JP2009064391A - Keyword extracting device, keyword extracting method, keyword extracting program, and recording medium with keyword extracting program recorded thereon - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a device, a method, and a program which extract a keyword in accordance with user's hobby, taste, and object. <P>SOLUTION: A presentation keyword control section 3 of a keyword extracting device of the present invention extracts a presentation keyword from a search keyword memorizing section 5 on the basis of a relation degree to a search keyword, a user's taste degree for each keyword, previously presented keywords, and the like. The extracted presentation keyword is memorized in a presentation keyword memorizing section 6 and is displayed on a display section 9 via a presentation keyword display preparing section 4 and a display control section 8. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a technique for extracting a keyword to be presented to a user, and in particular, can be applied to a search support apparatus that supports searching for any content including multimedia content.

Currently, keywords are used in various situations.
For example, Patent Literature 1 discloses a technique for extracting a keyword related to a keyword using a related word dictionary. Patent Document 2 discloses a method for searching for a document based on a search keyword input from a user.

  In addition, keywords are used to search for target content from a large amount of content that is becoming available on various devices (a large amount of text content distributed on the Internet, image content, music content, video content, etc.). Search using is used.

  In Patent Document 3, when a user inputs a keyword, the content providing server searches for content related to the input keyword, presents a list of searched content, and the user selects content from the presentation list. Then, a technique for supplying selected content is disclosed. In the technology disclosed in Non-Patent Document 1, initial keywords are displayed on the left side of the screen when activated, and when an arbitrary keyword is selected, program information that matches the selected keyword is displayed on the right side of the screen. . The keyword on the left side of the screen displays a word with a high frequency among words appearing in the information of the upper program of the program that matches the selected keyword.

In addition, in the web service on the Internet, it is possible to analyze the access of the web service and present the user with multiple keywords such as “Recently searched keywords” and “Keywords whose search number has been listed suddenly”. Has been. The user can perform a content search by selecting a desired keyword from the presented keywords and performing a search.
JP-A-5-282367 JP 2005-326482 A JP 2007-133688 A WISS2006 "Goromi-TV How to browse more than 1,000 videos taken with ease"

  However, when the user inputs a keyword for the purpose of searching or the like, it is a burden for the user to think of an appropriate keyword necessary for obtaining the target content.

  In some cases, the input of characters itself becomes a burden, and in particular, the input of characters with an input device that does not have a keyboard suitable for character input, such as a remote controller or a mobile phone, further increases the burden. Furthermore, in a search for selecting content from a large amount of content existing on the net, it is not always possible to obtain content that a user can satisfy with a single search. In the keyword-based search, steps are required to enter a keyword, and after completing the input, perform a search instruction and obtain a search result. However, even if the search result is obtained, the target content can be found in the search result. If not, it is necessary to repeatedly enter another keyword and the burden on the user increases. It is desirable to reduce the burden of inputting keywords in order to make it possible to repeat such operations smoothly, especially when searching for enormous contents.

  As a method for reducing the burden of inputting a keyword to the user, there are the above-described services on the Internet that allow a user to present and select a plurality of keywords, and the device described in Non-Patent Document 1. However, in the former, a search is possible only by the user selecting a keyword, but the presented keyword is not a keyword that matches the user's personal preference, and the keyword intended by the user is not presented. There's a problem. This is due to the macro trend in web service access analysis, which is not related to individual preferences. In the latter case, if there is no user's favorite keyword among the initial keywords, it may be difficult to reach the favorite content. When any keyword is selected on the initial keyword screen, subsequent keywords are extracted from keywords with high frequency of program information that matches the selected keyword, and thus there are often no keywords preferred by the user. Even if the selection of keywords is repeated, keyword extraction is performed in a manner unrelated to personal preference, so it is highly possible that a favorite keyword is not presented.

  Further, the method of using a related word dictionary when extracting related keywords as disclosed in Patent Document 1 is not sufficient for extracting related keywords that a user wants. For example, if there is a keyword of “meat”, if you want to show related keywords of “meat” as an ingredient, increase the keywords such as “beef”, “chicken”, etc. However, when a keyword related to “meat” as a recipe is desired to be presented, recipe names such as “hamburg” and “steak” are listed as keywords. As described above, there are cases where the user's search intention and the related keywords required by the user differ depending on the search content. Therefore, it is necessary to prepare a related word dictionary for extracting related keywords for each user's search intention and search content, but the user's search intention varies depending on the user and is managed by a dictionary or a table. The fixed related term information thus made cannot cope with various search intentions or search contents. In addition, the method disclosed in Patent Document 2 is also intended to present an expression representing the user's search purpose for the purpose of document search, and has no concept of user preference information. That is, there is a problem that keywords cannot be presented based on hobbies and preferences.

  Another content search method is directory search. In directory search, keywords or contents are classified in advance and a hierarchical database is prepared. First, a plurality of higher-level keywords (classification names) are presented to the user. Present lower keywords. Repeat until you reach the keyword or content you want. At the time of hierarchical structure, the hierarchical structure is determined based on keyword evaluation in linguistics such as keyword thesaurus information, content classification results, etc., but the search intentions of users are various, and each user's intention is Presenting appropriate keywords for each search is difficult with a fixed hierarchical database. In addition, in order to present appropriate keywords in such a fixed hierarchical database, it is necessary to adjust the type and hierarchical structure of the keyword depending on the search target content. However, when the search target content changes, the adjustment is performed again. There must be.

  As described above, it is used in the method of searching for related information between keywords set in advance as keywords to be presented, fixed related word information managed in a dictionary or table, or meta information of content. It is not possible to present keywords that match the preferences of individual users. In practice, it is desirable that the keywords to be presented be close to the user's hobbies / preferences and easy to select.

  On the other hand, a method of accumulating personal preference information and extracting a keyword is conceivable, but there is a problem in that an unintended keyword is presented only by preference information. For example, if a user who is interested in “Soccer” and “Shogi” wants to search for “Shogi” at the present time, it is expected that a keyword close to the purpose will be presented. The method used also presents keywords related to “soccer”. Thus, when the user's preference and purpose do not match, there is a problem that an appropriate keyword cannot be extracted only by storing preference information.

  The present invention has been made to solve the above-described problems, and an object thereof is to provide an apparatus, a method, and a program for extracting a keyword that matches a user's hobbies, preferences, and purposes.

  The present invention in one aspect is a keyword extraction device, a keyword storage unit that stores a plurality of keywords, a related storage unit that stores a degree of association between a plurality of keywords, and a plurality of keywords for each of the plurality of keywords. Preference storage means for storing the degree of preference that correlates with the frequency with which each has been specified, output means for outputting a selection candidate keyword, specification means for specifying a keyword from the keywords output by the output means, Extraction means for extracting a presentation keyword to be output next time from the plurality of keywords based on the degree of association with the keyword designated by the designation means and the degree of preference.

