at the time of search (designation by a label in this embodiment) and displaying the search result on the screen will be described. The search processing apparatus 101 includes a document and metadata storage unit, a metadata reception unit, and a search result output unit. Further, a database (hereinafter referred to as a metadata DB) 103 for metadata and a document And a database (document metadata DB) 105 relating to a combination of documents and metadata. The metadata DB 103, the document DB 104, and the document metadata DB 105 are databases used mainly for executing search processing.
[0010]
FIG. 2 is an operation explanatory diagram showing the relationship between the operation of the system and the use of the database when the information search system performs a search process, for example, when an information search process using metadata is performed. In practice, the system executes a program and performs the operations described below. However, a system that virtually achieves the processing purpose is constructed, and the system is described as having a configuration for processing. However, as described above, the system has a function and only performs a processing operation, and does not actually have a dedicated configuration as described below. The following dedicated units are operations or functions. FIG. 2 shows one processing procedure example of the present invention. First, in step S200 of FIG. 2, at the time of system construction, a metadata administrator creates a metadata name representing a metadata category and a metadata value belonging to the metadata name. The created metadata name and metadata value are registered in the metadata DB 103. Further, when creating metadata, it is not necessary for the administrator to set the upper and lower concepts of the metadata. Therefore, the burden on the metadata creator can be greatly reduced. In addition, since there is no fixed directory structure (that is, the upper metadata for any metadata is not limited to one), the directory structure is not different from the hierarchical relationship of the concept that the searcher has individually. It is also possible to create metadata that is a superordinate concept with the metadata name as the metadata value and to provide a hierarchical structure between the metadata. In this case, there is no need to fix the upper metadata name to one, and a plurality of metadata names can be freely set. Details of this content will be described with reference to FIG.
[0011]
Next, the flow of the entire system will be described using the flow of FIG. First, in step S201, the client inputs text from the input unit of the operation apparatus 102, and the server stores the document input and registered from the client in the document DB. At this time, in step S204, metadata to be added to the document by the metadata adding unit is associated with the document and registered in the document metadata DB. Note that the metadata to be registered in the document metadata DB is obtained from the metadata DB by the corresponding metadata extraction unit in step S202. In step S203, the system automatically specifies metadata to be added to the document by the corresponding metadata specifying means. At this time, it is also possible to specify metadata to be individually assigned by the system administrator without relying on the automatic processing of the system. Details of document registration and metadata assignment will be described later.
[0012]
At the time of document retrieval, the server returns the metadata to the operation device 102 of the client by the operation of the connection request from the client in the metadata presenting means in step S220. Next, in step S209, the client performs an operation of selecting metadata in the metadata input unit, and in step S210, the server identifies the document having the selected metadata by the metadata corresponding document search unit. In step S214, the metadata corresponding document output unit returns the result to the operation device 102 of the client.
[0013]
The processing contents of the information retrieval system can be broadly divided into accumulation processing, retrieval processing, and output processing. Processing at the time of storage and search is performed by the operation device 102 and the search processing device 101, the metadata DB 103, the document DB 104, and the document metadata DB 105. The automatic processing of the metadata at the time of document accumulation and the search processing are performed by the search processing device 101 as a server, and the document registration, the search key designation and the search command are mainly performed by the operation device 102 of the client. The search result output process is mainly performed by the operation device 102.
[0014]
FIG. 3 shows a configuration example of the category directory. The category here means a group of the same type of metadata, and the directory is a data structure of a multi-branch structure in which upper metadata is uniquely determined. As shown in the figure, a plurality of category directory structures usually exist for one category. Therefore, it is difficult to map the created category to a directory (hierarchy) as compared to creating a document category. For example, the category of the basic design document for product A can be “product A” as the product category and “basic design document” as the document category. When the category is classified into a directory, which of 311 and 312 should be classified, that is, which of “product” 301 and “document” 302 should be classified as a higher-level directory. It varies depending on the documents stored in and the usage method. For example, when searching for a document for a certain product A, a directory structure such as 311 in FIG. 3 is desirable. However, when searching for a sample of a basic design document, a directory structure such as 312 in FIG. 3 is desirable. Therefore, in one embodiment of the present invention, the category is not regarded as a directory, but the category information is held as document metadata, and the document is organized using the metadata. Here, the metadata of this system includes categories, that is, metadata names (for example, “product” 301, “document” 302, etc.) and information of each category, that is, metadata values (for example, “product A”, “basic design document”, etc. ). Further, as will be described later, an interface capable of effectively handling information (document or the like) to which metadata is added is provided.
[0015]
FIG. 4 shows an example of metadata. In the embodiment of the present invention, it is assumed that metadata as shown in FIG. 4 is handled. Here, the metadata includes a metadata name representing a category (for example, “product name” in 401 in FIG. 4) and a metadata value representing a type in the category of the metadata name (for example, “product in 401 in FIG. 4”). A ”,“ Product B ”,“ Product C ”,. The same applies to 402 and 403 in FIG.
