JP2019159883A - Retrieval system, retrieval method - Google Patents

Abstract

To provide a retrieval system which makes it easy to provide information to a keyword input by a user.SOLUTION: A retrieval system 100 retrieves a facility information database 45 having information regarding a facility by using a keyword input by a user, and has: retrieval intention estimation means 36 for estimating a retrieval intention for a user to retrieve the facility when there is no information matching with the keyword in the facility information database, or the facility having the keyword does not exist in the surroundings; retrieval range expansion means 38 for expanding a retrieval range of the facility information database based on the keyword and the retrieval intention estimated by the retrieval intention estimation means; retrieval means for retrieving the facility in the retrieval range expanded by the retrieval range expansion means in the surroundings from the facility information database; and facility provision means 17 for providing facility information retrieved by the retrieval means.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a search system and a search method.

  It is widely performed that a user searches for information by inputting a keyword. In recent years, voice recognition technology has improved, and it has become easy for users to search for information by voice. For example, since it is difficult for a user operating a moving body such as a vehicle to operate a keyboard, it can be said that retrieval of information by voice is preferable in an in-vehicle device.

  Information retrieval may not satisfy the user's request unless the information exactly matches the input keyword, such as the meaning of a certain word. Even if the information is related to the input keyword, May meet user requirements. For example, when a user searches for a POI (Point Of Interest) such as a facility, the POI of the same genre as the input keyword often satisfies the user's request (for example, different search for a coffee shop name) If coffee shop information is provided).

  In addition, a general name of the facility such as the above genre is stored in the search DB in which a part of the information in the facility information DB storing the facility information is registered for indexing (indexing). However, the information stored in the search DB and the keyword input by the user do not always match (for example, even if “McDonald (registered trademark)” is registered in the search DB, the keyword input by the user is “ In the case of "Mac", fast food facilities will not hit the search.) Alternatively, even if the meaning is the same as the keyword input by the user, it may not be determined to be the same due to the shaking of the notation.

  For this reason, a synonym DB in which information having a similar meaning to the information in the search DB is stored has been prepared in the past, and the search range is expanded by searching including the synonym DB during the search. A technique is known (for example, refer to Patent Document 1). Patent Document 1 discloses a search device that performs a similarity search using attributes related to an attribute associated with an input keyword.

JP 2014-170297 A

  However, conventionally, there is a problem that the search result cannot be provided to the user unless the keyword input by the user is defined in the search DB or the synonym DB. This will be described with reference to FIG.

  FIG. 1 is an example of a diagram illustrating a search using the search DB 41 or the synonym DB 42. First, FIG. 1A shows an example of the search DB 41, and FIG. 1B shows an example of the synonym DB 42. The search DB 41 stores general facility genre names, and the synonym DB 42 stores general facility genre name abbreviations, derived terms, subordinate concepts, paraphrases, and the like.

  As shown in FIG. 1C, the user says “I want to eat a beef pot”, but “beef pot” is not stored in the search DB 41. However, since it is stored in the synonym DB 42, the search system determines that the beef pot means sukiyaki, and a facility whose genre is sukiyaki can be searched from the facility information DB.

  On the other hand, as shown in FIG. 1 (d), when the user says “I want to eat mochi-nabe”, “mochi-nabe” is not stored in either the search DB 41 or the synonym DB 42. In this case, the facility that provides the hot pot is not provided to the user, but the user may want the information if the facility provides a meal similar to the hot pot. However, in order to obtain information about a facility that provides a meal similar to a pot, the user needs to change the keyword and perform a search again.

  As described above, according to the conventional technology, there is a situation in which information related to a keyword cannot be provided to the user even if the user's request is satisfied if the information is related to the input keyword.

  In view of the above problems, an object of the present invention is to provide a search system that facilitates providing information on a keyword input by a user.

  In view of the above problems, the present invention is a search system that searches a facility information database having information about facilities using a keyword input by a user, and whether there is information that matches the keyword in the facility information database, Alternatively, when there is no facility having the keyword in the vicinity, the search intention estimating means for estimating the search intention for the user to search for the facility, and the keyword and the search intention estimated based on the search intention estimated by the search intention estimating means. Search range expansion means for expanding the search range of the facility information database, search means for searching for facilities in the search range that are expanded by the search range extension means from the facility information database, and search by the search means And a facility providing means for providing the facility information.

  It is possible to provide a search system that makes it easy to provide information on a keyword input by a user.

It is an example of the figure explaining the search using search DB or synonym DB. It is an example of the figure explaining the outline of the search by a search system. It is an example of the schematic block diagram of a search system. It is an example of the functional block diagram which shows the vehicle-mounted apparatus which a search system has, and the function of a search server in a block form. It is an example of the figure which shows extended synonym DB typically. It is an example of the input understanding result shown with the data structure called a frame. It is an example of the sequence diagram which shows an example of the procedure in which a search system searches a facility. It is an example of the conceptual diagram which shows how a search range expansion part determines extended synonym DB. It is an example of the sequence diagram which shows an example of the procedure in which a search system re-searches a facility. It is an example of the sequence diagram which shows an example of the procedure in which a search system switches extended synonym DB. It is an example of the sequence diagram which shows an example of the procedure which a search range expansion part switches extended synonym DB by a user's request | requirement. It is a figure explaining the example of an update of the designated point in designation | designated history information. It is an example of the sequence diagram which shows an example of the procedure in which a search system selects extended synonym DB. It is an example of the flowchart figure which shows the procedure in which a priority update part updates the priority of extended synonym DB in a priority table.

  Hereinafter, as an example of an embodiment for carrying out the present invention, a search system and a search method performed by the search system will be described with reference to the drawings.

<Search outline>
FIG. 2 is an example of a diagram for explaining an outline of the search by the search system 100 of the present embodiment. In the present embodiment, a plurality of extended synonym DBs 43 prepared in accordance with a user's search intention are prepared in advance for a certain keyword (in principle, the user speaks). The extended synonym DB 43 stores specific facility names that have some relevance. Relevance includes, for example, a viewpoint of search intention and a viewpoint of facility type.

If the search DB 41 and the synonym DB 42 do not have information that matches the keyword spoken by the user, or the search DB 41 and the synonym DB 42 have information that matches the keyword spoken by the user, but the facility is not found in the peripheral search based on the facility information DB 45 If not, the peripheral search based on the facility information DB 45 is continued using the expanded synonym DB 43 determined according to the keyword and the user's search intention (what is important in the current search).
(1) The user says, for example, “I want to eat a hot pot” as a keyword. The search system 100 determines that “mochi pot” is the keyword.
(2) Although the search system 100 searches the search DB 41 and the synonym DB 42, this keyword is not stored in the search DB 41 and the synonym DB 42.
(3) In this case, the search system 100 specifies the search intention that the user regards as important in the current search. The search intention may be specified from a past conversation, may be specified from a system usage history, may be estimated from a user profile (user attempt), or the search system 100 may inquire a user. Good. In FIG. 2, “price” is estimated as the search intention.
(4) The search system 100 determines the extended synonym DB 43 prepared in association with the keyword (money pot) and the search intention.
(5) The search system 100 searches the facility information DB 45 for a facility having the facility name included in the extended synonym DB 43.

  Therefore, the search system 100 according to the present embodiment searches the extended synonym DB 43 even when no keywords are stored in the search DB 41 and the synonym DB 42, so that it is possible to easily provide a search result for the keyword input by the user. . Further, even when keywords are stored in the search DB 41 and the synonym DB 42 but are not found in the vicinity, the extended synonym DB 43 is searched, so that it is possible to easily provide a search result for the keyword input by the user.