  Preferably, the apparatus further includes an input unit that accepts an instruction from the outside, and a presentation unit that presents the keyword output by the output unit to the user. Specify a keyword from the inside.

  More preferably, it further comprises search means for searching for content associated with the keyword, and the presentation means presents the content searched by the search means.

  Preferably, the apparatus further includes an updating unit that updates the preference degree in accordance with the keyword specified by the specifying unit.

More preferably, the preference storage means stores a plurality of types of preference levels,
Update means for updating each of the plurality of types of preference degrees by different methods is further provided.

More preferably, the preference storage means stores the long-term preference and the short-term preference,
The updating means updates the preference information so that the update rate of the short-term preference degree is larger than the update rate of the long-term preference level.

  More preferably, the updating unit updates the preference level so as to prevent the presentation keyword previously extracted by the extracting unit from being extracted again.

  More preferably, the updating unit lowers the preference level of the presented keyword previously extracted by the extracting unit.

Preferably, the information processing apparatus further includes means for storing the keyword presentation times presented by the presentation means in association with the keywords,
Based on the stored presentation times, the extraction means preferentially extracts the keywords that are not presented during the first suppression count by the presentation means.

More preferably, in response to the instruction, re-presenting means for causing the presenting means to re-present the presentation keyword previously presented by the presenting means;
The apparatus further includes suppression release means for increasing or decreasing the first suppression count depending on whether an instruction to cause the re-presentation means to perform re-presentation or an instruction to cause the extraction means to extract the presentation keyword is given.

Preferably, the apparatus further comprises means for storing the keyword presentation times designated by the designation means in association with the keywords,
Based on the stored presentation times, the extraction means preferentially extracts the keywords that are not designated during the second suppression count by the designation means.

More preferably, in response to the instruction, re-presenting means for causing the presenting means to re-present the presentation keyword previously presented by the presenting means;
The apparatus further includes suppression release means for increasing or decreasing the second suppression count depending on whether an instruction to cause the re-presentation means to perform re-presentation or an instruction to cause the extraction means to extract the presentation keyword is given.

Preferably, the apparatus further comprises session determination means for determining a session indicating an operation unit based on the instruction,
The update unit updates the preference level based on the determination result of the session determination unit.

Preferably, the apparatus further comprises session determination means for determining a session indicating an operation unit based on the instruction,
The extraction means changes the number of times to suppress the keyword presentation according to the result of the session determination means.

  Preferably, the extracting unit extracts the presented keyword randomly from the plurality of keywords in accordance with the instruction.

  Preferably, the extraction unit sequentially extracts all the plurality of keywords in a predetermined order according to the instruction.

  In the present invention according to another aspect, a keyword storage unit that stores a plurality of keywords, a related storage unit that stores a degree of association between a plurality of keywords, and a plurality of keywords have been designated for each of the plurality of keywords. A computer including a preference storage unit that stores a preference degree having a correlation with a frequency, an output unit that outputs a keyword of a selection candidate, a keyword storage unit, a related storage unit, a preference storage unit, and a control unit that controls operations of the output unit. The keyword extraction method for extracting the presentation keyword that is output next time by the output unit using the step of specifying a keyword from the keywords output by the output unit, the keyword specified by the specification unit, Extracting a presented keyword from a plurality of keywords based on the degree of association and the degree of preference.

  In the present invention according to another aspect, a keyword storage unit that stores a plurality of keywords, a related storage unit that stores a degree of association between a plurality of keywords, and each of a plurality of keywords are designated. A preference storage unit that stores a preference level having a correlation with the frequency, an output unit that outputs a keyword of a selection candidate, a keyword storage unit, a related storage unit, a preference storage unit, and a control unit that controls operations of the output unit. A keyword extraction program for causing a computer to extract a presentation keyword to be output next time by the output unit, the step of specifying a keyword from the keywords output by the output unit, and the keyword specified by the specification unit Based on the degree of relevance and the degree of preference, the control unit executes the step of extracting the presented keyword from a plurality of keywords. Make.

  The present invention according to still another aspect is a computer-readable recording medium storing the keyword extraction program described above.

  According to the present invention, the presentation keyword is extracted based on the preference information for the keyword of the user and the related information between the keywords. Therefore, the user does not need to think about the keyword, and it is possible to search with a keyword that more matches the user's hobbies / preferences.

An embodiment of the present invention will be described below with reference to the drawings.
(1. Functional configuration of the present invention)
First, the structure of the keyword extracting device according to the present invention will be described with reference to FIG. FIG. 1 is a block diagram showing the structure of a keyword extracting apparatus according to the present invention. The keyword presentation device includes an input unit 1, an input receiving unit 2, a presentation keyword generation control unit 3, a keyword display creation unit 4, a search keyword storage unit 5, a presentation keyword storage unit 6, a selected keyword storage unit 7, and a display control unit 8. The display unit 9 is provided. Various selection information, request information, and designation information from the user are input to the input unit 1. The input reception unit 2 receives information from the input unit and passes the information to each processing unit according to the input information. The presentation keyword generation control unit 3 extracts a keyword to be presented from the search keyword storage unit 5 that stores the keyword to be presented to the user based on a predetermined method, and stores the extracted presentation keyword in the presentation keyword storage unit 6. The keyword display creation unit 4 creates a display screen with the extracted presentation keywords. For example, a search button for display is generated with the keyword extracted by the presented keyword generation control unit 3. The selected keyword storage unit 7 stores a search keyword selected by a user's operation. The display control unit 8 receives the display screen created by the keyword display creation unit 4 and displays it on the display unit 9. However, the keyword extraction device itself may not have the display unit. The keyword extracting device only needs to include a portion for extracting and outputting the presented keyword.

  FIG. 2 is a block diagram of a search support apparatus incorporating the keyword extraction apparatus. The search support apparatus includes an input unit 1, an input receiving unit 2, a presentation keyword generation control unit 3, a keyword display creation unit 4, a search keyword storage unit 5, a presentation keyword storage unit 6, a selected keyword storage unit 7, and a display control unit 8. In addition to the display unit 9, a search request unit 10, a search unit 12, a search result acquisition unit 13, and a search result display creation unit 14 are provided. In response to an instruction from the input receiving unit 2, the search request unit 10 creates a search expression for performing a search in the search unit 12 and sends a search request to the search unit 12. Here, the search formula is created according to, for example, the keyword attached to the search button selected by the user among the search buttons created by the keyword display creation unit 4. In response to a search request from the search request unit 10, the search unit 12 searches for content related to the keyword. Note that this search may be performed in an apparatus that performs search support or may be performed in an external apparatus connected via a network. The search result acquisition unit 13 acquires the search result searched by the search unit 9. The search result display creation unit 14 creates search result information for display from the search results acquired by the search result acquisition unit 13. In the following, an embodiment related to a content search support apparatus will be described.