[0016]
FIG. 5 shows an information holding method regarding metadata. FIG. 5 shows an example of document metadata and an example in which the document is embedded in the metadata. In this example, metadata is attached to a document in a format such as XML. As shown at 511 to 513 in FIG. 5, it is possible to represent document metadata in a form including a link (URL, full path, etc.) to the document. When the document is in an XML or HTML format, it can be expressed in an embedded form as indicated by reference numeral 514 in FIG. Further, as in 514 of FIG. 5, a plurality of metadata values (for example, the basic design document 502 and the detailed design document 503) can be assigned to one category (for example, “document type”). By this method, a document related to a plurality of directories (for example, a document related to the product B, in which the basic design document 502 and the detailed design document 503 are combined with the metadata document type) can be expressed. Therefore, in the case of directory storage, the problem that it cannot be determined in which directory such a document should be stored can be easily solved by the method of this embodiment. As described above, the problem that a document having a plurality of concepts in the same hierarchy cannot be easily stored in a folder having a directory structure whose upper concept is uniquely determined can be solved by the method of this embodiment. Therefore, by using the method of this embodiment, it is possible to efficiently carry out such work relating to document storage without being bothered.
[0017]
The metadata providing agent, which is a process at the time of accumulation, will be described with reference to FIGS. 6 to 15. In the following, the alphanumeric character “S” means a step that is a system operation procedure. First, FIG. 6 shows a flowchart of automatic assignment of metadata by the system. First, in step S <b> 601, a new document to which metadata is added is input from the operation apparatus 102. Details will be described later on the new document registration screen (FIG. 13). Next, processing regarding each category (for example, type, product name, etc.) of metadata is performed. In S602, a category to be processed is selected. In S603, first, processing is performed for each metadata of one category. In the process, for example, the similarity between the metadata and the new document is acquired, and the metadata and the similarity are stored. Subsequently, the stored metadata is sorted in descending order of similarity. This is A. Any method may be used as a method for obtaining the similarity. In step S604, it is checked whether the above-described metadata sorting processing for all categories scheduled for processing is completed. If not completed, the process returns to S602 to continue the sort process. In this process, metadata suitable for a new document is selected from metadata in a certain category. For example, if many words “production plan ABC” appear in the new document, it is determined that the product name metadata “production plan ABC” is appropriate for the new document. When the series of processes of S602 and S603 is completed, the process proceeds to S605.
[0018]
In step S605, a document similar to the input document is searched from the accumulated documents. Next, in S606, the following processing is performed on each retrieved document. This processing is an example. First of all, the metadata given to the retrieved document is acquired. Secondly, as a process for each acquired metadata, the similarity between the acquired metadata and the input document is stored. Next, the stored metadata is sorted in the descending order of similarity by the sum of scores (similarities expressed quantitatively by numerical values). This is B. In step S607, it is checked whether the sorting process in step S606 has been completed for the top N documents in the search result. If not completed, the process returns to S606 and continues. If completed, the process proceeds to S608. In the above processing, metadata of a document similar to the new document is selected. For example, in N documents similar to the new document, when a lot of metadata “basic design document” appears (for example, when it exceeds the set value), the type of metadata “basic design document” Judge that the metadata is appropriate for the document.
[0019]
In S608, the following processing is performed for each category of metadata. The metadata is sorted in ascending order of the sum of the ranks of A and B in the previous operation, and this is set as the estimated category. It is also possible to adjust the weights of A and B by multiplying A or B by a coefficient. That is, the estimated categories are provided in descending order of similarity score. In the system, A and B are combined to determine metadata suitable for the new document. If processing for all categories has not been completed, step S608 is repeated. When all categories have been processed, that is, if YES in S609, the process returns to the document list screen (FIG. 11). The document list screen (FIG. 11) will be described later.
[0020]
FIG. 7 shows a screen transition diagram related to the agent function for adding metadata. First, the user who performed the login operation is specified on the user authentication screen 701. Next, a transition is made to a library search screen 702 for individuals, where a search for inputting library conditions for individuals is performed. A screen 703 showing the search result of the personal library is displayed. On this screen, for example, a link to a library of people to be referred to is selected from a list of searched people. Next, the screen transitions to a document list display screen 704 which is a selection result screen. From the document list screen 704, the screen transitions to a screen 705 representing document contents, a screen 706 for newly registering a document, and a screen 707 for changing metadata. A document concept (metadata), a table of contents, and the like are displayed on a screen 705 representing the document contents. On the screen 706 for newly registering a document, a specified document can be newly registered. On the screen 707 for changing metadata, the metadata of the registered document can be changed. On the metadata change screen 707, for example, a detail screen 708 for changing metadata for specifying a product name related to “document” which is one of metadata is displayed, and a product name is selected from the screen. it can.
[0021]
FIG. 8 shows an example of the user authentication screen. The user authentication screen includes an input field 802 for a user's individual number (hereinafter referred to as ID), a password input field 803, a display for operating "login" (for example, a display button) 804, and a display for deleting input (for example, a "clear" button) ) 805. The operation procedure is as follows. First, the user inputs his / her user ID and password to the operation device 102 and operates (eg, clicks) the “login” button. Then, the screen changes to a personal library search screen (FIG. 9). On this screen (FIG. 8), a user who uses the system can be identified. You can also manage password security. When returning from the personal library search screen (FIG. 9) and performing the login process again, the previous login user is automatically logged out and can be logged in with the user designated this time.
[0022]
FIG. 9 shows an example 901 of a personal library search screen. The library search screen includes a field 902 for selecting or inputting a department name, a field 903 for inputting a keyword, a field 904 for inputting an update date, a “search” button 905 for executing a search, and an input “Clear” button 906 for deleting the content that has been set. The operation procedure is as follows. First, the user inputs or selects one or more search conditions (department, keyword and update date) in the above field, and executes the search, that is, clicks the “Search” button. Then, the screen changes to a personal library search result screen (FIG. 10). In the example 901 of the personal library search condition input screen, it is possible to narrow down the documents that the user wants to see from the viewpoints of department, keyword, and update date from the accumulated document group. The keyword can specify a document subject, a document type, a product name, a creator name, and the like. In this case, if the metadata category increases, the category value can also be specified.