  In addition, since a keyword (having pot) that is not stored in the synonym DB 42 may not be set as a keyword in the extended synonym DB 43, the search system 100 converts a keyword that is not stored in the synonym DB 42 into a synonym. You can do it. As a result, the extended synonym DB 43 can be applied.

<Terminology>
The facility of the present embodiment may be a specific point (point) on the map. Such a facility is called POI (point of interest). Alternatively, the facility may be a facility prepared for some purpose. For example, various things such as stores, buildings, terrain, and addresses can be facilities.

  Extending the search range means, for example, adding a facility that has not been searched in the facility information DB to the search range. It does not include expanding the regional scope of facility location.

  Peripheral facilities refer to a geographical range where the user can move without taking too much time.

  An example of the database associated with the keyword and the search intention is the extended synonym DB 43, for example.

<System configuration example>
FIG. 3 is an example of a schematic configuration diagram of the search system 100. The search system 100 includes a search server 50 and an in-vehicle device 10 that communicate with each other via a network N. The network N has a public line provided by a communication carrier such as a mobile phone network, a wireless LAN network, or a WiMAX communication network, and a provider network provided by a provider (a company that provides a service for connecting to the Internet). ing. A base station 9 is connected to the network N so that the in-vehicle device 10 is connected to the public line of the communication carrier.

  The search server 50 is an information processing device that provides search results to the driver. There may be one or more search servers 50. The search server 50 also has a voice recognition function for recognizing a driver's voice and generating text data, and a language processing function for extracting the meaning of keywords and statements by analyzing the text data. Moreover, it is preferable to have a dialogue management function for managing dialogue with the user.

  The in-vehicle device 10 may have one or more functions (search function, voice recognition function, language processing function) that the search server 50 has. For example, the in-vehicle device 10 can have a voice recognition function and a language processing function. When the in-vehicle device 10 has all the functions of the search server 50, the search server 50 becomes unnecessary. Further, a server different from the search server 50 may perform voice recognition and language processing, and another server may transmit the keyword to the search server 50.

  The in-vehicle device 10 is one of information processing devices that output search results. The in-vehicle device 10 may be called a navigation device, for example. The in-vehicle device 10 called navigation searches for a route from a departure point to a destination and sets it in a road map, displays the route and the current location on an electronic map displayed on the display, or before changing the course based on the route. To guide the appropriate course by voice guidance or animation on an electronic map. In addition, it may have an AV (Audio Visual) playback function, a communication function with the Internet, and the like. The driver can reach the destination by operating the vehicle 8 according to the guidance. A predetermined server may search for a route among the functions of the in-vehicle device 10 and provide the route to the in-vehicle device 10. Further, the in-vehicle device 10 may transmit the position information of the vehicle 8 to a predetermined server, so that the predetermined server may perform route search and route guidance.

  The in-vehicle device 10 may be a general-purpose information processing terminal or a navigation dedicated terminal (also called a PND (Portable Navigation Device)). Examples of the general-purpose information processing terminal include a smartphone, a tablet terminal, a mobile phone, a PDA (Personal Digital Assistant), a notebook PC, and a wearable PC (for example, a wristwatch type, a sunglasses type). These devices are usually used as information processing terminals, but when application software for navigation is executed, route search, route guidance, and the like are performed in the same manner as a dedicated navigation terminal.

  However, the in-vehicle device 10 of the present embodiment does not need to have a function specific to the in-vehicle device, and only needs to be able to search the search server 50 with a keyword spoken by the user as described later. Therefore, the in-vehicle device 10 is not limited to a navigation device.

  In-vehicle device 10 may be realized by cooperation (communication) of a general-purpose information processing terminal and a dedicated terminal on the vehicle side. An application installed in a general-purpose information processing terminal provides a function necessary for a substantial search, and a dedicated terminal on the vehicle side acquires and displays a screen generated by the application through communication. As such an application, CarPlay (registered trademark), Android Auto (registered trademark), and the like are known. Further, a dedicated terminal on the vehicle side may provide functions necessary for substantial search.

  Thus, when cooperating with a general-purpose information processing terminal, the in-vehicle device 10 (vehicle-side dedicated terminal) may be a device called display audio (or connected audio). Display audio is a device that mainly provides an AV function and a communication function without a navigation function.

  The in-vehicle device 10 may be capable of switching between an in-vehicle state and a portable state in both cases of a general-purpose information processing terminal and a vehicle-side dedicated terminal. That is, the in-vehicle device 10 may be detachable from the vehicle 8.

<Search system functions>
FIG. 4 is an example of a functional block diagram illustrating the functions of the in-vehicle device 10 and the search server 50 included in the search system 100 in a block form.

<< In-vehicle device >>
The in-vehicle device 10 includes a voice acquisition unit 15, a voice output unit 16, a search result output unit 17, a display control unit 18, an operation reception unit 19, a voice data transmission unit 20, and a data acquisition unit 21. Each of these functions of the in-vehicle device 10 is a function or means realized by the CPU of the in-vehicle device 10 executing a program expanded from the flash memory (or HDD) to the RAM.

  For example, while the steering switch is being pressed, the voice acquisition unit 15 performs A / D conversion on the voice collected by the microphone 11 and converts it into an electrical signal to acquire it as voice data. Or you may always acquire audio | voice data, without the trigger operation by a user. Since voice is continuous data, the voice acquisition unit 15 repeatedly acquires voice data for a predetermined time (for example, 20 milliseconds). The predetermined time is set as a time advantageous for voice recognition.

  The voice data transmission unit 20 repeatedly transmits the voice data for a predetermined time acquired by the voice acquisition unit 15 to the search server 50. The communication protocol is not particularly limited, and a communication protocol suitable for mobile communication may be used as an example. It is assumed that the IP address of the search server 50 is known in advance.

  The data acquisition unit 21 acquires search results and the like transmitted by the search server 50. When additional information is required for the search, the data acquisition unit 21 acquires inquiry information for inquiring the user for additional information from the search server. If the user asks a question, an answer to the question is acquired. Search results, inquiry information, and answers are sent to the search result output unit 17.

  The search result output unit 17 sends the audio information of the search results, inquiry information, and answers to the voice output unit 16, and sends the screen information of the search results, inquiry information, and answers to the display control unit 18. Note that the search results, inquiry information, and answers may include only voice information or only screen information.

  The voice output unit 16 outputs the search result, the inquiry information, and the voice information included in the answer from the speaker 12. If the search result is already audio data in a format such as MP3 or PCM (Pulse Code Modulation), it can be reproduced as it is. When speech information is transmitted as text data, speech synthesis is performed.

  The display control unit 18 displays the search result, the inquiry information, and the screen information included in the answer on the display 13 (or HUD: Head Up Display). For example, when a facility is searched, information about the facility (a facility icon, a name of the facility, a telephone number, a map including the location of the facility), and the like may be displayed.

  The output of audio information to the speaker 12 and the output of screen information to the display 13 of the screen information are preferably performed at substantially the same time. However, the screen information may be displayed when the vehicle stops later or when the user operates.

  The operation reception unit 19 receives various operations on the in-vehicle device 10 using a touch panel, hard keys, and the like. In the present embodiment, for example, a tap for a displayed search result (facility name or the like) is accepted. As a result, detailed information on the facility can be requested to the search server 50.

<< Search server >>
The search server 50 includes a voice data receiving unit 31, a voice recognition unit 32, a language processing unit 33, a data transmission unit 34, a dialogue management unit 35, and a search unit 37. These functions of the search server 50 are functions or means realized by the CPU of the search server 50 executing a program and cooperating with various hardware elements of the search server 50.