(2. System configuration of the present invention)
The keyword presentation device according to the present invention is substantially realized by a computer system including computer hardware, a program executed by the computer hardware, and data stored in the computer hardware. Hereinafter, this computer system will be described with reference to FIG. FIG. 3 is a diagram illustrating an example of a computer system for realizing the present invention.

  Referring to FIG. 3, this computer system 200 includes a computer 220, a monitor 203a, and a remote control 204.

  The computer 220 includes an optical disk drive 201, a magnetic disk drive 202, a display interface (I / F) 203b, a remote control interface (I / F) 205, a CPU (central processing unit) 206, a ROM (read-only memory, A read-only memory (208), a RAM (random-access memory) 209, and a bus 207 for connecting them together. Although not shown here, the computer 220 may further include a network adapter board that provides connection to a network, a communication device that performs wireless communication, and the like.

  The remote control interface 205 receives a signal from the remote control 204. The ROM 208 stores a bootup program and the like. The RAM 209 stores program instructions, system programs, work data, and the like. An optical disk 210 is inserted into the optical disk drive 201. A magnetic disk 211 is inserted into the magnetic disk drive 202.

  A program for causing the computer system 200 to realize the function of the search support apparatus and input information are both stored in the optical disk 210 or the magnetic disk 211 and further transferred to the hard disk 212. Alternatively, the program and input information are transmitted to the computer 220 through a network (not shown) and stored in the hard disk 212. The program stored in the hard disk is loaded into the RAM 209 at the time of execution. Alternatively, the program may be loaded directly into the RAM 209 from the optical disk 210, from the magnetic disk 211, or via a network, without going through the hard disk 212.

  Note that a program for realizing the function of the search support device and a recording medium on which input information is recorded are a CD-ROM (Compact Disc Read Only Memory), an MO (Magneto-Optical disc), an MD (Mini Disc), and It is not limited to the optical disk 210 such as a DVD (Digital Versatile Disc), the magnetic disk 211 such as an FD (flexible disk) and a hard disk. For example, a tape such as a magnetic tape or a cassette tape, an IC (Integrated Circuit) card or an optical card Type recording medium, mask ROM, EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Program) ble ROM), and it may be a semiconductor memory such as a flash ROM. However, the computer system 200 needs to be equipped with a reading device for reading from these recording media that store programs.

  This program includes a plurality of instructions for causing the computer 220 to realize the function of the search support apparatus according to the present embodiment. Some of the basic functions necessary to realize these functions are provided by operating system (OS) or third-party programs running on the computer 220 or modules of various toolkits installed on the computer 220. Therefore, this program does not necessarily include all the functions necessary for realizing the functions of the search support apparatus according to this embodiment. If this program contains only the instruction that executes the control of the search support device described above by calling an appropriate function or “tool” in a controlled manner so that a desired result is obtained. Good.

(3. Keyword presentation)
Next, keyword presentation according to the present invention will be described with reference to FIG. FIG. 4 is a diagram illustrating an example of a screen displayed on the monitor 203a. The display screen includes a search keyword 401, a presentation keyword 402, a search result 403, a PREV button 404, a NEXT button 405, a Genre button 406, and a switching button 407. However, FIG. 4 shows an example of a screen generated for display, and the display of the search keyword 401, the presentation keyword 402, the operation button arrangement, and the search result 403 can be arbitrarily changed.

  The search keyword 401 is a keyword specified by the user for searching for content. FIG. 4 shows an example screen when “news” is designated as the search keyword 401.

  The search result 403 is an image corresponding to the content searched by the search unit 12 according to the search keyword 401. In FIG. 4, images corresponding to six of the contents searched based on the keyword “news” are displayed. When an image is selected from the search results 403, content can be viewed.

  In addition, when there are a large number of contents searched according to the search keyword 401, only a part of them is displayed, and by selecting the switching button 407, the contents not displayed are changed to the contents currently displayed. It may be displayed.

  When the search keyword 401 is designated, the presentation keyword 402 is stored in the search keyword storage unit 5 based on the degree of association with the search keyword 401, the user's preference for each keyword, the keyword presented previously, and the like. The keywords extracted from the generated keywords and generated for display. Details of the keyword extraction method will be described later. In this figure, there are nine presentation keywords 402, but other numbers may be used, and the arrangement of keywords included in the presentation keywords 402 is not limited. For example, they may be arranged in a line.

  The user can select the next search keyword 401 from the presented keywords 402. When selection information for selecting a keyword from the presented keywords 402 is input to the input unit 1 by the user, the selected keyword is stored in the selected keyword storage unit 7 via the input receiving unit 2 as follows. It is stored as a search keyword 401. The method for determining the selection of the search keyword by the user is not particularly limited. For example, as shown in FIG. 4, when the remote control 204 includes an input device that can control focus movement in the cross direction, the focus of the search keyword is moved by the operation in the cross direction, and the search keyword of the favorite keyword is set. Select a keyword by focusing and selecting.

  According to the method of the present invention, by selecting and presenting the presentation keyword 402 based on information different for each user, such as operation history and preference, the user does not need to think about the keyword, and the user's hobbies / preferences can be improved. It is possible to search with a matched keyword. By performing the selection process, the user can present a keyword that matches the user's preference and is closer to the currently requested content.

  In this method, since the presented keyword is not fixed, the user's search intention is easily reflected. In order to present appropriate keywords in a fixed hierarchical database used in directory searches, it is necessary to adjust the keyword type and hierarchical structure according to the search target content. Must be done. On the other hand, according to this method, it is possible to present a keyword suitable for the user's search intention without limiting the search target content.

  Further, according to the selection information, the input receiving unit 2 requests the presentation keyword generation control unit 3 to generate the next presentation keyword 402. The presentation keyword generation control unit 3 stores the generated keyword in the presentation keyword storage unit 6, and the keyword display creation unit 4 generates the presentation keyword 402. Further, according to the selection information, the search request unit 10 creates a search formula and makes a search request to the search unit 12. And the search result acquisition part 13 acquires a search result, and the search result display preparation part 11 produces the search result 403 in FIG. 4 which is the search result information for a display.

  The display control unit 8 controls the display of the search keyword 401, the presentation keyword 402, and the search result 403, and causes the display unit 9 to display them. That is, when the search keyword 401 is designated, both the search result 403 based on the search keyword 401 and the presentation keyword 402 that is a candidate for the next search keyword are displayed. Therefore, it is possible to reduce the labor of user input. However, for example, the configuration may be such that after the next search keyword 401 is presented, the user operates the screen to request the extraction of the next presentation keyword 402 or the search request. The user can repeat the search of the content by viewing the search result displayed if he / she likes it, and repeating the selection of another keyword presented if he / she does not like it.

