JP2001022778A - Device and method for retrieval - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a retrieval device which permit designation of a retrieval condition through easy operation and actualizes higher retrieving capability than simple key word retrieval. SOLUTION: A database 10 has each of pieces of data mapped in a factor space composed of a specific factor group. A key word storage part 220 is stored with many key words used for retrieval from the database 10 so that they are made to correspond to position coordinates in the factor space. A retrieval UI(user interface) 210 presents those key word groups to a user and accepts a user's choice, obtains the position coordinates of the selected key word in the factor space from the key word storage part 220 and requests a data retrieval part 12 to retrieve data nearby the position.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for searching a database, and more particularly to a method for searching a database of a type in which each data is classified according to a plurality of factors and the data is searched by the position of each data in a factor space. Method and apparatus.

[0002]

2. Description of the Related Art Document databases, image databases,
In a music database or the like, a number of search keywords are assigned to each data (that is, document data, image data, music data, or the like) at the time of data registration, and a search keyword specified by a user is included at the time of search. The type that seeks data is more widely used than before. As is well known, in this type of database, the quality of the keyword assigned to each data and the consistency of the keyword among the data determine the search ability. In extreme cases,
Even if the keyword assigned to the data has the same or similar content as the keyword specified as the search key, if the character string differs, the data may not be searched.

[0003] Various approaches have conventionally been taken to solve such problems.

For example, in the prior art disclosed in Japanese Patent Application Laid-Open No. 7-200608, an associative keyword associated with each keyword is registered in a keyword database in advance.
At the time of the search, the search is performed not only for the keyword specified by the user but also for the associative keyword corresponding to the keyword.

Further, in the conventional document database system disclosed in Japanese Patent Application Laid-Open No. 9-305602, each of a plurality of pre-selected impression words (for example, "brightness", "power", etc.) Score (degree of matching with impression words)
Is set. That is, when constructing a database,
Instead of assigning a search keyword to a document, a score is set for each impression word such as “brightness” is 1 and “strength” is 4 or the like. And when searching,
The system requests the user to specify the degree of matching for each impression word, and searches for document data having a score distribution corresponding to the specification.

As a technique that further refines this method, there is a technique disclosed in Japanese Patent Application Laid-Open No. 9-34909. In this conventional technique, a plurality of adjectives are determined in advance as information for designating data, and a search instruction is received from a user in the form of coordinates in an adjective space constituted by the plurality of adjectives. On the other hand, in the database, each data is mapped to a factor space defined by a factor group obtained by performing a factor analysis on an adjective group. Data search is performed by converting coordinates in the adjective space designated by the user into coordinates in the factor space, and searching for data close to the coordinates in the factor space.

Further, as a similar prior art, there is a technique disclosed in Japanese Patent Application Laid-Open No. 10-40262.

[0008]

In the prior art disclosed in Japanese Patent Application Laid-Open No. 7-200608, not only an inputted keyword but also an associative keyword is searched, so that a search range is widened. However, the expansion of the search range is
It is only a spread at the level of the "dot" of associative keywords. That is, even if the keyword assigned to the data has the same or similar content as the associative keyword, if it is different as a character string, there remains a problem that the data is not searched.

In the prior art disclosed in JP-A-9-305602, JP-A-9-34909, and JP-A-10-40262, each data of a database is mapped to a certain space such as a factor space. Since the search can be performed using the concept of the distance in the space, the problem of the prior art described in JP-A-7-200608 can be avoided. However,
In these conventional techniques, when designating search conditions, it is necessary to input a degree for each of several to several tens of impression words (feeling expression words), and the operation tends to be complicated.

SUMMARY OF THE INVENTION The present invention has been made to solve such a problem, and provides a search apparatus and method capable of designating search conditions with a simple operation and realizing higher search performance than a simple keyword search. provide.

[0011]

To achieve the above object, a search device according to the present invention is a search device for searching data from a database consisting of a plurality of data arranged in a factor space. A keyword storage unit that stores a keyword for designating a target together with the position of the keyword in the factor space, and a list of keywords stored in the keyword storage unit, and a list of keywords used as a search key from the user is displayed. Accept the selection,
A search key specifying unit for obtaining the position of the selected search key in the factor space from the keyword storage unit, and searching the database for data within a predetermined distance from the position obtained by the search key specifying unit. A data search unit.