  The audio data receiving unit 31 receives audio data from the in-vehicle device 10. The voice data receiving unit 31 sequentially sends the received voice data to the voice recognition unit 32. The voice recognition unit 32 converts voice data into text data. There are various methods for conversion to text data, and in this embodiment, a known method is used. In speech recognition, there are a method of converting speech data into text data in character units and a method of converting into text data in word units. In this embodiment, it is assumed that text data is converted into text data. In this case, the text data is decomposed into words (for example, independent words) by morphological analysis.

  The language processing unit 33 performs morphological analysis on the text data converted from the speech data by the speech recognition unit 32 and decomposes it into words. Morphological analysis refers to breaking a sentence into the smallest meaningful units (words in this embodiment). The language processing unit 33 further performs syntax analysis and semantic analysis to determine the content of the sentence. The language processing unit 33 sends a meaningful word string extracted from the text data to the dialogue management unit 35.

  The dialogue management unit 35 manages dialogue with the user using the word string. Details will be described later. The dialogue management unit 35 estimates a user request from the management result of the dialogue, and, for example, if a facility search is a user request, specifies a keyword for searching for a facility from a word string. This keyword is sent to the search unit 37.

  The dialogue management unit 35 includes an intention estimation unit 36, and the intention estimation unit 36 estimates a user's search intention. There are a method of inquiring the user anew and a method of estimating from the contents of previous and subsequent conversations.

  The search unit 37 is responsible for overall processing related to search. First, a genre or facility name is specified from the keyword by the search DB 41 and the synonym DB 42, and a facility having this genre or facility name is searched from the facility information DB 45.

  The search unit 37 includes a search range expansion unit 38 and a priority update unit 39. When the keyword is not stored in the search DB 41 and the synonym DB 42, or when the facility is not found in the peripheral search by the facility information DB 45, the search range expansion unit 38 searches the extended synonym DB 43 determined from the keyword and the user's search intention. Extend the range. The search unit 37 searches the facility information DB 45 by the facility name stored in one or more extended synonym DBs 43.

  Further, the priority update unit 39 changes the priority of the extended synonym DB 43 used by the search unit 37 for the search according to the facility or genre designation history by the user and the user profile.

  The data transmission unit 34 transmits the search result, inquiry information, or answer sent from the dialogue management unit 35 or the search unit 37 to the in-vehicle device 10.

  In addition, the search server 50 includes a storage unit 40 constructed in the HDD, flash memory, RAM, or the like of the search server 50. The storage unit 40 stores a search DB (Data Base) 41, a synonym DB 42, an extended synonym DB 43, a user information DB 44, a facility information DB 45, an extended DB selection table 46, and a priority table 47. Hereinafter, these databases or tables will be described.

表１は、検索ＤＢ４１に記憶されている情報を模式的に示す。検索ＤＢ４１は各施設を索引的に検索できるデータベースであり、施設情報ＤＢ４５の施設と対応付けられている。また、検索ＤＢ４１には一般的なジャンルが格納されている。このジャンルで後述する施設情報ＤＢ４５が検索される。施設情報を格納した施設情報ＤＢ４５の各施設には施設の施設名及びジャンルが設定されている。したがって、検索ＤＢ４１にキーワードが格納されている場合、このキーワードをジャンルとする施設の情報を提供可能になる。なお、検索ＤＢ４１に施設名が登録されていてもよい。

Table 1 schematically shows information stored in the search DB 41. The search DB 41 is a database that can search each facility in an indexed manner, and is associated with the facility in the facility information DB 45. The search DB 41 stores general genres. The facility information DB 45 described later is searched with this genre. The facility name and genre of the facility are set for each facility in the facility information DB 45 storing the facility information. Therefore, when a keyword is stored in the search DB 41, it is possible to provide information on facilities that use this keyword as a genre. The facility name may be registered in the search DB 41.

表２は、類義語ＤＢ４２に記憶されている情報を模式的に示す。類義語ＤＢ４２には施設のジャンルごとにこのジャンルの省略語、派生語、下位概念、言い換え、などの類義語が格納されている。例えば、ファーストフードに対してはジャンクフード、ハンバーガー、牛丼屋などの類義語が登録されている。また、類義語ＤＢ４２には、施設名とこの施設名の類義語が対応付けて登録されていてもよい。

Table 2 schematically shows information stored in the synonym DB 42. The synonym DB 42 stores synonyms such as abbreviations, derivative words, subordinate concepts, paraphrases, and the like for each genre of facility. For example, for fast food, synonyms such as junk food, hamburger, beef bowl shop are registered. In the synonym DB 42, a facility name and a synonym of the facility name may be registered in association with each other.

  In the synonym DB 42, synonyms are to be searched. Therefore, even when the user searches by a synonym instead of a keyword in the search DB 41, the genre of the facility stored in the search DB 41 can be specified from the synonym DB 42, and the search system 100 can search for the facility in the facility information DB 45. Become. The items in the synonym DB 42 are associated with the search DB 41.

表３は、ユーザ情報ＤＢ４４に記憶されている情報を模式的に示す。表３（ａ）はプロファイル情報を示す。プロファイル情報には、ユーザがどのような施設を好むかに関する情報として各ジャンルごとのユーザの施設の訪問回数が登録されている。訪問回数が多いほど、ユーザがそのジャンルの施設を好むことが推定できる。この訪問回数には、ユーザが施設を検索することなく、訪問した場合の訪問回数である。ただし、ユーザが施設を検索した結果、訪問した場合の訪問回数を含んでもよい。なお、検索システム１００は車両が一定時間以上、駐停車した場合にユーザがその施設を訪問したと判断する。

Table 3 schematically shows information stored in the user information DB 44. Table 3 (a) shows the profile information. In the profile information, the number of visits to the user's facility for each genre is registered as information regarding what facility the user likes. It can be estimated that the more the number of visits, the more the user prefers a facility of that genre. The number of visits is the number of visits when the user visits the facility without searching for facilities. However, the number of visits when a user visits as a result of searching for facilities may be included. Note that the search system 100 determines that the user has visited the facility when the vehicle has parked or stopped for a certain time or longer.

  Note that the user may register in advance the genre of the facility he / she likes as a profile. If the attributes such as the user's age and gender are known, the search system can obtain statistics on the number of visits for each attribute. The number of visits for which statistics are taken with the same attribute may be used as profile information instead of the number of individual visits.

  Table 3 (b) shows the priorities of search intentions set by the user. The user can register in advance the priority order of the search intention that he / she attaches importance to the search server 50 from the in-vehicle device 10 or the like. Thereby, the tendency of the search intention which a user attaches importance can be understood.

  Table 3 (c) shows the designation history information. The designation history information includes a genre designated by the user with respect to the retrieval system 100, designated points, and the number of retrieval intentions estimated at the time of retrieval. The genre designated by the user includes the genre determined from the facility name designated by the user.

  The designated history information is a user response to the facility proposed in the search result. For example, when the search system 100 proposes a hamburger shop and the user sets the destination, the designated points are increased. If the hamburger shop is proposed as a result of the search of the extended synonym DB 43, the search intention when the hamburger shop is proposed is increased by one (for example, the numerical value associated with “speed” is increased by one). Become).

  The designated point may not be the number of times the user sets the facility or genre as the destination, but may be replaced by the point by some conversion.