  In the search support apparatus of the present invention, there are cases where the search keyword 401 is not specified at the time of the first search (at the time of initial startup) or the like. In this case, the presented keyword 402 is set to a predetermined fixed keyword that is presented in accordance with a call request from the user regardless of the previously presented keyword, the degree of preference, the degree of association, etc., and the user searches among them. The keyword 401 may be selected. In particular, if a fixed keyword is a keyword having a broad meaning that is used for a category name or the like, a plurality of categories are presented at the time of the initial search, so that the user can easily select the keyword.

  Alternatively, it is possible to request profile input such as the user's age, gender, favorite field, etc., and select a keyword to be presented based on the input result. It is also possible to categorize by age, gender, etc., and set initial presentation keywords for each category in advance. However, at the time of activation other than the first time, it is preferable that the presentation keyword is extracted based on a predetermined keyword extraction method.

  Further, the configuration may be such that the user can directly specify the search keyword 401 by key input or the like. Further, at that time, when the user inputs some characters, a keyword that matches the first character may be extracted from the search keyword storage unit 5 and displayed on the display unit 9. With this configuration, the user's trouble of key input is reduced. However, when the presentation keyword 402 is extracted and presented, it is preferable to select the search keyword 401 from among them.

  The NEXT button 405 is a button that is selected when the user requests a keyword different from the keyword being presented. When this button is selected, the presentation keyword generation control unit 3 extracts the number of keywords that are not presented from the keywords stored in the search keyword storage unit 5 and the keyword display creation unit 4 newly A display keyword 402 for display is generated for the extracted keyword. The display control unit 8 causes the display unit 9 to display a newly generated presentation keyword instead of the currently displayed presentation keyword. The newly presented keyword is stored in the presented keyword storage unit 6. If there is no appropriate keyword among the presented keywords, it is easily switched to another keyword, so that new search keywords are presented one after another, making it easy to move on to the next search.

  The PREV button 404 is a button that is selected when the presentation keyword before the change is presented again when the presentation keyword is changed by the NEXT button 405. When this keyword is selected, the presented keyword generation control unit 3 extracts the keyword presented previously from the presented keyword storage unit 6, and the keyword display creation unit 4 displays the presented keyword 402 for display with respect to the extracted keyword. Is generated. Then, the display control unit 8 causes the display unit 9 to display the extracted presentation keyword. The keyword presented again by the above processing may be stored in the presented keyword storage unit 6. By using the PREV button 404, the user can confirm again the keyword presented previously.

  In addition, when the presentation keyword is not changed by the NEXT button 405, the fixed keyword may be displayed when the PREV 404 is selected. Alternatively, the display control unit 8 may not display the PREV button 404 on the display unit 9. Further, although it is displayed, the PREV button 404 may not be selected. FIG. 4 is a diagram corresponding to such a case, and the PREV button 404 is displayed lightly to indicate that the user cannot select.

  The display of the search result 403 when the display of the presentation keyword 402 is changed by the PREV button 404 and the NEXT button 405 is not limited. The display of the search result 403 may not be changed before and after the display of the presented keyword 402 is changed, or another search result may be displayed.

  The Genre button 406 is a keyword that is selected when the user requests a fixed keyword. The fixed keyword refers to a predetermined keyword that is presented in accordance with a call request from the user regardless of the keyword, the degree of preference, the degree of relevance, and the like presented last time. The search keyword storage unit 6 also stores information on whether or not the keyword is a fixed keyword together with the keyword.

  When the Genre button 406 is selected, the presentation keyword generation control unit 3 extracts a fixed keyword from the search keyword storage unit 6, and the keyword display creation unit 4 generates a display keyword 402 for display of the fixed keyword. Then, the display control unit 8 displays a fixed keyword instead of the current presentation keyword 402. At this time, the display of the search result is not limited. The display of the search result may not be changed before and after the display of the search keyword is changed, or another search result may be displayed.

  In the method of extracting a keyword related to the keyword selected last time, if a search keyword is repeatedly selected, the keyword to be presented may be biased to some fields. In addition, the user's search intention may change during repeated searches, and the user may want to search with keywords in a completely different field that are not related to the currently presented keyword group. In this case, a fixed keyword is presented as a presentation keyword regardless of the previous selection keyword, thereby facilitating a search in another field.

  The fixed keyword is not limited, but may be, for example, a keyword used for an upper layer keyword, a category name for classifying content, a field name, a genre name, or the like among keywords hierarchized according to linguistic meaning.

(4. Processing performed by the present invention)
FIG. 5 is a flowchart showing the flow of processing of the search support apparatus according to the present invention. The specific operation of the present invention will be described with reference to this figure.

  First, in step S1, the CPU 206 determines whether or not there is an input to the input unit 1 (for example, the remote control I / F 205). When it is determined that there is no input in the input unit 1 (No in step S1), the process of step S1 is repeated.

  If it is determined that there is any input in the input unit 1 (Yes in step S1), the CPU 206 determines in step S2 whether the input is selection information indicating designation or selection of the search keyword 401 by the user. .

  If it is determined that selection information has been input (Yes in step S2), CPU 206 performs learning at the time of keyword selection, that is, updates preference information in step S3. Details of the preference information update method will be described later.

  After updating the preference information, the CPU 206 instructs the search unit 12 to execute content search using the selected keyword in step S4. Whether the content search is performed is within the apparatus executing the content search support or an external apparatus connected via a network does not affect the present invention.

  Subsequently, in step S5, the CPU 206 acquires the search result obtained in step S4, and causes the display unit 9 to display the acquired search result in step S6.

  Subsequently, in step S7, the CPU 206 performs presentation keyword extraction processing by a predetermined method. The presentation keyword extraction process will be described later. CPU206 memorize | stores the keyword extracted in step S7 in the presentation keyword memory | storage part 6 as an extracted keyword (step S8). In step S <b> 9, the CPU 206 displays the stored presentation keyword on the display unit 9. With this processing, a series of processing ends. After the process is completed, the process from step S1 can be repeated.

  Here, it is described that when a search keyword is specified, content search and display (steps S4 to S6) are performed before extraction and display of presentation keywords (steps S7 to S9). However, this order does not matter, and when the present invention is not incorporated in the content search apparatus and is simply used for presenting keywords, the processing in steps S4 to S6 is not necessary.

  When CPU 206 determines that the input information is not search keyword selection information (No in step S2), CPU 206 determines whether another keyword request is made in step S10, that is, whether NEXT button 405 has been selected.