According to this configuration, each data mapped on the factor space can be easily searched using the keyword. Further, in this configuration, not only the keyword strictly corresponding to the keyword but also the keyword occupying a position close to the keyword in the factor space is searched, so that the search ability is much higher than that of a normal keyword search. Note that the “factor space” is not always formed by a group of factors determined by so-called factor analysis. It is sufficient for a factor to have the property of an index that expresses the attribute of data, and each data is spatially mapped by considering the degree to which each data satisfies the condition of such a factor as coordinate values. If possible.

In a preferred embodiment, first, a limited representative keyword is presented as an option, a keyword group similar to the selected representative keyword is presented, and a search key is selected from the keyword group. To be configured. According to this aspect, it is possible to efficiently search for what the user desires from a large number of keywords.

In a further preferred aspect, the representative keyword group can be specified by the user, so that the user interface for specifying the keyword can be easily understood by the user.

[0015] A search method according to the present invention is a search method for searching data mapped on a factor space composed of a plurality of factors, wherein a plurality of keywords selected for designating a search target are provided. Is mapped on the factor space, a data search key is selected from the plurality of keywords, and data is searched based on a distance of the selected data search key in the factor space. .

According to this method, it is possible to provide a more powerful search capability than a normal keyword search by simply specifying search conditions.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention (hereinafter abbreviated as embodiments) will be described below with reference to the drawings.

FIG. 1 is a functional block diagram showing the configuration of a search device according to the present invention. Here, a search of a database of sound (song) data such as popular music will be described as an example.

The database 10 is a database storing sound data. In the database 10, each sound data is mapped in a factor space composed of predetermined sensitivity factors. In the present embodiment, for example, 20 sensitivity factors are defined, and a 20-dimensional factor space is defined using each of these 20 factors as a coordinate axis. Each sound data is mapped to a 20-dimensional factor space by setting the coordinate values of each of the 20 sensitivity factors.

FIG. 2 shows positional information of each sound data held in the database 10 in the factor space. In FIG. 2, the items "snd-C-0001" and "snd-C-000"
"2" and the like are the file names of the sound data, and the items in each column "feeling of power" and "feeling of compassion" are the names of sensitivity factors.
Each numerical value set at the intersection of a row and a column indicates a coordinate value on the sensitivity factor axis of the sound data. In the illustrated example, a 20-dimensional factor space of (feelings of power, affection,..., Freshness) is formed. For example, the sound data “snd-C-0001” has coordinates (0. 005,
0.7803,..., 0.0205). Each coordinate value is a value of 0 or more and 1 or less, and the closer to 1, the stronger the sensitivity factor is.

Each sensibility factor exemplified here is obtained by factor analysis and cluster analysis of a sensitivity keyword group described later.

Since the database 10 is configured as described above, if a position in the factor space is designated, sound data close to the position can be searched. The data search section 12 performs this search processing.

The search UI (user interface) processing unit 20 is a processing unit that performs processing for a user interface when searching the user database 10. The search UI processing unit 20 is connected to an input device 30 such as a mouse or a keyboard, a display device 32 such as a CRT display or a liquid crystal display, and an audio output device 34 such as a speaker. The search UI processing unit 20 receives a user's input from the input device 30, executes a predetermined process according to the input, and outputs the processing result from the display device 32 or the audio output device 34. The search UI processing unit 20 includes a search UI
210, keyword storage unit 220, representative key registration unit 23
0, a main search menu change unit 240 is provided.

The search UI 210 receives an input of a search condition from the user, requests the data search unit 12 to perform a search process according to the search condition, and presents the search result returned from the data search unit 12 to the user. And other processing. The acquisition process of the search condition input is performed by the search condition acquisition unit 212.
The search result output process is performed by the search result presentation unit 2
18.

Unlike the prior art, such as Japanese Patent Application Laid-Open No. 9-34909, in which a search condition is specified by a set of degrees (numerical values) of each impression word, in the present embodiment, the search condition can be specified by a keyword. . In the present embodiment, in order to express the impression of hearing the sound data, more than 200 various impression words (kansei expression words) are employed as keywords (kansei keywords). These keywords are stored in the keyword storage unit 220. FIG. 3 shows an example of data contents stored in the keyword storage unit 220. In FIG. 3, the items “competence” and “comfort” in each column are
As in the case of FIG. 2, the names given to the sensibility factors are shown, and the items “adorable”, “bright”, etc. in each row are sensibility expression words used as keywords. The numerical value shown at the intersection of each row and each column indicates the coordinate value of each keyword for each sensitivity factor axis. As described above, the keyword storage unit 220 stores, for each keyword, position coordinate information of the keyword in the factor space.