  Whether or not the user has specified is finally determined when the destination is set, but in this embodiment, for the convenience of explanation, the specified point is increased or decreased even if the user clarifies the intention or request in the dialogue. And However, the amount of increase of the designated point may be changed depending on the case where the destination is actually set and the case where the user clarifies the will or request in the dialogue.

  It can be estimated that the larger the designated point is, the more the user likes the facility of the genre. The designated point can be simply set to the number of times a facility or genre is designated, for example. Although it set to the destination, it does not matter whether the user has actually moved to the facility. This is because there is a possibility of going to another facility or entering a schedule while moving.

  Also, for example, when the user discards the facility proposal and designates another genre or a facility of that genre, the priority update unit 39 greatly reduces the points of the discarded genre and increases the genre of the designated facility You may let them.

  Each genre is associated with the number of times the search intention is adopted. That is, when the user sets the facility of the searched genre as the destination, what kind of search intention is searched is registered. Thereby, the tendency of the search intention which a user attaches importance can be understood.

表４は、施設情報ＤＢ４５に記憶されている情報を模式的に示す。施設情報ＤＢ４５には、各施設に関する施設情報が登録されている。例えば、施設名、店名、住所、電話番号、及び、ジャンルなどが登録されている。施設名はいわばブランド名であり、チェーン店の名称、屋号などである。店名は同じ施設名を区別するための名称であり、ない場合もある。住所は所在を示す。電話番号は連絡先である。ジャンルはこの施設を何らかの属性でグループ分けする際のグループ名であり、検索ＤＢ４１の一般的なジャンル、及び、拡張類義語ＤＢ４３のジャンルと対応付けられている。なお、１つの施設が複数のジャンルに属する場合がある。

Table 4 schematically shows information stored in the facility information DB 45. Facility information related to each facility is registered in the facility information DB 45. For example, a facility name, a store name, an address, a telephone number, a genre, and the like are registered. The name of the facility is a brand name, such as the name of a chain store or a store name. The store name is a name for distinguishing the same facility name and may not be present. The address indicates the location. The telephone number is the contact information. The genre is a group name when the facilities are grouped by some attribute, and is associated with a general genre in the search DB 41 and a genre in the extended synonym DB 43. One facility may belong to a plurality of genres.

表５は、拡張ＤＢ選択テーブル４６の一例を示す。拡張ＤＢ選択テーブル４６は、キーワードと検索意図から検索範囲拡張部３８が検索の拡張範囲となる拡張類義語ＤＢ４３を決定するためのテーブルである。表５に示すように、行方向に検索意図が、列方向にキーワードがそれぞれ対応付けられている。キーワードはユーザが入力するものなので、キーワードの種類は、施設名、ブランド名、カテゴリー名、メニュー名、商品名・・・など様々なものが考えられる。

Table 5 shows an example of the extended DB selection table 46. The extended DB selection table 46 is a table for the search range expansion unit 38 to determine an extended synonym DB 43 to be an extended search range from the keyword and the search intention. As shown in Table 5, the search intention is associated with the row direction and the keyword is associated with the column direction. Since keywords are input by the user, there are various types of keywords such as facility names, brand names, category names, menu names, product names, and so on.

  For each keyword, an appropriate extended synonym DB 43 is registered for each search intention. For example, even if the user searches with the keyword “McDonald's (registered trademark; omitted below)”, the search should be based on what the user places importance on, such as product / menu, speed, price, or atmosphere (search intent) This is because the extended synonym DB 43 is considered to be different. That is, the extended synonym DB 43 of A1 to A4 stores facility names that are relevant in terms of products / menus, speed, price, or atmosphere.

  For this reason, various extended synonym DBs 43 are prepared for one keyword, but it becomes difficult for the search server 50 to uniquely identify the extended synonym DB 43. Therefore, the search range expansion unit 38 determines the extended synonym DB 43 to be searched according to the user's search intention.

  Furthermore, the extended synonym DB 43 such as A1 to A4 registered in one cell of the extended DB selection table 46 is often not one. That is, even if the user's search intention can be specified, the extended synonym DB 43 is often not limited to one. There are a plurality of extended synonym DBs 43 that are considered to correspond to search intentions and keywords, and are classified for each relevant facility from the viewpoint of the type of facility, for example.

  As described above, since there are a plurality of extended synonym DBs 43 corresponding to one search intention and one keyword, it becomes easy to provide a search result desired by the user. A priority table 47 is prepared for the search server 50 to select one extended synonym DB 43 from a plurality of extended synonym DBs 43 corresponding to one search intention and one keyword.

  In addition, as a user's search intention, an undefined search intention (that is, other than product / menu, speed, price, or atmosphere) may be input in the search system 100, so an undefined intention is input. It is preferable that an extended synonym DB 43 to be used in the case is also defined.

表６は優先度テーブル４７の一例を示す。優先度テーブル４７は、１つの検索意図と１つのキーワードに対応する複数の拡張類義語ＤＢ４３のうちどれを優先するかを検索範囲拡張部３８が決定するためのテーブルである。表５で説明したように、１つ検索意図と１つのキーワードに対応する拡張類義語ＤＢ４３が複数あるため、表６のような優先度テーブル４７を用いて、検索範囲拡張部３８が優先して検索する拡張類義語ＤＢ４３を決定する。

Table 6 shows an example of the priority table 47. The priority table 47 is a table for the search range expansion unit 38 to determine which of the plurality of extended synonym DBs 43 corresponding to one search intention and one keyword is prioritized. As described in Table 5, since there are a plurality of extended synonym DBs 43 corresponding to one search intention and one keyword, the search range extension unit 38 uses the priority table 47 as shown in Table 6 to search with priority. The extended synonym DB 43 to be determined is determined.

  The keywords in the priority table 47 in Table 6 are all “McDonalds”. In the priority table 47 of Table 6 (a), when the search intention is a product / menu, four extended synonyms DB 43 are registered in order of priority. In the extended synonym DB 43 of priority 1, for example, another facility name that provides the same product as the product with the facility name “McDonald” is registered. In the extended synonym DB 43 of priorities 2 to 4, facility names of facilities that are likely to be able to purchase products similar to the product having the facility name “McDonald” are registered for each genre.

  In the priority table 47 of Table 6 (b), when the search intention is fast, four extended synonyms DB 43 are registered in order of priority. In the extended synonym DB 43 of priority 1, fast food facilities with a reputation for speed are registered. In the extended synonym DB 43 of priority 2, the facility names of noodles that are supposed to be provided quickly are registered. In the extended synonym DB 43 of priority 3, the facility name of a fast convenience store is registered in that it can be purchased and eaten as it is. In the extended synonym DB 43 of priority 4, the names of facilities of snacks that are not included in the priorities 1 to 3 but provide meals quickly are registered.

  In the priority table 47 of Table 6 (c), when the search intention is price, four extended synonyms DB 43 are registered in order of priority. In the extended synonym DB 43 of priorities 1 and 2, the names of facilities that provide products (beef bowls, noodles) whose price range is similar to the products of the facility name “McDonald” are registered. In the extended synonym DB 43 of priority 3, the facility names of the stores in general whose price range is similar to the product with the facility name “McDonald” are registered. In the extended synonym DB 43 with priority 4, the name of a facility that provides a product with a low price range is registered.

  When the search range expansion unit 38 expands the search range using the extended synonym DB 43, the extended synonym DB 43 to be used is determined according to the priority table 47 in descending order of priority.