  If the CPU 206 determines that the input information is another keyword request (Yes in step S10), the CPU 206 performs learning at the time of another keyword request, that is, updates the preference in step S11. Details of the preference update method will be described later. After the learning process, the processes from step S7 to step S9 described above are performed.

  If the CPU 206 determines that the input information is not a request for another keyword in step S10, the CPU 206 determines in step S12 whether the input information is a past keyword request, that is, whether the PREV button 404 has been selected.

  If it is a past keyword request (Yes in step S12), the CPU 206 performs learning at the time of the past presentation keyword in step S13, that is, updates the preference information described later, and in step 14 determines whether there is a past presentation keyword. judge.

  If there is a keyword presented in the past (Yes in step S14), CPU 120 extracts the keyword presented once in step S15, and presents the keyword to the user in step S9.

  If it is determined in step S14 that there is no keyword presented in the past (No in step S14), CPU 206 extracts a fixed keyword determined in advance in step S18. And the extracted fixed keyword is memorize | stored in the presentation keyword memory | storage part 6 (step S8), and it is made to show on the display part 9 by step S9. In this flowchart, the CPU 206 presents a fixed keyword when there is no keyword presented in the past, but the CPU 206 does not change the presented keyword and displays a message that there is no past keyword to be presented. May be.

  If the CPU 206 determines in step S12 that the input information is not a past keyword request (No in step S12), it determines in step S16 whether the input information is a fixed keyword request.

  If it is determined that the request is a fixed keyword request (Yes in step S16), the CPU 120 performs learning at the time of the fixed keyword request in step S17, that is, updates preference information described later, and then extracts a fixed keyword in step S18. Subsequently, after the presentation keyword is stored in step S8, the presentation keyword is displayed on the display unit 9.

  If it is not a fixed keyword request (No in step S16), the process designated by the user is executed in step S19. If the designated process is a content viewing instruction, display control is performed so that the content can be viewed. If the designated process is an end operation, the content search support process is terminated.

(5. Method for extracting presentation keywords)
Here, a method used when the presentation / presentation keyword generation control unit 3 extracts presentation keywords will be described.

  In the present invention, when extracting the presentation keyword, preference information for the keyword of the user and the degree of association information between the keywords are used. The preference information and the relevance information are stored in the search keyword storage unit 5 for each keyword.

  In the present embodiment, the relevance information includes the relevance of each keyword with other keywords as shown in FIG. Each keyword is identified by an ID. For example, the keyword “sports” has a relevance level of “music” of 20, a relevance level of “concert” of 10, a relevance level of “baseball” of 50, a relevance level of “soccer” of 50, and “fishing”. The degree of association with “Japan” is 30, and the degree of association with “Japan representative” is 20. Assume that the higher the relevance value, the greater the degree of relevance.

  In the present embodiment, the relevance information is given in advance. For example, it may be given using a relevance dictionary as described in Patent Document 1. However, the relevance information need not be fixed. For example, it is possible to search for content including both of the two keywords and use the number of searched content to determine the content.

  In addition, the degree of relevance may be calculated each time a presentation keyword is extracted. Specifically, as shown in FIG. 8, an element indicating a feature amount is held in keyword units. As the degree of association between a certain keyword and another keyword, the similarity of this feature amount can be used. That is, the difference in the feature amount is obtained for all the elements between the keywords in FIG. 8, and the similarity is obtained from this difference. For example, a value obtained by multiplying the sum of the squares of the difference between the feature quantities by a constant so that the value falls within a certain range and normalized can be used as the similarity. In this case, the relevance is higher as the sum is smaller. As an advantage of this method, it is not necessary to store the degree of association between all the keywords, so that there is an effect that the capacity for storing the degree of keyword association can be reduced. When the number of keywords is N, if the degree of association between each keyword and another keyword is simply expressed, N (N-1) degree of association values are required. On the other hand, in the case of this method, if the number of feature quantities is k, the number of data to be stored for calculating the relevance is Nk. Therefore, if k <N−1, the number of data, and hence the storage capacity to be used is set. Can be reduced.

  Also, the preference information for the user's keyword is determined based on the specification of the search keyword by the user. Therefore, when the user repeats the keyword selection operation, keywords closer to the user's preference can be extracted and presented. In this embodiment, as shown in FIG. 8, the preference information includes items such as a long-term preference level, a short-term preference level, the most recent presentation count, the most recent selection count, and the last selection for each keyword.

  The preference level is a value having a correlation with the selection frequency of the user. Here, the higher the preference value, the greater the degree of preference. That is, it is assumed that a keyword having a higher selection frequency by the user is updated to have a higher value.

  In this embodiment, it is assumed that there are two types of preference levels, the long-term preference level and the short-term preference level. However, the type of preference is not limited to this. For example, only one kind of preference degree may be given to one keyword. There may be three or more types of preference.

  Further, in FIG. 7, the preference value is quantified in units of keywords, but may be held by other methods. For example, keywords having the same characteristics may be grouped to have a value for each group.

  The most recent number of presentations indicates how many times the keyword extraction process was performed when the keyword was presented as the last presentation keyword. For example, up to now, ten keyword extractions have been performed by the presentation keyword generation control unit, and when the keyword is extracted and presented by the second, fifth, and eighth keyword extractions, , 8 The most recent presentation count indicates how many times the keyword is selected as the search keyword at the time of the keyword extraction process. For example, up to now, 10 keyword extractions have been performed by the presentation keyword generation control unit, and the keyword has been extracted and presented in the second and fifth keyword extractions, and has been selected by the user. The number of selections is 5. It is assumed that the number of keyword extraction processes necessary for determining the most recent presentation count and the most recent selection count is stored in the presentation keyword storage unit 6, for example.

  The last selection indicates that the selection was performed in the previous operation, “1” is given to the keyword selected immediately before, and “0” is given to the other keywords.

  FIG. 7 shows a case where sports and music are extracted as presentation keywords in the first keyword extraction, are presented, and sports are selected as the next search keywords.

  The presented keyword generation control unit 3 preferentially extracts keywords with higher preference and keywords with higher relevance. In this embodiment, for each keyword, a score defined by (score) = A × (long-term preference) + B × (short-term preference) + W1 × (relevance) is calculated, and a score having a high score is given priority. It shall be extracted. Here, A, B, and W1 are weights. In the following description, A = 1, B = 1, and W1 = 5, but the values are not limited to this.