The search condition acquisition unit 212 presents these keyword groups to the user and accepts selection of a keyword to be used as a search key. In the present embodiment, in order to make it easy for the user to select a keyword, a mechanism is first provided for selecting a keyword by a two-stage operation. That is, the first
At the stage, selection of a keyword category is received, and at the second stage, selection of a keyword group to be used as a search key is received from a group of keywords belonging to the category. That is, in the present embodiment, since a large number of keywords exceeding 200 are used, the number of candidate keywords is once narrowed down by selecting a category. The main search processing unit 214 performs the first-stage processing, and the sub-search processing unit 216 performs the second-stage processing.

That is, the main search processing unit 214 displays a plurality of categories prepared in advance as a main search menu, and passes the category selected by the user on this menu to the sub search processing unit 216. The sub search processing unit 216 extracts the keyword group belonging to the selected category from the keyword storage unit 220, and displays a list of these keywords as a sub search menu. And the sub search processing unit 2
16 fetches one or a plurality of keywords selected by the user on the sub search menu as a search key (search condition).

In this embodiment, the user can customize the category of the keyword. That is, in the present embodiment, the user is allowed to select some keywords from among a large number of keywords registered in the keyword storage unit 220, and these selected keywords can be set as representative keywords representing each category. I have. The selected representative keyword group is registered in the representative key registration unit 230 as main search menu information. The main search processing unit 214 reads main search menu information (that is, a list of representative keywords) from the representative key registration unit 230, and generates a main search menu based on this.
Display and ask user for selection. In the present embodiment, a plurality of pieces of main search menu information can be registered in the representative key registration unit 230, and the user can select and use desired information from these. This category classification customizing process is performed by the main search menu change UI 240. The main search menu change UI 240 provides a user interface for setting and changing the category classification.

In this embodiment, a category classification based on a group of sensitivity factors is provided as a default category classification. That is, a list of the names of the 20 sensitivity factors is
This list of kansei factor names is used as the main search menu unless the user instructs to use another main search menu, which is stored in the representative key registration unit 230 as main search menu information.

The search condition acquisition section 212 is a main search processing section 2
14 and the search conditions received by the sub-search processing unit 216 are converted into a query format that can be processed by the data search unit 12 (that is, the condition of the position coordinates to be searched in the factor space). I do. A search result of the data search unit 12 is presented (for example, displayed) to the user by the search result presentation unit 218.

The schematic configuration of the search device according to the present embodiment has been described above. This search device is typically constructed as software on a computer such as a personal computer and a workstation. In this case, the database 10 and the data search unit 12 are connected to the search UI processing unit 20.
May be configured on a computer separate from the computer that configures, and a search inquiry may be made via a network.

Next, a search procedure in the apparatus of the present embodiment will be described with reference to a display example and a flowchart of a user interface screen.

FIG. 4 shows an example of a search UI screen generated and managed by the search acquisition unit 212. A search window 401 as a search UI screen includes a message column 41
2. Main search menu 409, sub search menu 410, search result display list 411, search range setting field 404, search execution button 402, end button 403, audition button 40
5, a detail display button 406, a record button 407, a save button 408, and a main search menu change button 413. FIG. 7 shows a procedure of a search process using the search window 409.

FIG. 4 shows the search window 401 in the initial state immediately after the start of the search device (S10 in FIG. 7). In this state, the main search menu 409 displays a default category list. That is, the main search menu 409 displays a list of the names of the sensibility factors (“feeling of power,” etc.) constituting the factor space of the database 10. On the right side of the main search menu 409, a scroll bar 409a for scrolling this list is displayed. If your list is short and does not need to be scrolled,
Scroll bar is not displayed. Search range setting field 40
4 includes a main search menu 409 and a sub search menu 41
A search range for the search key selected at 0 is set. In the present embodiment, data near the search key is searched in the factor space. The search range setting field 404 sets the range of “nearby”. In the present embodiment, assuming that the distance from the search key in the factor space is r, a numerical value obtained by subtracting r from 1 is referred to as “similarity”, and the search range is set by the lower limit (cut-off value) of the similarity. I do. Note that r is a value normalized to be 1 at the maximum. In the search, only data having a similarity greater than or equal to the lower limit is searched. The value 0.6 shown in FIG. 4 is an example of a default value. To change this value, bring the mouse cursor into this field and enter the value with the keyboard. Note that the definition of “similarity” used here is only an example. Further, instead of using the similarity, the determination may be made based on the distance itself in the factor space. In this case, the upper limit of the distance considered to be nearby is set in the “search range”.