[0023]
FIG. 10 is a screen showing an example of the search result of the personal library. The search result screen 1001 of the personal library displays the search condition so that the search condition can be seen by displaying the same search area as the search screen of the personal library described in FIG. . This search area includes a department selection field 1002, a keyword input field 1003, an update date input field 1004, an operation button 1005 for executing a re-search, or a “clear” button for performing an operation for deleting input contents. 1006 is provided. The operation procedure is the same as the personal library search operation method described with reference to FIG. If the contents of the department selection field 1002, the keyword input field 1003, and the update date input field 1004 are changed and the “re-search” button 1005 is clicked, the search result of the newly changed search condition is displayed in the search result display area. Can be reflected. The personal library search result screen (FIG. 10) is provided with a name column 1007, an affiliation column 1008, and a link column 1009 to the personal library in the column direction as display areas for the search results. A plurality of such data are displayed in the row direction. The operation procedure of the search result display area is as follows. First, the user clicks on a link 1010 for one personal library in the name column. Then, the screen changes to a document list screen (FIG. 11). Although the display for returning is omitted in FIG. 11, it is possible to return to FIG. 10 by performing the returning operation. As described above, the contents of the selection fields and input fields 1002 to 1004 are displayed. By changing and clicking the “re-search” button 1005, the search can be performed again and again. On this screen, it is possible to narrow down the documents that the user wants to see from the viewpoints of department, keyword, and update date from the accumulated document group, and to refer to a list of information such as document registrants. Note that the name in the name column 1007 is the name of the document registrant.
[0024]
FIG. 11 shows an example 1101 of a document list screen. The document list screen includes, for example, an agent's utterance 1111, an explanation of who is registered, a new document registration 1102, change of document information 1103, and a link to usage help 1104. The document list includes a document subject column 1106, a document type column 1107, a product name column 1108, and a creator name column 1109 in the column direction, and a plurality of these data are displayed in the row direction. As shown at 1110, information on a creator different from the registrant (logged-in person in this embodiment) can be referred to. The operation procedure of the document list screen is as follows. For example, the user clicks one document subject “basic design document (overall)” 1112 in the document subject column 1106. Then, the screen changes to a screen (FIG. 12) for displaying the document contents. On the document content screen (FIG. 12), information such as the type, product name, author, and text related to the selected document subject, for example, “Basic Design Document (Overall)” is displayed. In the function of the document list screen (FIG. 11), when the user at the time of login does not match the registrant of the personal library, the agent utters a screen for newly registering a document or changing metadata. Do not display a link to the screen. For this reason, it is possible to prevent a document or metadata registered by another person from being changed. In the example of FIG. 11, the AA that is “A list of documents registered by Mr. AA” in the utterance 1111 of the agent is the name of the registrant. The creator name 1109 is a document creator (author), and is distinguished from the document registrant. Also, on this screen, a list of documents registered by one user and metadata of each document can be referred to. Furthermore, since the agent provides guidance on operations that are currently possible according to the type of screen, the user is less confused about the operations.
[0025]
FIG. 12 shows an example 1201 of a document content screen. The document content screen is composed of an agent utterance 1206 and document content 1205. The agent utterance 1206 includes metadata 1202, a link 1203 to a document list, and a link 1204 to help on usage, and may be displayed in other formats. The operation procedure is as follows. The user clicks a link 1203 to the document list screen in the utterance of the agent. Then, the screen changes to a document list screen (FIG. 11). On this screen, the contents (title, table of contents, etc.) and metadata (type, product name, creator, etc.) of one document can be referenced.
[0026]
FIG. 13 shows an example 1301 of a screen for newly registering a document. The new document registration screen includes an agent utterance 1306, a document file path input field 1304, and a “register” button 1305. The utterance 1306 of the agent includes a link 1302 to a document list and a link 1303 to help on how to use, in addition to the guidance “Register new document”. It may be displayed in other formats. The operation procedure is as follows. First, the user inputs the document file path in the input field 1304 and clicks a “register” button 1305. Then, the agent performs the following processing described in the flowchart of FIG. In the first step, the text is extracted from the document file. In the second step, metadata included in the text of the document is searched, and a weight is added according to the number of appearances for each metadata. In a third step, a stored document whose text is similar to the document is searched. In the fourth step, for each metadata that has been assigned to each document of the search result, a weight is added according to the similarity of each document. In the fifth step, the metadata having the largest weight for each metadata type is assigned to the newly registered document. The document and metadata assignment information is registered in the system. When the link 1302 to the document list is selected, the screen changes to the document list screen (FIG. 11). At this time, the metadata of the newly registered document is reflected on the document list screen (FIG. 11). With the function of the new document registration screen (FIG. 13), the user's client PC, that is, the document in the operation device 102 can be registered in the system. Furthermore, metadata that the agent determines to be most appropriate can be automatically assigned to the document. This eliminates the need for the user to manually add metadata to the document, greatly reducing the labor involved in registration work.