  Now, assume that the keyword “sports” is selected by the user. When “sports” is passed as the keyword information of the selected search keyword, the presented keyword generation control unit 3 refers to the search keyword storage unit 6, and relates to the “sports” for each keyword and the keyword itself. Points are calculated from the degree of preference. Specifically, in the case of the keyword “baseball”, since the long-term preference for baseball is 50, the short-term preference is 100, and the relevance to baseball sports is 25, the score is 50 + 100 + 5 × 25 = 275. . Similarly, a score is calculated for each keyword, sorted in descending order of score, and stored in the presented keyword storage unit 7. It is good to extract the high-order keywords when sorted in descending order of the score obtained as described above. An arbitrary threshold value may be provided, and a keyword that exceeds the threshold value may be extracted.

  In addition, when extracting a presentation keyword, it is also possible to suppress presentation of a keyword that has been presented as a presentation keyword in the past n times or selected as a search keyword in the past m times. Here, n and m can be set to arbitrary numbers.

  In order to perform the above-described presentation suppression, the latest presentation count and the latest selection count are used. The presentation suppression process will be described with reference to FIG.

  When receiving an instruction for extracting the presentation keyword, the CPU 206 updates the keyword extraction count N, the most recent presentation count, and the most recent selection count stored in the search keyword storage unit 5 in step S201.

  Next, in step S <b> 202, the CPU 206 extracts the latest presentation number n ′ of each keyword.

  In step S203, the CPU 206 compares N−n ′ and the magnitude of n. When N−n ′ ≦ n (Yes in step S203), the CPU 206 sets the score of the keyword to 0 in step S206. This is because when N−n ′ ≦ n, the keyword has been presented as a presentation keyword in the past n times. In step S206, the score is set to 0. However, this process is not limited to this, and any process that suppresses the presentation of the keyword may be used. A process of applying a constant of 0 or more and less than 1 or subtracting the constant may be used.

  When N−n ′ ≦ n is not satisfied (No in step S203), the process proceeds to step S204 and subsequent steps. In step S204, the CPU 206 extracts the most recent presentation count m ′ of each keyword.

  In step S205, the CPU 206 compares N−m ′ and the magnitude of m. When N−m ′ ≦ m (Yes in step S205), the CPU 206 sets the score of the keyword to 0 in step S206. This is because when N−m ′ ≦ m, the keyword has been selected as the search keyword in the past m times. When N−m ′ ≦ m is not satisfied (No in step S205), the keyword score is not changed.

  As described above, it is possible to reduce the number of operations until a keyword closer to the target content is reached by suppressing the repeated presentation of the same keyword based on the operation history.

  In the present embodiment, the most recent presentation (or selection) count is defined as indicating the number of times the keyword was extracted (or selected) at the end of the keyword extraction process. Although processing is performed, it is also possible to suppress keyword presentation by other methods. For example, it is possible to define the number of times of the most recent presentation (or selection) as indicating how many times it was presented (or selected) before, and to compare with n (or m), it is possible to suppress the fixed time. . In this case, for example, in step S105, the CPU 206 updates the preference level and updates the latest selection count. The most recent selection count is updated so that the most recent selection count of the keyword selected immediately before is incremented by 1, the most recent selection count of the previously selected keyword is increased by 2, and thereafter. Similarly, the latest number of presentations is updated.

  In addition, the number (n, m) of suppressing the keyword presentation can be dynamically changed. For example, the CPU 206 suppresses the keyword presentation depending on the number of times that the operation of presenting the “previous keyword” by selecting the PREV button 404 and the operation of presenting the “other keyword” by selecting the NEXT button 405 are continuously performed. The number of times n and m is reduced. When the search keyword is designated, the CPU 206 returns n and m to predetermined values. Specifically, for example, every time the PREV button 404 or the NEXT button 405 is selected without specifying a search keyword, or every time the PREV button 404 or the NEXT button 405 is selected s times (s ≧ 2). Decrease both n and m by 1. At this time, lower limits of n and m (for example, 2 etc.) may be set.

  Alternatively, every time a keyword selection instruction from the presented keywords is selected s times (s ≧ 2), both n and m can be reduced by one. At this time, it is desirable to determine the upper limit values of n and m in advance.

  Performing operations such as “previous keyword” and “other keyword” without selecting a keyword presented by the user includes a state of searching for a keyword presented previously. In addition, the fact that only the keyword selection is continued is presumed to be a state where the user does not seek the keyword presented before. Accordingly, the number of times of not presenting in this way is dynamically changed, that is, an instruction to re-present the “previous keyword” and “other keywords”, or a keyword is designated from the presented keywords and a new presented keyword is selected. By increasing or decreasing the number of times the previously presented keyword is not presented depending on which of the instructions to be extracted is made, it is possible to make it easier for the user to reach the desired keyword.

  When extracting keywords, in addition to using the score as described above, use a method of extracting keywords using random values, or a method of extracting all the presented keywords in order in a predetermined order. Is also possible. The random extraction and the method of extracting all the keywords in order are preferably performed according to a user instruction.

  Further, it is possible to select only a part of the keywords presented to the user at random. At this time, it is preferable that a keyword selected at random is extracted from a keyword layered according to a linguistic meaning, because a higher range of keywords is included, so that a user can select more frequently. . For example, a keyword covering a wider area such as “car” is preferable to a keyword having a limited meaning such as “sport car”.

  It is possible to provide an opportunity for the user to select a new keyword different from the preference information by random presentation of keywords or list presentation. This has the effect of providing a new discovery opportunity that prevents bias due to user preferences.

(6. Preference information update method)
In the present invention, the preference information is updated according to the user's operation. Therefore, it is possible to present a keyword closer to the user's preference by repeating the operation by the user. Further, when the same keyword is selected at a different point in time, the preference information has been updated since the previous operation, and therefore does not necessarily match the keyword presented at the previous selection. As a result, different keywords can be presented while matching the user's hobbies, and the user can be guided to a wider range of content by performing a search based on different keywords.

  FIG. 10 is a flowchart of preference learning according to a user operation performed by the search device according to the present invention, that is, updating of preference information. A method for updating preference information will be described with reference to this flowchart.

  First, in step S <b> 101, the CPU 206 determines whether or not preference information is stored in the search keyword storage unit 5. If the CPU 206 determines that there is stored preference information (Yes in step S101), the CPU 206 reads the saved preference information onto the ROM 208 in step S102. If it is determined that there is no saved preference information (No in step S101), the preference information initialized in step S103 is created.

  In step S104, the CPU 206 determines whether the user has selected a keyword. This process corresponds to the determination performed in step S2, step S10, step S12, and step S16 in FIG. If it is determined in step S2 in FIG. 5 that the search keyword has been selected from the presented keywords, the result is Yes in step S104. A search keyword is not selected from the presented keywords, but when an instruction for extracting another keyword (another keyword, a past keyword, a fixed keyword) is given, that is, in steps S10, S12, and S16. A case where the determination is Yes corresponds to No in step S104.

  When it is determined that a keyword has been selected (Yes in step S104), the CPU 206 performs the processes of steps S105 and S106.