  The priority is changed according to the profile information and specified history information in the user information DB 44. For example, if the user has a history of discarding the search result based on the extended synonym DB 43 with priority 1 and selecting the search result based on the extended synonym DB 43 with priority 2, the priority update unit 39 determines the occurrence frequency, etc. Since the designated point of the designated history information is updated according to, the extended synonym DB 43 of priority 1 and priority 2 can be exchanged.

  FIG. 5 is a diagram schematically showing the extended synonym DB 43. The extended synonym DB 43 stores facility names (for example, trademarks) of facilities belonging to this genre for each genre of facilities. That is, it is determined that the facilities registered in one extended synonym DB 43 are similar from the viewpoint. For example, the facility name of a hamburger chain store is registered in the extended synonym DB 43 called burger shop. The heading “Burger Shop” is a name representative of a facility stored in the extended synonym DB 43 (an example of a database name). In the present embodiment, a genre name or a name that can be regarded as a genre name is used.

  A plurality (any number) of extended synonym DBs 43 can be defined. The definition itself may be performed by the operator of the search system 100. Moreover, it is not a problem that the same facility name is stored in different extended synonym DBs 43. For example, the extended synonym DB 43 of the genre of fast food may include the contents of the extended synonym DB 43 of the genre of beef bowl chain. Moreover, another extended synonym DB 43 may be defined for each of one or more facility names in a certain extended synonym DB 43. For example, in the extended synonym DB 43 of the burger shop, Mos Burger (registered trademark; hereinafter omitted) is defined, but an extended synonym DB 43 having the facility name of Mos Burger as a representative name (for example, genre name) may be defined.

<Language processing and dialogue management>
Next, language processing and dialogue management will be briefly described. First, the language processing unit 33 further parses words extracted from text data by morphological analysis to determine dependency between words. For example, it is analyzed that “Near” modifies “Mac” and “A” is a predicate for voice data “Is there a Mac near?”.

  Next, the language processing unit 33 performs semantic analysis. Semantic analysis is performed with reference to a concept dictionary in which correct dependency is registered. Ambiguity does not occur in a short sentence such as “Is there a Mac near?”, But the meaning is resolved by semantic analysis and the meaning of the statement is specified.

  From the above, the dialogue management unit 35 determines that a request to provide a facility of the genre or facility name “Mac” near the current location is input. The dialogue manager 35 advances the dialogue to satisfy this request. By understanding the meaning of the user's remarks, it is possible to specify the dialogue action type for managing the dialogue. The dialogue action type is a format for dialogue management in which items necessary for dialogue are predetermined. In response to the above user request, the dialogue action type is determined to be a facility search request. Items necessary for the answer are specified by the dialogue action type.

  FIG. 6 is an example of an input understanding result shown by a data structure called a frame 70. The frame 70 is a set of symbols and attributes determined in advance according to the dialogue action type. The dialogue management unit 35 manages the dialogue using the frame 70. That is, the attribute associated with the symbol included in the frame is updated according to the progress of the dialogue. In FIG. 6, the left side of “:” is a symbol, and the right side is an attribute. When the attributes are determined for all symbols, one dialogue action is completed. Note that such an input understanding method is merely an example, and for example, a probability model such as a Bayesian network may be used.

  A frame 70 in FIG. 6 corresponds to the interactive action type corresponding to the facility search request. The genre is “Mac”, the area is “Near”, and the search intention is currently unknown. The dialogue management unit 35 can inquire about the search intention if necessary. The search intention is filled when the search range is extended by the extended synonym DB 43.

  Note that the dialogue management unit 35 holds not only frame management but also information such as past user comments or which of the user and the search system 100 should speak.

<Operation procedure>
FIG. 7 is an example of a sequence diagram illustrating an example of a procedure in which the search system 100 searches for a facility.

  S1: The user said, “Is there a Mac nearby?” To search for “McDonalds” nearby.

  S2: The voice acquisition unit 15 of the in-vehicle device 10 acquires this message and sends it to the voice data transmission unit 20. The voice data transmission unit 20 transmits voice data to the search server 50.

  S3: The voice data receiving unit 31 of the search server 50 receives the voice data, and the voice recognition unit 32 performs voice recognition to generate text data.

  S4: Next, the language processing unit 33 performs morphological analysis on the text data and converts it into words with parts of speech.

  S5: Next, the language processing unit 33 performs syntax analysis and semantic analysis to determine the user's request. The dialogue manager 35 determines the dialogue action type from the user's request and starts managing the dialogue.

  S6: The dialogue management unit 35 determines that the user requests a facility of the genre “Mac”. If the search unit 37 does not hit even when searching the search DB 41 by Mac, the search unit 37 can determine that Mac is McDonald's by searching the synonym DB 42. From the above, the search unit 37 performs a peripheral search with the keyword “McDonald”. That is, a facility name of McDonald's is searched from the facility information DB 45, and among those, McDonald's within a predetermined distance from the current location is narrowed down. The periphery is about 2 to 3 km in the traveling direction, for example. In addition, since the synonym DB 42 is registered with the official name of the genre or the official name of the facility (McDonald is the official name of Mac), the facility name is searched from the facility information DB 45, but the official name of the genre is When specified, a peripheral search is performed based on the genre of the facility information DB 45.

  When McDonald's is found in the vicinity, the data transmission unit 34 of the search server 50 transmits to the in-vehicle device 10 that McDonald's is found.

  If no McDonald's is found in the vicinity, the search unit 37 sends a message to that effect to the dialogue management unit 35. In addition, in the case where “I want to eat mochi nabe” is not included in either the search DB 41 or the synonym DB 42 for a statement such as “I want to eat mochi nabe”, it may not be possible to search around. In these cases, the dialogue management unit 35 determines that the search range is expanded because the facility search that is the purpose of the dialogue is not completed.

  S7: Therefore, the intention estimation unit 36 inquires of the user about the search intention. For example, the intention estimation unit 36 sends the in-vehicle device 10 via the data transmission unit 34 with inquiry information “McDonald's is located in the vicinity, so an alternative facility is searched. For what reason did you want McDonald's?” Send to.

  S8: The data acquisition unit 21 of the in-vehicle device 10 acquires inquiry information and sends it to the search result output unit 17. Since the search result output unit 17 sends the inquiry information to the voice output unit 16, the inquiry information is output from the speaker 12. On the other hand, the user says, for example, “I want to finish the meal soon!”.

  S9 to S12: Processing is performed in the same manner as steps S2 to S5, and the language processing unit 33 analyzes the meaning of the statement. For example, the intention estimation unit 36 detects the search intention of the user “speed” from the meaning of the statements “early” and “finish”.

  S13: As a result, the user's search intention is determined, and the search range expansion unit 38 determines the extended synonym DB 43 based on the keyword and the search intention. When a plurality of extended synonym DBs 43 are registered for a combination of a user's search intention and a keyword, the search range expansion unit 38 determines an extended synonym DB 43 with a priority of 1. As shown in Table 6 (b), it is assumed that the fast food extended synonym DB 43 is determined from “McDonald's” and “Speed”. As a result, the search range in the facility information DB 45 is expanded.

  In addition, in the case where “I want to eat motsunabe” is not included in either the search DB 41 or the synonym DB 42 for a statement such as “I want to eat mochi nabe”, “Motonabe” is included in the keyword for determining the extended synonym DB 43 in the first place It may not be. In such a case, the search range expansion unit 38 converts “Motonabe” into a synonym using a technique such as Word2Vec (Skip-Gram) and determines the extended synonym DB 43 using it as a keyword.

  S14: Since the extended synonym DB 43 is determined, the search unit 37 searches the vicinity. First, the facility information DB 45 is searched with each facility name registered in the extended synonym DB 43. When there is a facility hit in the search, the facilities within a predetermined distance from the current location are narrowed down.