When the initial screen is displayed, the process enters a loop waiting for an input from the user (S12). Input on the search window 401 includes key selection on the main search menu 409 (S14), key selection on the sub search menu 410 (S20), depression of the search execution button 402 (S24),
Audition button 405, detail button 406, record button 40
7, pressing of a button for instructing various kinds of processing on the search result such as the save button 408 (S32), pressing of the main search menu change button 413 (S36), or pressing of the end button 403 (S40).

When a certain category is selected on the main search menu 409 by a mouse click operation or the like (S1).
The sub-search processing unit 216 that has received the selection result from the main search processing unit 214 acquires the keywords (candidate keywords) belonging to the selected category from the keyword storage unit 220 (S16), A list of these candidate keywords is displayed on the sub search menu 410 (S18). In the initial state shown in FIG. 4, the main search menu 409 displays a list of sensitivity factor names,
When the user selects a certain sensibility factor, the sub-search processing unit 216 has a high score for the sensibility factor from the keyword group stored in the keyword storage unit 220 (for example, a predetermined threshold). A group of keywords (more than the value) is extracted and displayed on the sub search menu 410. Also,
When a list of the representative keywords selected by the user is displayed in the main search menu 409 (that is, when a customized main search menu is used), when a representative keyword is selected, the sub search processing unit 216 The similarity of each keyword in the keyword storage unit 220 with the representative keyword is calculated, and a keyword group in which the value of the similarity is equal to or greater than a preset threshold is extracted and displayed on the sub search menu 410. That the similarity is equal to or more than the predetermined value, from another viewpoint, means that the similarity is within a predetermined distance from the representative keyword in the factor space. In the present embodiment, a threshold value serving as a keyword extraction criterion of the sub search menu 410 can be set freely. Therefore, by changing the threshold value, the number (range) of candidate keywords displayed in the sub search menu 410 can be changed even when the same representative keyword is selected in the main search menu 409.

When the list of candidate keywords is displayed on the sub search menu 410 in this manner, a search key can be selected on the sub search menu 410.

When a keyword in the sub search menu 410 is selected by an operation such as a mouse click (the determination result in S20 is Y), the sub search processing unit 216 stores this keyword as a specified search key ( S22). In the present embodiment, a plurality of keywords in the sub search menu 410 can be selected as search keys. When a plurality of search keys are selected, a search is performed as an AND condition of the plurality of search keys. When the keyword once selected on the sub search menu 410 is further clicked, the sub search processing unit 216 releases the selected state of the keyword and deletes it from the stored search key group.

When the search execution button 402 is selected by a mouse or the like (the determination result in S24 is Y), the search condition obtaining unit 212 uses the position coordinates in the factor space of each keyword stored as a search key in S22 as a keyword. Storage unit 22
0 (S26), a description of the search query is created based on the position coordinates, and the description is sent to the data search unit 12 to make a query. For example, when the number of search keys is one, the search condition acquisition unit 212 uses the position coordinates of the search keyword and the lower limit value of the search range set in the search range setting column 404 as the center of the search keyword. Create a search condition description for searching for data in. Upon receiving this inquiry, the data search unit 12 searches the database 10 for data that satisfies the condition (S
28), and passes the search result (for example, a file name list of the searched data file group) to the search result presentation unit 218 of the search UI 210. In this search (S28), for each data in the database 10, the distance from the keyword of the search key is calculated, and this distance is converted into a similarity (similarity of the data to the keyword). Then, data in which the value of the similarity falls within the designated search range is extracted. The search result may include information such as the degree of similarity of each data calculated at this time. Search result presentation unit 218
Displays a list of the received search results in the search result display list 411 (S30). In the example of FIG. 4, the search result display list 411 displays the file name of each data of the search result and the value of the similarity of each data as a “score”.