  In step S105, learning about the selected keyword, that is, preference information of the selected keyword is updated. An example of the update method is shown below.

  In the present embodiment, the long-term preference level and the short-term preference level are updated by the following equations (1) and (2), respectively.

  Here, t is a time parameter (natural number) indicating the number of times of presentation keywords, L (t) is a long-term preference level, and S (t) is a short-term preference level. WL is a long-term preference learning constant, Ws is a short-term preference learning constant, and each takes a value between 0 and 1. In this method, in order to increase the update rate of the short-term preference, it is desirable that Ws> WL. α is a preference constant. Since the asymptotic value of the selected keyword is desirably higher than the asymptotic value of the related keyword, α is desirably a value larger than the maximum value of R. In addition, α is preferably the maximum value that the preference level can take so that the preference level of the selected keyword does not decrease. In this embodiment, as will be described later, it is assumed that α = 100 in order to normalize the maximum degree of preference to 100.

  However, the method of updating the preference level is not limited to this method. In general, the preference level of the selected keyword is increased so that the preference of each keyword of the user is reflected in the preference information. Update it. For example, an update method such as adding a constant to the preference level of the selected keyword may be used.

  Next, in step S106, related keywords are learned. Specifically, the long-term preference level and the short-term preference level are updated by the following formulas (3) and (4) for each keyword.

  Here, R is the degree of association between the selected keyword and the keyword to be learned. By updating the preference level of the keyword related to the keyword selected in this way, the preference of the user is reflected more. However, it is not always necessary to realize the processing of the present invention.

  Next, in step S107, the CPU 206 learns keywords that have not been selected from the presented keywords. In this embodiment, in order to suppress the presentation of unselected keywords, updating is performed so as to lower the preference level by the following equation (5).

  Here, Wx is a short-term preference limiting constant and takes a value between 0 and 1. By using Expression (5), it is possible to reduce the short-term preference of keywords that are presented but not selected. Here, only the short-term preference level is updated, but the long-term preference level may be updated by a formula similar to formula (5) by providing a long-term preference limit constant Wy. However, in that case, it is preferable that Wx> Wy so that the influence of the short-term preference is weakened quickly. Further, the expression (5) may be applied to all presented keywords.

  By updating the preference level in this way, it is possible to prevent the same keyword as the previously presented keyword from being extracted as a presentation keyword and repeatedly presented to the user in the subsequent processing. By suppressing the repeated presentation of the same keyword, the number of operations until reaching a keyword closer to the target content can be reduced.

  By having two preferences, a long-term preference that is maintained for a long period of time and a short-term preference that loses its influence in a short period of time, it is possible to strongly reflect the user's intention at the time the keyword is presented Thus, more appropriate keywords can be presented, the time to reach the content can be shortened, and at the same time, long-term preferences can be learned, and appropriate keywords can be presented even when the search assistance is executed next time. It is possible to present more appropriate keywords by handling the preference level divided into the short-term preference level that is presumed to be the current preference and the long-term preference level that is presumed to be the user's preference. .

  In this embodiment, the constants WL = 0.01, Ws = 0.05, and Wx = 0.99 are used, but these values may be other values. Further, it may be dynamically changed based on the operation history of the user such as the number of times of presentation t and the number of times of selection.

  In step S108, normalization of short-term preference and long-term preference is performed. For example, simple normalization is performed by multiplying constants so that the maximum value becomes 100 for both the long-term preference and the short-term preference. Finally, the preference information data updated in step S109 is stored in the preference information storage unit.

  In the above, an example of a method of updating differently for the short-term preference level and the long-term preference level based on the operation history of the user has been shown. In this method, by increasing the types of learning constants and limit constants, even when there are three or more types of preference levels, different updates are made for each type of preference level. Can do. Thereby, a user's preference can be reflected more correctly. Moreover, the preference degree may be one type.

  Moreover, this content search assistance apparatus may perform session determination, and may update preference information based on the result. Here, a session indicates a unit of operation, and one session is a period from when the apparatus receives an instruction to start using the search support function and receives an instruction to end use of the search support function.

  The instruction to start using the search support function is, for example, an instruction for starting a search support application. The instruction to end use of the search support function is, for example, an instruction to start viewing content, an instruction to end an application, or an instruction to perform a search restart operation. Also, an instruction for requesting a fixed keyword indicating a genre may be used as an instruction to end use of the search support function. The CPU 206 accepts these usage start or usage end instructions and determines a session.

  If the CPU 206 determines that the session has ended, it updates the preference information for each keyword. For example, the short-term preference of each keyword is reduced. In particular, it is possible to initialize the short-term preference level and completely eliminate the influence of the past. On the other hand, by not updating the long-term preference level, the influence of short-term preference information is weakened, and the presentation keyword is extracted based on the user's general preference when using the next content search support. It becomes possible. Since the user's keyword selection is strongly reflected in the short-term preference in the session, the time to reach the target keyword or content can be shortened.

  When an operation such as resuming a search is performed, it is considered that the target content cannot be reached at the present time. Therefore, by updating the preference information in that case, it is possible to resume a new search and reach the content desired by the user with a few operations.

  In addition, it is determined that a series of searches has been completed as a result of the user viewing the content, and the preference information is updated and stored, so that the user's preference information is reliably stored and stored, and the user's preference is The next new keyword presentation reflected can be executed.

  When it is determined that the session is ended, the CPU 206 may initialize the latest presentation count and the latest selection count. By canceling the suppression of the keyword selection by the previous selection, it is possible to extract the presented keyword according to the general preference of the user at the next use of the content search support. When there is a specific operation, for example, an operation that intends to redo the search, by canceling the suppression of the presentation, it is possible to realize the start of the search according only to the preference regardless of the past operation. This realizes both restart of search and continuous search.

(7. Other embodiments)
In the present embodiment, the example in which the present invention is applied to the content search support apparatus has been described. However, the present keyword presentation apparatus can also be applied to other apparatuses and systems.

  For example, in the case of searching various dictionaries such as Japanese-English, English-Japanese, Japanese, etc., the present invention can be applied when it is desired to know other related keywords after inputting a search keyword. In this related keyword search, it is desirable to increase the number of times of presentation suppression so that a keyword selected in the past is not re-presented and to have a function of easily listing the keywords selected in the past.

  It can also be used for selecting data stored in a hard disk connected to a computer or a database connected via a network. In addition to papers and electronic books, this data includes various statistical information and product specification tables. At this time, it is efficient that the relevance information of the keyword is obtained from the text information of the data included in the database. For example, as the degree of association between a certain keyword and another keyword, a probability value simultaneously included in a certain content can be used. The present invention can also be applied to a case where an appropriate keyword is searched for a target for which a keyword needs to be newly set, such as a content title or catch phrase.