[0027]
FIG. 14 shows an example screen 1401 for changing metadata. The metadata change screen includes an agent utterance 1412 and information for changing metadata. The utterance 1412 of the agent includes a link 1402 to a document list, a link 1403 to a new registration screen for a document, and a link 1404 to help on how to use, in addition to the guidance “change text information”. It may be displayed in other formats. Information for changing metadata includes a document subject input field 1406, a document type selection / input field 1407, a product name selection / input field 1408, and a creator name selection / input field 1409. The operation procedure for changing the metadata is as follows. First, the user changes one metadata in the document metadata (document subject, document type, product name and creator name). For example, assume that the product name 1414 in the proposal material 1413 of XX Corporation is changed. The metadata candidates are metadata having a large weight (that is, having high similarity and suitable as metadata) among the metadata acquired by the agent on the new document registration screen. Since the metadata candidates are displayed in the list 1410, an arbitrary product name is selected from the list 1410 and set to the product name 1414. Metadata candidates are displayed in the list 1410 in descending order of weight. As an operation for selecting a metadata candidate, for example, click Other ... 1411 from the list 1410. Then, a metadata change detail screen (FIG. 15) is displayed. On the metadata change details screen (FIG. 15), metadata candidates are presented for each document from the existing metadata group. Details will be described later. Thereby, it is possible to reduce the time and effort for the user to add metadata to the document. In the agent utterance 1412 area, for example, the user clicks a link to the document list screen. Then, a transition is made to the document list screen (FIG. 11). At this time, the metadata changed on the metadata change screen (FIG. 14) is reflected on the document list screen (FIG. 11). If a link 1403 for registering a new document in the agent's utterance is selected, the screen changes to a new document registration screen (FIG. 13).
[0028]
FIG. 15 shows an example 1501 of a detailed screen for changing metadata. The metadata change details screen is composed of a list of metadata values. The operation procedure is as follows. First, on the screen for changing metadata (FIG. 14), for example, it is assumed that the subject sets the product name in the row of the proposal material 1413 of XX Corporation. At that time, one of the product name input fields, for example, a combo box is selected to display a list 1410 of product names. For example, “Others” 1411 is selected. Then, a detailed screen 1501 for changing metadata is displayed. From the screen, one piece of metadata, for example, “HH system” 1502 is clicked. Then, the screen returns again to the metadata change screen (FIG. 14). At this time, the metadata selected by the user, that is, the “HH system” is reflected in the field 1414. As described above, since “HH system” is selected in the example, “HH system” is newly input in the column 1414 of the metadata change screen (FIG. 14). In this embodiment, an example is shown in which an existing metadata group is displayed, and the user selects appropriate metadata from the group and assigns it to a document. As a result, the user can save time and effort to think of appropriate metadata even when changing the metadata to be added to the document. Further, as metadata, there is an effect of preventing the mixing of synonyms and different expressions and reducing the noise of the metadata. For example, if the metadata is selected only from the list in the above example, it can be unified with the “HH system”. On the other hand, if the input to the input area is performed manually, there is a high risk that different metadata will be added to indicate the same item such as “HHSYSTEM”, “HH-system”, and “HH system”. By displaying the list in this way and selecting metadata from the list, it is possible to reduce the noise of the metadata. The processing at the time of accumulation has been described above. In the present embodiment, in addition to the embodiment (embodiment embodiment), the implementation effect of the method for assigning metadata has been described.
[0029]
In FIG. 16 and subsequent figures, processing at the time of search and output will be described. In the following, an embodiment of a system that effectively searches for documents to which metadata is attached will be described. In this embodiment, in order to effectively handle a document with metadata, a screen (called a metadata map) in which labels representing metadata names are distributed on the screen is used as shown in FIG. . Here, it is assumed that the documents 511 to 514 in FIG. 5 and document metadata are handled as metadata. First, 1611 in FIG. 16 is an initial state of the metadata map. This metadata map has a meaning as a visual search screen that allows visual search. In the metadata map, in the initial state, a metadata name which is a higher-level concept representing the metadata category, for example, the highest concept, is displayed. In this example, there are three displays, but one display is also possible. However, the use effect of the present invention is larger when two or more. The user selects a metadata name representing a category to be narrowed down by document search from these metadata names (clicks a label), and starts document search. For example, taking FIG. 3 as an example, the highest metadata name is “product” and “document”. Even if the components of the directory are the same, the configuration (directory configuration) also changes if the classification criteria change, so that a plurality of top-level metadata can exist for one category. For example, if there are 10 configurations (directory configurations) for one category, a maximum of 10 metadata names are displayed on the initial screen (1611 in FIG. 16). A case will be described in which this initial screen is displayed using a directory instead of a data map. The method of the present invention can take a plurality of configurations for one category. For example, it is assumed that ten directory configurations can be taken for one category. In that case, 10 directories are displayed. Therefore, if it is assumed that one directory is displayed in one window, ten windows are prepared. Compared to this, the method of this embodiment for displaying the topmost metadata name in the initial state has few windows, for example, only one, and is very easy to see and very convenient.