  S15: The data transmission unit 34 of the search server 50 sends the search result (for example, “Yoshinoya (registered trademark, omitted below) found 2 km away. Do you want to set the destination?”) To the in-vehicle device 10. Send.

  S16: The data acquisition unit 21 of the in-vehicle device 10 acquires the search result and sends it to the search result output unit 17. The search result output unit 17 sends the audio information of the search results to the audio output unit 16 and sends the screen information to the display control unit 18. Thereby, a search result is output from the speaker 12 and the display 13. Note that the output to the display 13 may not be performed. For example, you may display at the timing which the vehicle stopped.

  For example, the in-vehicle device 10 outputs audio information of a search result such as “Yoshinoya is found 2 km away. Do you want to set it as a destination?” From the speaker 12.

  As described above, the search system 100 according to the present embodiment prepares the extended synonym DB 43 for the search intention and the keyword, thereby expanding the search range even when the facility based on the search DB 41 and the synonym DB 42 cannot provide the facility. This makes it easy to provide search results to the user.

<User search intent>
In the description of FIG. 7, the search server 50 inquires the user about the search intention, but the intention estimation unit 36 can acquire the search intention by other methods. Several methods will be described below.
(1) At the time of search, the user is inquired from the search system 100 (explained in FIG. 7).
(2) Judgment is made based on the user's comments before and after the keyword is input. For example, if a statement “search for a cheap place” is made at the time of a previous facility search, it can be estimated that the current search intention is also “price”. Or, if you say, “Moss is good, cheaper?” After “Is there a Mac nearby?”, It can be estimated that Mac's search intention is also “price”.
(3) Inferred from profile information or specified history information in the user information DB 44. As shown in Table 3 (b), since the user has registered in advance the priority order of the search intention in the search server 50, the search intention with the highest priority can be adopted. Moreover, since the search intention referred when the user designated the proposed facility as shown in Table 3 (c) is recorded for each genre, the number of times of adoption of the same search intention for each genre is totaled, The search intention with the largest value is adopted. That is, the number of adoptions for each product / menu, for each speed, for each price, and for each atmosphere is summed across genres.

When the search intentions determined in Table 3 (b) and Table 3 (c) are different, either may be adopted, but the user may be inquired.
(4) Link schedules and estimate based on free time and future schedule. The user may register a schedule on the cloud. The search server 50 can read a user's future schedule in cooperation with the schedule system. For example, if there is a schedule such as meeting a person one hour later, it is estimated that the user's search intention is fast. If there is no plan, it is presumed that products / menus, atmosphere, etc. will be given priority. If it is not clear from the schedule, it may be confirmed with the user.

<Selection of extended synonym DB>
How the extended synonym DB 43 is switched will be described with reference to FIG. FIG. 8 is a conceptual diagram showing how the search range expansion unit 38 determines the extended synonym DB 43. FIGS. 8A to 8D are extended synonym DBs 43 having priorities 1 to 4, respectively.
(1) First, the search unit 37 searches the facility information DB 45 for a facility name stored in the extended synonym DB 43 for fast food having priority 1.
(2) If no facility is found around the vehicle in the fast food extended synonym DB 43, the search range expansion unit 38 determines the next highest priority noodle extended synonym DB 43, and the search unit 37 has a priority of 2. The facility name stored in the extended synonym DB 43 for noodles is searched from the facility information DB 45.
(3) When no facility is found around the vehicle in the extended synonym DB 43 of noodles, the search range expansion unit 38 determines the extended synonym DB 43 of the next highest priority convenience store, and the search unit 37 has a priority 3 convenience store. The facility name stored in the extended synonym DB 43 is searched from the facility information DB 45.
(4) If no facility is found around the vehicle in the convenience store extended synonym DB 43, the search range expansion unit 38 determines the next highest priority snack extended synonym DB 43, and the search unit 37 has a priority 4 snack The facility name stored in the extended synonym DB 43 is searched from the facility information DB 45.
(5) When the search unit 37 searches the fast food extended synonym DB 43 and the dialogue management unit 35 provides the facility, when the user specifies the genre of noodles or the facility of this genre, the search range expansion unit 38 sets the noodles The extended synonym DB 43 is determined, and the search unit 37 searches the facility information DB 45 for a facility name stored in the extended noodles synonym DB 43.
(6) The search unit 37 searches the extended synonym DB 43 for fast food and the dialogue management unit 35 provides the facility. However, when the user designates a genre called a convenience store or a facility of this genre, the search range expansion unit 38 sets the convenience store The extended synonym DB 43 is determined, and the search unit 37 searches the facility information DB 45 for a facility name stored in the extended synonym DB 43 of the convenience store.
(7) When the search unit 37 searches the extended synonym DB 43 for fast food and the dialog management unit 35 provides the facility, when the user designates a genre of snacks or a facility of this genre, the search range extension unit 38 The extended synonym DB 43 is determined, and the search unit 37 searches the facility information DB 45 for a facility name stored in the extended synonym DB 43 for snacks.

  In the search according to the present embodiment, the extended synonym DB 43 can be switched in this way, so that it is easy to provide the facility desired by the user.

<Re-search in the same extended synonym DB>
When the user requests another facility in step S16 of FIG. 7, the search unit 37 provides stores other than the store proposed this time from the fast food extended synonym DB 43. For example, if “Yoshinoya” is proposed, the next search is performed except for “Yoshinoya”.

FIG. 9 is an example of a sequence diagram illustrating an example of a procedure in which the search system 100 searches for a facility again. Note that the processing in FIG. 9 will be described as being executed following step S16 in FIG.
S21: The user was not satisfied with “Yoshinoya” and, for example, said, “Another shop is good”.

  S22 to S25: Processing is performed in the same manner as steps S2 to S5, and the language processing unit 33 analyzes the meaning of the statement. For example, the language processing unit 33 detects a user's request to search for another store from the dependency of the words “other” and “store”.

  S26: As a result, the dialogue management unit 35 determines that the request is a facility search request excluding “Yoshinoya”, and the search unit 37 searches the surrounding area with the same fast food extended synonym DB 43. Although the extended synonym DB 43 can be changed at this time, the same extended synonym DB 43 is used because a clear intention for designating another genre is not extracted from the user. When the intention to specify another genre is extracted from the user's statement, the search range expansion unit 38 can change the extended synonym DB 43.

  S27: The data transmission unit 34 of the search server 50 transmits the search result (eg, “Mos Burger was found 2.4 km away. Do you want to set it as the destination?”) To the in-vehicle device 10.

  S28: The data acquisition unit 21 of the in-vehicle device 10 acquires the search result and sends it to the search result output unit 17. The search result output unit 17 sends the search result audio information and screen information to the audio output unit 16 and the display control unit 18, respectively, so that the search result is output from the speaker 12 and the display 13.

  For example, the in-vehicle device 10 outputs audio information of a search result such as “Mos Burger found 2.4 km away. Do you want to set it as a destination?” From the speaker 12.

  In this way, even if the user does not change the keyword, another facility similar to the keyword facility in terms of search can be searched using the extended synonym DB 43.

<Automatic switching of extended synonym DB>
In the description of FIG. 9 or FIG. 10, if there are no stores in the vicinity other than the store suggested last in the fast food extended synonym DB 43, the search range is determined according to the priority table 47 of Table 6 (b). The extension unit 38 switches to the extended synonym DB 43 of the next priority. For example, when there is no fast food facility other than the Yoshinoya in the vicinity in step S26 of FIG. 9, the search unit 37 searches for the vicinity using the extended synonym DB 43 for noodles with priority 2.