When a plurality of keywords are selected on the sub-search menu 410 and the search execution button 402 is pressed, the search condition acquisition unit 212 generates a query description representing an AND condition of the plurality of keywords. create.
That is, a search condition for instructing to search data included in an intersection area of each search range centering on each of the keywords is created. Data retrieval unit 12 receiving this
Extracts data in the intersection area and returns the data to the search UI 210 as a search result. For example, when two keywords “dense” and “three-dimensional” are selected, data having a similarity of 0.6 or less is retrieved from both keywords.

The search processing based on the keyword selected on the sub search menu 410 has been described above.
When a sensitivity factor name or a representative keyword is selected in the main search menu 409, a search can be performed by pressing the search execution button 402 without making a selection in the sub search menu 410. In this case, the search condition obtaining unit 212 determines the sensibility factor or the representative keyword selected in the main search menu 409 as a search key, and creates a search condition description for searching for data similar to the search key. .

When a keyword or the like is not selected in either the main search menu 409 or the sub search menu 410, it goes without saying that the pressing of the search execution button 402 is invalidated.

The above processing and operation will be described with reference to specific examples. For example, when the sensibility factor “fertility” is selected on the main search menu 409 in the initial state shown in FIG. 4, the search window 401 becomes the state shown in FIG. In this state, in the main search menu 409, the selected “richness” is selected.
Are displayed in a display form that can be distinguished from the others, and the sub-search menu 410 displays a list of keywords that are close to the “richness”. In this state, for example, when the keyword “mellow” is selected in the sub search menu 410 and the search execution button 402 is pressed, the database 10 is searched, and the search results are displayed in the search result display list 411.
(See FIG. 6).

When the search results are displayed as described above, various processing instructions can be given to the search results. In the present embodiment, as the processing for the search result, processing functions such as audition, detailed data display, file name recording, and file saving are provided. In the search window 401, buttons 405 to 408 corresponding to the respective processing functions are provided. Is provided. The trial listening function is a function provided based on a request for use of a sound database. A file name on the search result display list 411 is selected, and the buttons 405-4
08 is pressed, the determination result of S32 becomes Y
(Yes), the search result presentation unit 218 performs a process corresponding to the button (S34).

That is, when the user enters the search result display list 4
When a file name is selected on the file 11 and the details button 406 is pressed, detailed information on the file is displayed in the message field 4.
12 is displayed. When a file name is selected and the record button 407 is pressed, the file name is recorded in a predetermined search result recording file. Use this function when you want to record the file name of the sound data you want. When a file name is selected and a preview button 405 is pressed, the search result presentation unit 218 extracts sound data of the file from the database 10, converts the sound data into an audio signal, and outputs the audio signal from the audio output device 34. When a file name is selected and the save button 408 is pressed, the file is extracted from the database 10 and stored in a local directory separately specified by the user.

When the user presses the end button 403 with a mouse or the like (the determination result in S40 is Y), the search UI processing unit 2
0 ends the process, and as a result, the search window 401 is closed.

When the user wants to switch or customize the main search menu 409, he / she can click the main search menu change button 413. When the main search menu change button 413 is clicked, FIG.
In the procedure (3), the determination result in S36 is Y (Yes). Then, the main search menu change UI 240 is activated, and a main search menu change process is performed (S38). In this process, the main search menu change window 701 shown in FIG. 8 is displayed on the display device 32, and the user's instruction is received using the window 701 as a user interface. The main search menu change window 701 includes a menu setting button 702, a menu name input field 703, a menu addition button 704, a menu deletion button 705, a processing stop button 706, a key insertion button 707, a key insertion release button 708, a completion button 709, A main search menu name list 710, a main search key list 711, a search key candidate list 712, and a message column 713 are provided. In the main search menu name list 710,
The names of one or more main search menus registered in the representative key registration unit 230 are displayed in a list. In the example of FIG.
The default main search menu is registered with a menu name of "initial setting". “User definition 1” and the like are the names of the main search menu separately created (customized) by the user. When activated, the main search menu change UI 240 opens a window 701, reads a list of registered main search menus from the representative key registration unit 230, and displays the list on the main search menu name list 710.
At this time, the search window 401 (search acquisition unit 212)
The main search menu name currently used is displayed in a form that can be distinguished from other menu names, such as reverse display. Then, a list of representative keywords (or sensitivity factor names) belonging to the currently used main search menu is displayed in the main search key list 711.

Initially, the main search menu name list 710 is active (accepts input), but if another list or button is pressed with a mouse or the like, the pressed object becomes active. In addition, objects that become active can be sequentially switched by an operation such as pressing a TAB key. The operation around this is the same as the operation in the widely used window environment.