[0030]
1612 in FIG. 16 is an example of a metadata map showing a state in which the label “product” as the metadata name is clicked, that is, selected from the state 1611 in FIG. When the metadata name “product” label is clicked, the value of the metadata belonging to the metadata name is displayed. Here, metadata values of the metadata name “product” in 1612 of FIG. 16 are “product A”, “product B”, and “product C”. This metadata value label is displayed around the display position of the metadata name so as to prevent overlap, and the metadata name and the metadata value are connected by a line so that the relation can be understood. . If there are many metadata values, the number of documents to which metadata is assigned, the selected metadata and related metadata are evaluated as indicators, the importance of metadata is determined, Display metadata values on the metadata map in order of importance. Other metadata values with low importance may be combined into one label as a label such as “others” and displayed on the metadata map. In this case, when the “other” label is selected, other metadata values are displayed in a list format, for example, and the metadata value can be selected therefrom. In addition, metadata that can be grouped can be defined with a group name representing them as one label. Therefore, when the group name is selected, metadata in the group is displayed in a list format, and a metadata value can be selected therefrom. Further, by re-clicking the selected label, the label of the selected “metadata name” can be brought into a non-selected state. In that case, the label color of the “metadata value” can be made lighter, the label size can be reduced, or the label itself can be hidden. When the label itself is hidden, it can be displayed again by selecting the label of the superordinate concept again.
[0031]
Next, the user selects a corresponding label from among the labels representing the displayed metadata values. By this operation, documents having the selected metadata value are narrowed down. When a plurality of metadata values are selected, narrowing down is performed using an OR condition or an AND condition. At this time, the OR condition and the AND condition can be set by a pop-up menu or environment setting. Alternatively, in the process of selecting a metadata value, the conditional expression can be displayed on a separate window or the like as a conditional expression display screen at the same time so that the conditional expression can be changed. As a specific example, when AND is set in the environment setting or the like, for example, when A, B, or C is selected, “A AND B AND C” is displayed on the conditional expression display screen. At this time, the AND expression can be changed by a click operation or the like. It is also possible to insert parentheses by dragging parenthesis items or the like into this conditional expression. In addition, by clicking again on the selected label, it is possible to deselect the value of the selected metadata and narrow down the selection again. It is also possible to deselect labels that you do not want to select (for example, while holding down the control key).
[0032]
1613 in FIG. 16 is an output example of a metadata map (that is, a visual search screen) when the metadata value “product A” is clicked in 1612 of FIG. At this time, the metadata value of “product A” is assigned to document 1, document 2, and document 3 of 511 to 513 in FIG. 5, as can be seen from FIG. 5. Therefore, document 1, document 2, and document 3 are search results. Further, the document 4 (514 in FIG. 5) that does not have the metadata of “Product A” is not extracted as a search result. Although details will be described later, examples of extraction of search results are as indicated by 1701 in FIGS. When the label of “Product A” is clicked, the label of “Product A” selected by, for example, displaying the color of other labels in the same category in a light color is highlighted. Depending on the operation mode, other labels in the same category that are not selected can be made smaller or hidden. By this operation, it is possible to narrow down the information and to make it easy to see the screen by taking into consideration that the information that needs to be viewed at the same time can be viewed without scrolling as much as possible.
[0033]
1614 of FIG. 16 shows an example of a metadata map when the label of the metadata name “document” is clicked in 1613 of FIG. The value of the metadata belonging to the metadata name of “Document” is displayed in the same manner as when the “Product” label is clicked. Here, the metadata values belonging to the document are “basic design document”, “detailed design document”, and “test”. At this time, as shown in 511 to 513 in FIG. 5, in addition to the metadata of “Product A” selected in 1613 in FIG. ”Metadata. Therefore, as shown at 1614 in FIG. 16, these labels can be connected by lines 1621 to 1622 to emphasize the relationship between the metadata values. In addition to connecting related labels with lines as methods for emphasizing the degree of association between metadata values, "Representation of degree of association by background color, tone, and shading of labels", "Relationship by giving identification marks to labels "Representation of degree", "Representation of degree of association by label border type and line width", "Expression of degree of association by label shape and size", "Expression of degree of association by character font", " “Relationship expression based on distance difference” or the like may be used. Thus, it is desirable to display the relevance or degree of relevance by a visual method.
[0034]
1615 of FIG. 16 shows a display example of the metadata map when the label “basic design document” which is the metadata value of the metadata name “document” is selected in 1614 of FIG. On the metadata map, “Product A” and “Basic Design Document” are designated. In the example of FIG. 5, there are two documents, 511 document 1 and 512 document 2 that have metadata of “product A” and “basic design document”. Therefore, document 1 and document 2 are search results. An example of the search result extraction is as shown in 1701 in FIGS.
[0035]
According to the method of the present embodiment, it becomes easy to narrow down a target document by selecting a label representing metadata displayed in the metadata map in this way. In addition, the metadata name to be selected can be selected from “Product”, “Document”, and “Delivery destination” as desired. Therefore, when there are a plurality of classification criteria as shown in 311 and 312 of FIG. 3, it is troublesome that the directory structure desired by the searcher must be searched by turning a plurality of screens or the like. It is possible to avoid work. For example, in the present embodiment, there are three directory configurations of “product”, “document”, and “delivery destination”. For that purpose, if one directory can be displayed in one window, it is necessary to prepare three windows and turn them. If there are 100 directories, 100 pages will be turned. Furthermore, in the process of searching for a document from among them, there arises a problem that the association with other directories cannot be grasped. In this embodiment, all these problems can be solved. First, since only the metadata label at the top of the directory is displayed on the initial screen, even if the directory structure is 100, in the initial state, one appropriate one of 100 labels in one screen is selected. Just search. Furthermore, since the metadata labels that are related to each other on the metadata map are connected by a line in the form of a network, the relationship with the metadata of other directories that are of interest can be easily grasped as described above. Furthermore, since it is possible to save which label is selected in the metadata map, the previous state is reproduced when the system of this embodiment is started after restarting the computer. It is also possible.