FIG. 10 is an example of a sequence diagram illustrating an example of a procedure for the search system 100 to switch the extended synonym DB 43. Note that the processing in FIG. 10 will be described as being executed following step S16 in FIG.
S31 to S36: The same as steps S21 to S26 in FIG.

  S37: Even if the search unit 37 searches the vicinity of the same fast food extended synonym DB 43 in step S36, the facility cannot be found in the vicinity. For this reason, the search range expansion unit 38 refers to the priority table 47 and switches to the extended synonym DB 43 of the next priority.

  S38: The search unit 37 searches the facility information DB 45 with the facility name stored in the extended synonym DB 43 of the next priority, and narrows down facilities within a predetermined distance from the current location. Also in this case, “Yoshinoya” is preferably excluded. The subsequent steps S39 and S40 may be the same as steps S27 and S28 in FIG.

  The voice output unit 16 of the in-vehicle device 10 outputs, for example, a voice “Sorry for Kakuraku 3 km away. Do you want to set it as the destination?” From the speaker 12.

  As described above, the search system 100 can automatically switch to the next priority extended synonym DB 43 when there is no facility requested by the user in the priority synonym DB 43, so that the search range is automatically expanded. It becomes easier to provide search results to the user.

  Similarly, when the extended synonym DB 43 with priority 2 is used and no facility exists in the vicinity, the search range expansion unit 38 can switch to the extended synonym DB 43 with priority 3. Switching to the extended synonym DB 43 with priority 4 or later can be performed in the same manner.

<Change of extended synonym DB by user>
In step S16 of FIG. 7, even when the user designates the change of the extended synonym DB 43, the search range expansion unit 38 switches the extended synonym DB 43.

  FIG. 11 is an example of a sequence diagram illustrating an example of a procedure in which the search range expansion unit 38 switches the extended synonym DB 43 in response to a user request. Note that the processing in FIG. 11 will be described as being executed following step S16 in FIG.

  S41: The user was surprised by the proposal “Yoshinoya” and, for example, said, “Why is Yoshinoya?”.

  S42 to S45: The same as steps S2 to S5 in FIG. For example, the language processing unit 33 determines that the reason for selection of the extended synonym DB 43 is being asked from the dependency of the words “why” and “what?”.

  S46: The dialogue management unit 35 sets a dialogue action type that answers the reason for selection, and generates a reason for selection. Using the search intention and the genre name of the extended synonym DB 43, for example, an answer “searched for fast food as a store where meals can be completed quickly” is generated.

  S47: The data transmission unit 34 of the search server transmits the answer to the in-vehicle device 10. The data acquisition unit 21 of the in-vehicle device 10 acquires the response and sends it to the search result output unit 17. As a result, the voice output unit 16 outputs the answer from the speaker 12.

  S48: In response to this, the user remarked “Make it another genre!”.

  S49 to S52: The same as steps S2 to S5 in FIG. For example, the language processing unit 33 determines to change the extended synonym DB 43 from the dependency of the words “other”, “genre”, and “do”.

  S53: First, the search range expansion unit 38 determines that the user has discarded the genre of fast food. For this reason, the designated point of the fast food in the designated history information is reduced. Thereby, the search range expansion unit 38 may switch to the extended synonym DB 43 of the next priority. However, in this embodiment, in order to make an inquiry to the user, the dialogue management unit 35 uses the genre names of other extended synonym DBs 43, for example, “I suggest noodles, convenience stores, and snacks. Generate a query.

  S54: The data transmission unit 34 of the search server transmits the inquiry to the in-vehicle device 10. The data acquisition unit 21 of the in-vehicle device 10 acquires the response and sends it to the search result output unit 17. As a result, the voice output unit 16 outputs the answer from the speaker 12.

  S55: In response to this, the user replied, “It ’s OK at a convenience store.” Since this statement is transmitted to the search server 50, the search range expansion unit 38 of the search server 50 can switch to the extended synonym DB 43 whose genre is a convenience store and perform a peripheral search. In addition, the search range expansion unit 38 determines that the user has specified a genre of a convenience store. For this reason, the designated point of the convenience store in the designated history information is increased.

<< Update of designated point >>
As described above, since the user discards the genre of family restaurant and selects the genre of convenience store in the search of FIG. 11, the search range expansion unit 38 updates the specified history information in the user information DB 44.

  FIG. 12 is a diagram for explaining an example of updating the designated point in the designated history information. FIG. 12 is an excerpt of part of the specified history information. FIG. 12A shows the designation history information before update, and FIG. 12B shows the designation history information after update. Since the user cancels the proposal of the fast food store and designates the genre called convenience store, the priority update unit 39 decreases the designated point of the fast food genre by 20%, for example, and increases the designated point of the convenience store by 20%.

  By doing so, the designated point is updated every time the user discards or designates the facility. Therefore, if the priority of the extended synonym DB 43 is changed according to the designated point, the extended synonym is repeated as the search is repeated. The priority of the DB 43 matches the user's preference.

  In the description of FIG. 12, discard and selection are performed at the same time, but only discard or only selection may be performed. Further, the decrease or increase of 20% is merely an example, and the amount of variation may be set as appropriate. Moreover, you may update with an absolute value instead of a percentage.

  Note that it is also preferable to change the increase amount of the designated point between the case where the destination is set and the case where the genre is simply designated.

<Specifying facilities by the user>
In FIG. 11, the search server 50 inquires “I suggest noodles, convenience stores, and snacks. Which genre do you want to use?”, The user replied “No convenience store”. However, it will be convenient if the user can communicate his / her preference to the search server 50. That is, even if the user's utterance does not match the inquiry, it is preferable that the extended synonym DB 43 appropriate for the user's utterance can be determined.

  FIG. 13 is an example of a sequence diagram illustrating an example of a procedure for the search system 100 to select the extended synonym DB 43. Note that the processing in FIG. 13 will be described as being executed following step S54 in FIG.

  S61: The user said, “Make ramen”.

  S62 to S65: The same as steps S2 to S5 in FIG. For example, the language processing unit 33 determines that the user is requesting a facility that provides ramen from the dependency of the words “ramen” and “to do”.

  S66: The dialogue management unit 35 sets a dialogue action type of facility search request. The genre is ramen. First, the search unit 37 searches the search DB 41 and the synonym DB 42 but cannot find them. Therefore, the search range expansion unit 38 determines the extended synonym DB 43 from the keyword (ramen) and the search intention. First, if there is an extended synonym DB 43 associated with the keyword ramen, the search range expansion unit 38 adopts it. If there is no extended synonym DB 43 associated with the keyword ramen, the ramen is converted into another keyword by the synonym DB 42 and the expanded synonym DB 43 associated with this keyword is adopted. Alternatively, each extended synonym DB 43 is searched with the word ramen (an example of information specifying a facility), and the extended synonym DB 43 with the largest number of hits is employed. Alternatively, a synonym of ramen may be estimated using a natural language processing algorithm such as Word2Vec, and the extended synonym DB 43 may be determined using this synonym as a keyword. By doing so, the extended synonym DB 43 can be determined even if the user communicates his / her preference to the search server 50. In the following, for example, it is assumed that the extended synonym DB 43 for noodles can be determined.

  S67: The search unit 37 performs a peripheral search in the extended synonym DB 43 for noodles. At this time, simply searching the extended synonym DB 43 for noodles may include a soba restaurant, so it is better to search for facilities that provide ramen. For example, a facility having a synonym of ramen (Chinese etc.) in the store name is preferentially searched.