To switch the key list of the main search menu 409 displayed in the search window 401, the user only has to select another main search menu name on the main search menu name list 710. Then, a list of the representative keywords belonging to the newly selected main search menu is read from the representative key registration unit 230, and the main search key list 711 is read.
Is switched to this new list of representative keywords. The user checks the display of the main search key list 711, and if the user determines that the desired representative keyword group is displayed, the user can press the menu setting button 702.
By pressing this button, the list of the representative keywords corresponding to the currently selected main search menu name is searched U
It is passed to I210. As a result, the search window 401
Of the main search menu 409 is switched to a list of newly received representative keywords.

If there is no desired main search menu in the registered main search menu, a new main search menu can be created. In this case, the user may input a character string of a new menu name in the main search menu name input field 703 and press a menu addition button 704. Then, new main search menu information having the character string as a menu name is generated in the representative key registration unit 230. Then, the character string is displayed below the registered menu name in the main search menu name list 710, and an empty list is displayed in the main search key list 711. By incorporating desired keywords into the main search key list 711, a main search menu can be created. The keywords included in the main search key list 711 are limited to the keywords registered in the keyword registration unit 220. The search key candidate list 712 displays a list of all the registered keywords. When the user selects a desired keyword from this list 712 with a mouse or the like and clicks on the incorporate button 707, the keyword is added to the main search menu information of the selected menu name, and the main search key list 71
1 is displayed. On the other hand, to delete an unnecessary keyword from the main search menu, the user may select the keyword in the main search key list 711 and press the release button 708. As a result, the keyword is deleted from the main search menu information. In this way, the representative keywords registered in the main search menu can be selected. When the desired main search menu is configured, the main search menu 409 in the search window 401 can be switched to the menu by clicking the setting button 702.

As described above, in this embodiment, any keyword registered in the keyword registration unit 220 can be registered in the main search menu. The sentiment factor name shown in the default main search menu is a name given to the sentiment factor obtained by performing a factor analysis on a keyword group after considering the meaning of the factor. Can be confusing. On the other hand, in the present embodiment, the main search menu can be configured by selecting a keyword that is familiar to the user, so that the user can specify the category he or she wants to search for using a keyword whose semantic content is well known. Can be.

However, when a plurality of keywords having very similar meanings are registered in the main search menu as representative keywords, when those similar representative keywords are selected in the main search menu 409 (FIG. 4), they are displayed in the sub search menu 410. Of the candidate keywords are almost the same, resulting in increased waste. Therefore, in the present embodiment, the keyword incorporated in the main search key list 711 and the keyword having a high similarity to this keyword are not displayed in the search key candidate list 712. That is, a threshold value is set for the similarity, and a keyword whose similarity to a registered keyword is equal to or greater than the threshold is deleted from the search key candidate list 712. Thereby, it is possible to prevent a keyword having a similar meaning from being registered in the main search menu.

In the above, an example has been described in which a new main search menu is created and registered. However, the same operation can be used to edit a registered main search menu.

When the registered main search menu becomes unnecessary, the name of the menu is changed to the main search menu name list 71.
It is sufficient to make a selection on 0 and press the menu delete button 705. Then, the selected main search menu name is deleted from the main search menu name list 710. However, at this point, the information of the main search menu remains without being deleted, and the menu name can be restored to the main search menu name list 710 by pressing the stop button 706. To confirm the deletion, click the Finish button 7
Just press 09.

When the completion button 709 is pressed, the changes that have been made in the window 701 up to that time are confirmed, the changed items are registered in the representative key registration unit 230, and the window 701 is closed. For example, when the main search menu is deleted, information on the main search menu (such as a list of menu names and representative keywords) is deleted from the representative key registration unit 230. At this time, if the deleted main search menu is the menu currently used in the search window 401, the main search menu change UI 240
Informs the search UI 210 that the menu has been deleted. Then, the search UI 210 deletes the representative keyword list of the deleted menu from the main search menu 409 of the search window 401, and displays a default menu. When the main search menu is newly registered or the registered main search menu is edited, the result of the processing is registered in the representative key registration unit 230 by pressing the completion button 709, and is determined.

In this manner, the main search menu change UI2
When the setting change of the main search menu is completed by 40, the process returns to the search operation in the search window 401 again.