[0036]
FIG. 17 shows an example of displaying a document including the selected metadata. A metadata map 1615 is shown in the left area of the figure, a label (metadata value) selected in the upper area 1702, and a list of documents having the selected metadata value in the right area 1701. In the area where the metadata map 1615 is displayed, the screen transition described in FIG. 16 is displayed based on the procedure of the search operation, and the processing described in FIG. 16 is performed. The metadata value of the upper area 1702 is created based on the operation procedure, that is, the label selection order. In order to simplify the information, the number of hierarchies (hierarchy depth, the number of “•” in the embodiment + 1) is fixed to two in this embodiment. It is also possible to release the fixed number of layers and output all. The metadata value of the upper area 1702 can be manually input. When all of the metadata values in the upper area 1702 are output, the metadata is in the same category (directory structure), and the individual metadata values from the manipulated in the process of transition from the upper level to the lower level of the hierarchy Are sequentially extracted, and metadata values (for example, metadata 1, metadata 2,...) Are output.
[0037]
Further, instead of the metadata map 1615 of FIG. 17, metadata selected in the form of a directory can be displayed as shown in 1801 of FIG. When the metadata value (metadata having a hierarchy) of the upper area 1702 is manually input, partial matching is performed between the metadata value and the metadata values of all metadata stored in the system, and the metadata values match. Displays the directory stored on the system. When the displayed directory is not what is expected, that is, when there are a plurality of metadata having the same hierarchy, other matching directories are displayed in order. The display order at that time includes the order registered earlier in the system, the order registered later, and the alphabetical order of the metadata names of the highest hierarchy. According to the method of the present embodiment, for example, it is possible to select a document from the concepts based on the configurations of both the directories 311 and 312 in FIG. In order to reduce such a problem, it is possible to separately specify the top metadata name. The designation method may be such that @ (top metadata name), metadata name, metadata value, etc. are prefixed with a symbol such as @ mark, or the top metadata name may be displayed in another area.
[0038]
FIG. 19 is also an example of a search function interface using metadata. The metadata is represented in a tabular form 1901. By clicking a metadata name such as “Product” 1911 and “Document” 1912 in the metadata table, it is possible to sort and display by the value of the metadata. It is also possible to change the sorting priority by exchanging metadata columns (such as “product” 1911 and “document” 1912) by a drag operation or the like.
[0039]
In addition, the following functions are useful from the viewpoint of user assistance. For example, in the process of selecting metadata in FIG. 16, the user's selection history so far is recorded, and the metadata names and metadata information that are easy for the user to select (that is, frequently selected) are made conspicuous. Etc. are also possible. As a method to make it stand out, “Representation of the degree of association by the background color, tone, and shading of the label”, “Expression of the degree of association by adding an identification mark to the label”, “Relationship by the type and width of the label border "Representation of relevance by label shape and size", "Representation of relevance by character font", and the like.
[0040]
Regarding the relationship between metadata, a relationship that co-occurs (reacts together) and an exclusive relationship can be defined in the metadata. Thus, when a metadata label is selected, a metadata label having a co-occurrence relationship can be automatically selected (for example, when a metadata value “JAVA (R)” is selected). In addition, “object-oriented” is automatically selected), and it is also possible to automatically deselect metadata in an exclusive relationship.
[0041]
From the viewpoint of application to a relational database (referred to as RDB), the present invention can be applied not only to metadata attached to a document but also to database data. For example, each field name such as RDB can be used as a metadata name, and a field value (data) can be used as a metadata value. In addition, by defining the field to be referred to as the text, the above search method can be used. Further, by outputting RDB or the like in XML format, the present invention can be applied as it is. As a result, the information stored in the database can be narrowed down as desired while visually confirming the relationship between the data without fixing the hierarchical structure in a system using a metadata map.
[0042]
Regarding the rearrangement of metadata, a method for making the screen easier to see and operate is provided by the following method. That is, regarding the display of the metadata map, when the number of metadata labels increases, the maps are rearranged automatically so that the labels do not overlap. In order to make the map easier to see, the labels are automatically rearranged so that the lines are not crowded. There are a plurality of label rearrangement methods as described below. As a first method, overlapping and congestion are solved by shifting only the labels in the vicinity of each label. This method rearranges the labels without greatly changing the current arrangement and makes it easier to see. As a second method, all labels are evenly arranged. This method is a method for rebalancing the whole while maintaining the current arrangement to some extent. By this method, the entire label is rearranged evenly to make it easier to see. Third, there is a method of completely rearranging by using random numbers. In this case, the current arrangement of the metadata map is not maintained, but it is effective when the current arrangement is not desired. Further, as a fourth method, there is a method of arranging similar labels in the vicinity. Rearrangement of the metadata map labels can be done manually during the search process. It is also possible to automatically rearrange labels according to environment settings and option settings. In addition, the size of characters and figures can be changed as the display information increases or decreases. When the number of displayed metadata exceeds a certain value, the number of points that is the size of the character display is reduced, or when the overlapping display is possible or the display interval is narrowed, the number of points that is the size of the character display is reduced. This makes it easy to see. Further, when the relation is displayed as a line as in the example 1614 in FIG. 16, it is preferable to display the relation so that the line does not cross.
[0043]
17-19, especially in FIG. 17, when a large amount of information is displayed in the display area 1615, the display size of characters and the like must be reduced. In this case, it is desirable to enable an operation for expanding the display area of 1615 by making the boundary between the display areas 1615 and 1701 movable. It is also desirable to automatically increase the display characters in area 1615 as the display area expands. Conversely, in the operation of reducing the area 1615, the characters in the area 1615 are automatically reduced. Alternatively, it is also effective to automatically erase the display of a thin line with a low relevance, that is, a relevance line.