  S68: The data transmission unit 34 of the search server 50 transmits the search result to the in-vehicle device 10.

  S69: The data acquisition unit 21 of the in-vehicle device 10 acquires the search result and sends it to the search result output unit 17. Since the search result output unit 17 sends the search result to the audio output unit 16 and the display control unit 18, the search result is output from the speaker 12 and the display 13.

  Also in this case, the designated point in the extended synonym DB 43 for the genre of noodles is updated so as to increase, and the designated point of the extended synonym DB 43 for the genre of fast food is updated so as to decrease.

<Updating priority table>
Since the priority table 47 determines the order of the extended synonym DB 43 used by the search unit 37 for the search, it is important for providing the user with a favorite facility at an early stage. In the present embodiment, the priority of the extended synonym DB 43 in the priority table 47 is updated using the user profile or specified history.

  FIG. 14 is an example of a flowchart illustrating a procedure in which the priority update unit 39 updates the priority of the extended synonym DB 43 in the priority table 47. The processing in FIG. 14 is periodically executed, for example, once a week or once a month. Further, it is executed for each of the plurality of extended synonym DBs 43 associated with the same keyword and the same search intention. For example, the priority of the extended synonym DB 43 such as fast food, noodles, convenience store, and snack belonging to A2 associated with McDonald's and speed in the extended DB selection table 46 is updated.

  First, the number of visits of each genre in the profile is pointed (S101). That is, in order to adjust the weight with the designated point, for example, the number of visits is multiplied by a coefficient. When the number of visits is more important than the designated point, a larger coefficient is multiplied. When the designated point is more important than the number of visits, a smaller coefficient is multiplied.

  Next, the priority update unit 39 acquires designated points for each genre from the designated history information (S102).

  Moreover, the priority update part 39 totals the point and designated point of step S101 for every genre (S103).

  Then, the priority update unit 39 changes the priority in descending order of the total value (S104). Thus, by periodically updating the priority of the extended synonym DB 43, it becomes possible to provide the user's favorite facility at an early stage. Note that at least one of the number of visits and the designated point may be used.

<Summary>
As described above, in the search system according to the present embodiment, a plurality of extended synonym DBs are prepared for one keyword from different viewpoints (search intention), so that what is important for the search is searched. The extended synonym DB to be referred to can be changed according to the intention.

  By preparing multiple extended synonym DBs for one keyword and the user's search intention and switching the extended synonym DBs according to the priority, the facility can be found by using only one extended synonym DB. Even when there is not, it is possible to easily provide the search result by appropriately expanding to the extended synonym DB.

  In consideration of the user's search intention, by determining the extended synonym DB for the keyword, a result close to the user's search intention can be provided even if the keyword spoken by the user is not an official facility name or the like. it can.

  By changing the priority of the extended synonym DB based on the facility designated by the user and the extended synonym DB, search results suitable for the user's search intention can be provided at an early timing.

<Other application examples>
The best mode for carrying out the present invention has been described above with reference to the embodiments. However, the present invention is not limited to these embodiments, and various modifications can be made without departing from the scope of the present invention. And substitutions can be added.

  For example, since the keyword input by the user may not be associated with the extended synonym DB 43, in this case, the keyword is converted into another similar keyword using a natural language processing algorithm such as Word2Vec, and the keyword is changed to another keyword. The associated extended synonym DB 43 may be determined.

  In the present embodiment, an example in which a keyword is input by voice has been described. However, a keyboard or a touch panel may be used for inputting a keyword.

  In this embodiment, an example in which the user searches for facilities from the vehicle 8 has been described. However, the user may be on a light vehicle, a powered motorcycle (referred to as a motorcycle), or the like. That is, the moving body is not limited to a vehicle. In addition, the user may move on foot, and in this case, the user may input a keyword to a general-purpose information processing device instead of the in-vehicle device 10.

  The intention estimation unit 36 is an example of a search intention estimation unit, the search range expansion unit 38 is an example of a search range expansion unit, the search unit 37 is an example of a search unit, and the search result output unit 17 is a facility provided. The priority update unit 39 is an example of a priority update unit, and the facility information DB 45 is an example of a facility information database.

DESCRIPTION OF SYMBOLS 10 In-vehicle apparatus 50 Search server 100 Search system

Claims (11)

A search system for searching a facility information database having information on facilities using a keyword input by a user,
If there is no information that matches the keyword in the facility information database, or there is no facility having the keyword in the vicinity, search intention estimation means for estimating a search intention for the user to search for a facility;
Search range expansion means for extending the search range of the facility information database based on the search intention estimated by the keyword and the search intention estimation means;
Search means for searching the facility information database for facilities in the search range expanded by the search range extension means, and surrounding facilities;
Facility providing means for providing facility information searched by the search means;
A search system comprising: One or more databases that are facilities related to the keyword and registered with the name of the facility corresponding to the search intention estimated by the search intention estimation means are stored in association with the keyword and the search intention. ,
The search range expansion means uses the facility name registered in the database associated with the search intention estimated by the keyword and the search intention estimation means to search the facility information database. The search system according to claim 1, wherein the range is expanded. When the user inputs a request to provide a facility other than the facility information provided by the facility providing means,
The search means is a facility different from the facility information provided by the facility providing means, and the facility name registered in the database in which the facility name provided by the facility providing means is stored is the facility. The search system according to claim 2, wherein the search system is used for searching an information database. A plurality of the databases are associated with the keyword and the search intention,
When the search range is expanded to a certain database, but the search means cannot find a facility within a predetermined distance from the current location from the facility information database,
The search range extending means extends the search range by using the facility name registered in another database associated with the keyword and the search intention for searching the facility information database. The search system according to claim 2 or 3, wherein A priority is set for each of the plurality of databases associated with the keyword and the search intention,
When the search range expansion means has expanded the database to the search range, but the search means cannot find a facility within a predetermined distance from the current location from the facility information database,
The search system according to claim 4, wherein the search range expansion unit switches the database in descending order of priority. When the user inputs that the database is specified,
The search system according to claim 2 or 3, wherein the search range extending means extends the database designated by the user to the search range. When information specifying a facility that does not match the name of the database is input from the user,
The search range extending means extends the database associated with information designating the facility or the database associated with information converted from information designating the facility to the search range. The search system according to claim 2 or 3, wherein the search system is characterized. The number of times the facility has been used in association with the database of information about the facility that the user has used in the past, or
The number of times that the user inputs the designation of the database, recorded in association with the database,
6. The search system according to claim 5, further comprising priority update means for updating the priority of the plurality of databases using at least one of the database. The search means records the search intention when the user searches for a facility using the database in the past,
The said search intention estimation means estimates the search intention in which the said user searches a facility based on the said past search intention recorded, The description in any one of Claims 2-8 characterized by the above-mentioned. Search system.   The schedule intention in which the user's schedule is registered communicates with the search intention estimation means, and the search intention for the user to search for a facility is estimated based on the user's schedule. The search system according to any one of the above. A search method of a search system for searching a facility information database having information on facilities using a keyword input by a user,
When the search intention estimation means has no information that matches the keyword in the facility information database, or there is no facility having the keyword in the vicinity, a step of estimating the search intention for the user to search for the facility;
A search range expansion unit extending the search range of the facility information database based on the keyword and the search intention estimated by the search intention estimation unit;
A step of searching from the facility information database for a facility in the search range expanded by the search range expansion unit and surrounding facilities;
A facility providing means providing the facility information searched by the search means;
A search method characterized by comprising:

Images