The configuration and processing procedure of the search device according to the present embodiment have been described above. Next, the keyword storage unit 22
0 and the procedure of constructing the database 10 will be described.

FIG. 9 is a flowchart showing a procedure for constructing the keyword storage unit 220. This procedure starts with keyword selection (S50). In this step, as many keywords as possible are picked up according to the content of each data registered in the database 10 and the purpose of the search. In the present embodiment, sound data is set as a search target,
Since the purpose is to search for the impression of listening to the song, a number of impression words (kansei expression words) representing the impression of the song are selected as keywords. Next, a group of keywords picked up in S50 is presented to a sufficient number of subjects selected by random sampling or the like, an impression of each keyword is input, and the results are totaled (S52). The input of the keyword impression is performed by, for example, selecting a similar keyword for each keyword or giving a score to each predetermined evaluation item for each keyword. And
A factor group that explains the correlation between the keyword groups well is obtained by performing a factor analysis on the total result. Here, since the emotion expression word is used as a keyword, the obtained factor is called an emotion factor. An appropriate name is devised and given to each of these sensitivity factors. Then, a factor load for each factor of the keyword is obtained for each keyword, and the set of factor loads is regarded as coordinates, and each keyword is mapped to a factor space (S56). Then, information of each of these keywords and their position coordinates in the factor space is registered in the keyword storage unit 220 (S58). Through such processing, keyword information for search is constructed in the keyword storage unit 220.

In the above example, the coordinate values of each keyword in the factor space are represented by numerical values from 0 to 1, but all keywords are, for example, “beautiful” and “not beautiful”.
, The coordinate value in the factor space is represented as a value from -1 to +1.

FIG. 10 is a flowchart showing a procedure for constructing the database 10. To construct the database 10, first, the subject is asked to listen to each sound data as a material, and the impression is input (S70). The impression of the sound data may be described in the same manner as the impression of the keyword. For example, the keyword storage unit 2
For example, the user may select some of the keyword groups registered in 20 that express the impression of the sound. Then, based on the input result of the impression, the coordinates of each sound data in the factor space are obtained (S72). Then, information on the position coordinates of each data obtained in this way is registered in the database 10. Thus, it is possible to search the database 10 for desired data by designating the position of interest and the search range (radius from the position of interest) in the factor space.

As described above, according to this embodiment, when a data group mapped in the factor space is searched, if a keyword that is easy for the user to specify is specified, data that is impressively close to the keyword is searched. can do. That is, according to the present embodiment, it is not necessary to specify the position of the search target in the factor space in the form of coordinates or the like, and the search conditions can be easily specified.

Further, in the present embodiment, the keywords are presented step by step in the main search menu 409 and the sub search menu 410, so that even when a large number of keywords are used, the burden of the user on selecting keywords can be reduced.

In this embodiment, the main search menu 4
By making it possible for the user to designate a representative keyword to be taken up in 09, the ease of search is improved.

In the above example, the main search menu 40
Although the similarity (or distance) of each keyword with respect to the representative keyword selected in step 9 is obtained, and the keyword whose similarity is equal to or more than the threshold is displayed on the sub search menu 410, another method is also possible. For example, a correlation matrix as shown in FIG. 11 obtained at the time of factor analysis of a keyword group may be stored in the keyword storage unit 220, and the sub search menu 410 may be created based on the correlation matrix. In this case, a correlation coefficient of each keyword with respect to the representative keyword selected in the main search menu 409 is obtained from the correlation matrix, and a keyword whose correlation coefficient value is equal to or larger than a predetermined threshold is displayed in the sub search menu 410. do it. Also, when accepting designation of a representative keyword to be registered in the main search menu on the main search menu change UI 240, a keyword having a predetermined or higher correlation strength with respect to the registered representative keyword is referred to by referring to the correlation matrix. Selection can also be prevented.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a schematic configuration of a search device according to an embodiment.

FIG. 2 is a diagram showing mapping information of each data in a database to a factor space.

FIG. 3 is a diagram showing mapping information of each keyword to a factor space in a keyword storage unit.

FIG. 4 is a diagram illustrating an example of a user interface screen for search processing.

FIG. 5 is a diagram illustrating an example of a user interface screen for search processing.

FIG. 6 is a diagram illustrating an example of a user interface screen for search processing.

FIG. 7 is a flowchart illustrating an example of a processing procedure of the search device.