[0044]
As described above, as an embodiment of the present invention, a method has been described in which document metadata is not defined in a directory structure, but is defined in a free format structure that is not conscious of upper concepts and lower concepts. This method can be regarded as a document storage method in which a document is stored in a state to which metadata is added without being stored in a folder (directory). According to this method, it is not necessary to be aware of the directory structure, and it is possible to easily define document metadata even when there are a plurality of superordinate concepts. Therefore, it is possible to accurately register metadata envisioned by a document registrant. In the embodiment of the present invention, the function has been described in which information can be narrowed down like directory search by selecting a label without being aware of the directory on the metadata map. At this time, since the metadata map has elements that can be expanded into a plurality of directory configurations, it is possible to perform a search assuming a directory configuration envisioned by the searcher. In addition, when performing this search, metadata can be displayed in the form of a metadata map, directory, and tabular, where you can search from your favorite label (location) or directory or value or field. A search method was presented. In addition, a method has been described in which, in the process of searching for a document, the relationship with a document included in another directory configuration (category) can be easily discriminated by connecting labels with lines. According to this method, it is possible to search the related information widely by tracing the relationship between the metadata, and it is possible to investigate the necessary information sufficiently and to improve the search accuracy, which is very convenient. Further, the contents of the search method that can easily search the contents of the DB using the DB field information as metadata have been described. If the contents of the DB can be searched as they are by the search method of the present embodiment, it is possible to search a DB that has already been constructed visually and comfortably without using effort in determining keywords.
[0045]
【The invention's effect】
According to the present invention, the operability of the operation of assigning metadata to a document as the operation of categorizing information for search processing is improved.
Further, according to the present invention, in the search method using the metadata label or the like, the operability of the search process is improved or the search accuracy is improved.
[Brief description of the drawings]
FIG. 1 is a system configuration diagram of an information search apparatus.
FIG. 2 is a process flowchart of the entire system.
FIG. 3 is a configuration example of a category directory.
FIG. 4 is an example of metadata.
FIG. 5 is an example of how to have metadata.
FIG. 6 is a flowchart of document accumulation and metadata assignment processing.
FIG. 7 is a screen transition of document accumulation and metadata assignment processing.
FIG. 8 is an explanatory diagram of a user authentication screen for document accumulation and metadata assignment processing;
FIG. 9 is an explanatory diagram of a personal library search screen for document accumulation and metadata assignment processing;
FIG. 10 is an explanatory diagram of a personal library search result screen for document accumulation and metadata assignment processing;
FIG. 11 is an explanatory diagram of a document list screen for document accumulation and metadata assignment processing;
FIG. 12 is an explanatory diagram of a document content screen for document accumulation and metadata assignment processing;
FIG. 13 is an explanatory diagram of a new document registration screen for document accumulation and metadata assignment processing;
FIG. 14 is an explanatory diagram of a metadata change screen for document accumulation and metadata assignment processing;
FIG. 15 is an explanatory diagram of a metadata change detail screen for document accumulation and metadata assignment processing;
FIG. 16 is a flow of retrieval by a metadata map.
FIG. 17 is a search result by a metadata map.
FIG. 18 shows metadata directories and search results.
FIG. 19 shows a metadata table and search results.
[Explanation of symbols]
101 server (search processing device)
102 Client (operating device)
103 metadata database
104 Document database
105 Document metadata database
201 Document input means
202 Corresponding metadata extraction means
203 Corresponding metadata identification means
204 Corresponding metadata giving means
209 Metadata input means
210 Metadata corresponding document search means
214 Metadata Applicable Document Output Means
220 Metadata presentation means
1111 An example of the utterance of the metadata grant agent
Display example of 1615 metadata map
1701 Example of document search result (example of search document list)
1801 Example of directory display of metadata
1901 Display example of metadata in tabular format

Claims (3)

The relationship between the metadata value that is a search key when searching for information to be searched and the metadata name that is the category of the metadata value, and
A search processing device having a display unit and an input unit for searching for information while displaying the metadata value and the association between the metadata value belonging to the other metadata name;
A metadata database storing the metadata value in association with the metadata name;
A document database storing the information in association with an identifier identifying the information; and
A storage device comprising a document metadata database storing a plurality of the metadata names and the metadata values belonging to each metadata name in association with an identifier for identifying information;
  An information retrieval system having
The search processing device includes:
When receiving input of information to be searched through the input unit, for each metadata name, from the metadata database, to obtain all metadata values included in the metadata database related to the metadata name,
For all the obtained metadata values, obtain the similarity to the information that received the input,
For each metadata name, create first similarity information in which the similarity ranking is associated with a metadata value,
Among the information included in the document database, the document metadata database is searched for a predetermined number of pieces of information determined to be similar to the information that has received the input, and all the metadata values related to the information are acquired,
For all metadata values obtained by searching the document metadata database, calculate the similarity with the information that received the input,
For each metadata name, create second similarity information in which the similarity ranking is associated with a metadata value,
For each metadata name, determine a metadata value based on the first similarity information and the second similarity information;
The determined metadata value is stored in the document metadata database in association with an identifier for identifying information and a metadata name.