  The present invention can be used for a search support technique for multimedia contents existing in large quantities.

It is a block diagram which shows the structure of the keyword extraction apparatus which concerns on this invention. It is a block diagram of the content search support apparatus incorporating this keyword extraction apparatus. It is a figure which shows an example of the computer system for implement | achieving this invention. It is a figure which shows an example of the screen displayed on the monitor 203a. It is a flowchart figure which shows the flow of a process of the search assistance apparatus which concerns on this invention. It is a figure which shows an example of the memorize | stored relevance information. It is a figure which shows an example of the memorize | stored feature-value. It is a figure which shows an example of the stored preference information. It is a flowchart which shows the flow of the process for keyword presentation suppression. It is a flowchart which shows the flow of the process performed at the time of keyword learning.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Input part, 2 Input reception part, 3 Presentation keyword production | generation control part, 4 Keyword display production part, 5 Search keyword memory | storage part, 6 Presentation keyword memory | storage part, 7 Selection keyword memory | storage part, 8 Display control part, 9 Display part, DESCRIPTION OF SYMBOLS 10 Search request part, 11 Specification information memory | storage part, 12 Search part, 13 Search result acquisition part, 14 Search result display preparation part, 200 Computer system, 201 Optical disk drive, 202 Magnetic disk drive, 203a Monitor, 203b Display interface, 204 Remote control 205 Remote control interface, 206 CPU, 207 bus, 208 ROM, 209 RAM, 220 Computer, 401 Search keyword, 402 Presented keyword, 403 Search result, 404 PREV button, 405 NEXT button, 06 Genre button, 407 switching button.

Claims (19)

Keyword storage means for storing a plurality of keywords;
Association storage means for storing the degree of association between the plurality of keywords;
For each of the plurality of keywords, preference storage means for storing a degree of preference having a correlation with the frequency with which each of the plurality of keywords has been designated;
An output means for outputting selection candidate keywords;
Designation means for designating a keyword from among the keywords output by the output means;
A keyword extraction device comprising: an extraction unit that extracts a presentation keyword that is output next by the output unit from the plurality of keywords based on the degree of association with the keyword specified by the specifying unit and the preference level . An input means for receiving instructions from the outside;
Further comprising a presentation means for presenting the keyword output by the output means to the user,
The keyword extracting device according to claim 1, wherein the specifying unit specifies a keyword from among the presented keywords presented by the presenting unit in accordance with the instruction. And further comprising search means for searching for content associated with the keyword,
The keyword extraction device according to claim 2, wherein the presenting unit presents the content searched by the search unit.   4. The keyword extracting device according to claim 1, further comprising an updating unit configured to update the preference degree in accordance with a keyword specified by the specifying unit. 5. The preference storage means stores a plurality of types of preference levels,
The keyword extracting device according to claim 4, further comprising updating means for updating each of the plurality of types of preference degrees by a different method. The preference storage means stores a long-term preference level and a short-term preference level,
The keyword extraction device according to claim 5, wherein the updating unit updates the preference information so that the update ratio of the short-term preference degree is larger than the update ratio of the long-term preference degree.   The keyword according to any one of claims 4 to 6, wherein the updating unit updates the preference degree so as to suppress the presentation keyword previously extracted by the extracting unit from being extracted again. Extraction device.   The keyword extracting device according to claim 7, wherein the updating unit lowers a preference level of the presented keyword previously extracted by the extracting unit. Means for storing the keyword presentation times presented by the presenting means in association with the keywords;
The said extraction means extracts the said presentation keyword preferentially for the keyword which is not presented by the said presentation means during the 1st suppression frequency based on the said stored presentation time. Keyword extractor. Re-presenting means for causing the presenting means to re-present the presenting keyword previously presented by the presenting means in response to the instruction;
The apparatus further comprises means for increasing or decreasing the first number of suppressions depending on whether an instruction to cause the re-presentation means to perform the re-presentation or an instruction to cause the extraction means to extract a presentation keyword is made. The keyword extraction device described in 1. Means for storing the keyword presentation times designated by the designation means in association with the keywords;
4. The extraction means according to claim 2, wherein, based on the stored presentation times, the extraction means preferentially extracts the presentation keywords for a second suppression count by the designation means. 5. The keyword extracting device according to the above. Re-presenting means for causing the presenting means to re-present the presenting keyword previously presented by the presenting means in response to the instruction;
The apparatus further comprises means for increasing or decreasing the second number of suppressions depending on whether an instruction to cause the re-presentation means to perform the re-presentation or an instruction to cause the extraction means to extract a presentation keyword is made. The keyword extraction device described in 1. Based on the instruction, further comprising session determination means for determining a session indicating an operation unit;
The keyword extraction device according to claim 4, wherein the update unit updates the preference degree based on a determination result of the session determination unit.
Based on the instruction, further comprising session determination means for determining a session indicating an operation unit;
The keyword extracting device according to any one of claims 9 to 12, wherein the extracting unit changes the number of times the keyword presentation is suppressed according to a result of the session determining unit.   The keyword extraction device according to claim 1, wherein the extraction unit extracts the presentation keyword randomly from the plurality of keywords in accordance with the instruction.   The keyword extracting device according to claim 1, wherein the extracting unit sequentially extracts all the plurality of keywords in a predetermined order in accordance with the instruction. A keyword storage unit for storing a plurality of keywords;
An association storage unit that stores association degrees between the plurality of keywords;
For each of the plurality of keywords, a preference storage unit that stores a degree of preference having a correlation with the frequency with which each of the plurality of keywords has been designated;
An output unit for outputting the selection candidate keywords;
A keyword extraction method for extracting a presentation keyword output by the output unit next time using a computer including a control unit that controls operations of the keyword storage unit, the related storage unit, the preference storage unit, and the output unit Because
Designating a keyword from keywords output by the output unit;
Extracting the presented keyword from the plurality of keywords based on the degree of association with the keyword designated by the designation unit and the preference degree. A keyword storage unit for storing a plurality of keywords;
An association storage unit that stores association degrees between the plurality of keywords;
For each of the plurality of keywords, a preference storage unit that stores a degree of preference having a correlation with the frequency with which each of the plurality of keywords has been designated;
An output unit for outputting the selection candidate keywords;
A keyword extraction program for causing a computer including a control unit that controls operations of the keyword storage unit, the related storage unit, the preference storage unit, and the output unit to extract a presentation keyword that the output unit outputs next time. And
Designating a keyword from keywords output by the output unit;
A keyword extraction program that causes the control unit to execute a step of extracting the presentation keyword from the plurality of keywords based on the degree of association with the keyword designated by the designation unit and the preference degree.   A computer-readable recording medium storing the keyword extraction program according to claim 18.

Images