FIG. 8 is a diagram showing an example of a user interface screen for changing a main search menu.

FIG. 9 is a flowchart illustrating an example of a procedure of a construction operation of a keyword storage unit.

FIG. 10 is a flowchart illustrating an example of a procedure of a database construction operation.

FIG. 11 is a diagram illustrating an example of a correlation matrix between keywords.

[Explanation of symbols]

Reference Signs List 10 database, 12 data search unit, 20 search UI (user interface) processing unit, 30 input device, 32 display device, 34 audio output device, 210 search front end, 212 search condition acquisition unit, 214
Main search processing unit, 216 sub search processing unit, 218 search result presentation unit, 220 keyword storage unit, 230 representative key registration unit, 240 main search menu change processing unit.

Claims (11)

[Claims] 1. A retrieval apparatus for retrieving data from a database composed of a plurality of data arranged in a factor space, wherein a keyword for designating a retrieval target is designated by a position of the keyword in the factor space. And a keyword storage unit that stores the keyword group stored in the keyword storage unit, accepts selection of a keyword to be used as a search key from a user, and determines the position of the selected search key in the factor space. A search device, comprising: a search key specifying unit obtained from a keyword storage unit; and a data search unit searching the database for data within a predetermined distance from the position obtained by the search key specifying unit. 2. The search key specification unit includes: a representative registration unit that stores a group of representative keywords selected from the keywords; and when the user instructs a database search, the group of representative keywords is used as a main search menu. The first to display and accept the user's selection for these representative keyword groups
A selection unit, obtaining a keyword having a similarity degree equal to or more than a predetermined value with respect to the representative keyword selected by the user, displaying the obtained keyword group as a sub-search menu, and receiving a user selection for this keyword group 2. The search device according to claim 1, further comprising: a unit configured to select a keyword selected in the sub search menu as a search key. 3. The search device according to claim 2, further comprising a representative selection unit that receives a selection of a keyword used as the representative keyword from a user and registers the keyword in the representative registration unit. 4. The apparatus according to claim 1, wherein the representative selecting unit includes means for preventing a keyword having a similarity of a predetermined value or more from a keyword already selected as a representative keyword from being selected as a representative keyword. The search device according to claim 3, wherein 5. The second selection unit obtains, from the keyword storage unit, a keyword located within a predetermined distance in the factor space with respect to the representative keyword selected by the first selection unit. 3. The search device according to claim 2, wherein the group is used as a keyword group having a similarity of a predetermined value or more to the representative keyword. 6. The apparatus according to claim 1, further comprising a correlation table in which strengths of correlation between the respective keywords registered in the keyword storage unit are registered, wherein the second selection unit selects the representative keyword selected by the first selection unit. A keyword having a correlation strength equal to or more than a predetermined value is obtained from the correlation table, and the obtained keyword group is used as a keyword group having a similarity of a predetermined value or more to the representative keyword. The search device according to claim 2, wherein 7. A retrieval method for retrieving data mapped on a factor space composed of a plurality of factors, wherein a plurality of keywords selected for specifying a retrieval target are mapped on the factor space. A search method for receiving a selection of a data search key from the plurality of keywords and searching for data based on a distance in the factor space from the selected data search key. 8. A representative keyword group selected from the keywords is displayed in a main search menu, and a selection of a representative keyword as a category search key is accepted in the main search menu. 8. A search method according to claim 7, wherein a keyword group having a similarity degree equal to or more than a predetermined value is displayed in a sub search menu, and selection of a data search key is accepted from the keywords displayed in the sub search menu. Method. 9. A keyword within a predetermined distance from the category search key in the factor space is obtained, and the obtained keyword group is defined as a keyword group having a similarity to the category search key with a predetermined value or more. 9. The search method according to claim 8, wherein the search method is displayed in the sub search menu. 10. A keyword having a correlation strength not less than a predetermined value with respect to the category search key is obtained from a correlation table in which correlation strengths between respective keywords are registered. 9. The search method according to claim 8, wherein a keyword group having a similarity of a predetermined value or more to a category search key is displayed in the sub search menu. 11. A factor analysis of a plurality of keywords selected for designating a search target to form a factor space including a number of factors smaller than the number of keywords, and mapping each of the keywords to the factor space. Each data to be searched is mapped to a factor space consisting of each factor obtained by the factor analysis, and a selection of a search key from among the plurality of keywords is received. A search method that searches for data based on distance.