And
A label indicating at least one metadata name stored in the document metadata database is displayed via the display unit;
Accepting an initial selection for the displayed label via the input unit;
Search the document metadata database to obtain all the metadata values belonging to the metadata name indicated by the first selected label;
An initial label indicating the acquired metadata value is displayed via the display unit while showing a relationship with a label indicating the metadata name by a line,
Among the labels indicating the displayed metadata names, the selection following the first selection is accepted for the unselected labels through the input unit,
Search the document metadata database to obtain all metadata values belonging to the metadata name indicated by the selected label following the first selection;
A label based on a selection subsequent to the first selection indicating the acquired metadata value is displayed via the display unit with a line indicating a relationship with the label selected following the first selection;
If information including a combination of each metadata value indicated by the first label and each metadata value indicated by a label based on the selection following the first selection exists in the document metadata database, the combination is configured. Displaying a line indicating a relationship between labels indicating the metadata value to be displayed via the display unit;
Information retrieval system characterized by The relationship between the metadata value that is a search key when searching for information to be searched and the metadata name that is the category of the metadata value, and
The association between the metadata value and the metadata value belonging to another metadata name
A search processing device having a display unit and an input unit for searching information while displaying;
A metadata database storing the metadata value in association with the metadata name;
A document database storing the information in association with an identifier identifying the information; and
A storage device comprising a document metadata database storing a plurality of the metadata names and the metadata values belonging to each metadata name in association with an identifier for identifying information;
  An information search method using an information search system having
The search processing device includes:
When receiving input of information to be searched through the input unit, for each metadata name, from the metadata database, to obtain all metadata values included in the metadata database related to the metadata name,
For all the obtained metadata values, obtain the similarity to the information that received the input,
For each metadata name, create first similarity information in which the similarity ranking is associated with a metadata value,
Among the information included in the document database, the document metadata database is searched for a predetermined number of pieces of information determined to be similar to the information that has received the input, and all the metadata values related to the information are acquired,
For all metadata values obtained by searching the document metadata database, calculate the similarity to the information that received the input,
For each metadata name, create second similarity information in which the similarity ranking is associated with a metadata value,
For each metadata name, determine a metadata value based on the first similarity information and the second similarity information;
The determined metadata value is stored in the document metadata database in association with an identifier for identifying information and a metadata name.
And
A label indicating at least one metadata name stored in the document metadata database is displayed via the display unit;
Accepting an initial selection for the displayed label via the input unit;
Search the document metadata database to obtain all the metadata values belonging to the metadata name indicated by the first selected label;
An initial label indicating the acquired metadata value is displayed via the display unit while showing a relationship with a label indicating the metadata name by a line,
Among the labels indicating the displayed metadata names, the selection following the first selection is accepted for the unselected labels through the input unit,
Search the document metadata database to obtain all metadata values belonging to the metadata name indicated by the selected label following the first selection;
A label based on a selection subsequent to the first selection indicating the acquired metadata value is displayed via the display unit with a line indicating a relationship with the label selected following the first selection;
If information including a combination of each metadata value indicated by the first label and each metadata value indicated by a label based on the selection following the first selection exists in the document metadata database, the combination is configured. Displaying a line indicating a relationship between labels indicating the metadata value to be displayed via the display unit;
Information retrieval method characterized by The relationship between the metadata value that is a search key when searching for information to be searched and the metadata name that is the category of the metadata value, and
The association between the metadata value and the metadata value belonging to another metadata name
A search processing device having a display unit and an input unit for searching information while displaying;
A metadata database storing the metadata value in association with the metadata name;
A document database storing the information in association with an identifier identifying the information; and
A storage device comprising a document metadata database storing a plurality of the metadata names and the metadata values belonging to each metadata name in association with an identifier for identifying information;
  An information search program for functioning an information search system having
For the search processing device,
When receiving input of information to be searched through the input unit, for each metadata name, from the metadata database, to obtain all metadata values included in the metadata database related to the metadata name,
For all the obtained metadata values, obtain the similarity to the information that received the input,
For each metadata name, create first similarity information in which the similarity ranking is associated with a metadata value,
Among the information included in the document database, the document metadata database is searched for a predetermined number of pieces of information determined to be similar to the information that has received the input, and all metadata values related to the information are acquired.
For all metadata values obtained by searching the document metadata database, calculate the similarity with the information that received the input,
For each metadata name, create second similarity information in which the similarity ranking is associated with a metadata value,
For each metadata name, determine a metadata value based on the first similarity information and the second similarity information;
The determined metadata value is stored in the document metadata database in association with an identifier for identifying information and a metadata name.
Function and
A label indicating at least one metadata name stored in the document metadata database is displayed via the display unit;
Accepting an initial selection for the displayed label via the input unit;
Search the document metadata database to obtain all the metadata values belonging to the metadata name indicated by the first selected label;
An initial label indicating the acquired metadata value is displayed via the display unit while showing a relationship with a label indicating the metadata name by a line,
Among the labels indicating the displayed metadata names, the selection following the first selection is accepted for the unselected labels through the input unit,
Search the document metadata database to obtain all metadata values belonging to the metadata name indicated by the selected label following the first selection;
A label based on a selection subsequent to the first selection indicating the acquired metadata value is displayed via the display unit with a line indicating a relationship with the label selected following the first selection;
If information including a combination of each metadata value indicated by the first label and each metadata value indicated by a label based on the selection following the first selection exists in the document metadata database, the combination is configured. Realizing a function of displaying a relationship between labels indicating the metadata value via the display unit;
An information retrieval program characterized by

Images