JP4921330B2 - Search result snippet generation device, generation method, generation program and recording medium on which the generation program is recorded in consideration of the range that the place name means - Google Patents

Description

  The present invention relates to a method for searching a text set existing in a computer or accessible via a computer network, and outputs a snippet of a search result particularly in a search including position information in a text to be searched or a query. Further, the present invention relates to a search result snippet generation device, a generation method, a generation program, and a recording medium technique in which the generation program is recorded in consideration of a range that a place name means.

  In a general text search system, a plurality of texts including a search keyword input by a user are specified, and a list thereof is presented to the user as a search result. In the search result, a snippet showing an outline of the text is shown together with the title and URL of each text. As a result, the user can grasp the outline of the text without selecting an actual document and can select a desired document.

  The simplest snippet generation method is a method of acquiring a document of a specific size from the beginning of text. News articles are useful tools because they often show the outline of the article at the beginning of the sentence. However, in many texts existing on the Web, the summary is not always shown at the beginning of the sentence, so the snippet is often not useful to the user.

  In order to cope with this problem, there is a method of extracting a partial document (KWIC) including a keyword input by a user from each document as a search result and making it a snippet (see Non-Patent Document 1). According to this, since the text in the vicinity of the keyword that the user is interested in can be presented, it becomes easy for the user to select what is really desired from the individual search results when browsing the search results.

  Currently, there is an enormous amount of text data on the Web, and there is a large amount of text data that refers to a specific place. Further, it is said that about 20% of inquiries in the Web search engine are inquiries about a specific place or region (see Non-Patent Document 2).

  Under such circumstances, by extracting the place name notation from the text and converting the place name into latitude and longitude, a technique for linking the text to the position (see Non-Patent Document 3) can be practically used. Yes. By using this technique, it becomes possible to provide latitude and longitude information in the text, and even if the place name keyword does not match, it is possible to search for text information according to the location desired by the user. In other words, in response to a search request “place: Shibuya, keyword: restaurant”, a search including a keyword “restaurant” and referring to the vicinity of “Shibuya” can be performed.

Incidentally, the block level position reference information used in the embodiment of the present invention is disclosed in Non-Patent Document 4.
Drori, O .; , The Benefits of Displaying Additional Internal Document Information on Textual Database Search Result Lists, Proc. of ECDL 2000 (2000). Zhou, Y .; , Et al. : “Hybrid Index Structure for Location-based Web Search”, Proc. of CIKM2005. Clow, P.M. "Extracting Metadata for Spatially-Aware Information Retrieval on the Internet", Proc. of GIR2006 “Block level location reference information download service, block level location reference information”, data that has the latitude and longitude of block units (XX town △ address □ address) Ministry of Land, Infrastructure, Transport and Tourism National Land Planning Bureau Search on November 20, 2011] Internet <URL: http: // nlftp. mlit. go. jp / isj / about. html>

  When searching using the geographical information as described above, the given geographical information is not necessarily a place name, but may be a latitude and longitude obtained from GPS. Even if geographical information is given by the place name, once it is converted into latitude and longitude, a completely matching keyword is not always present in the text. In this case, when the conventional technique using the KWIC described above is used, a snippet including a keyword input as a search keyword can be obtained, but there may be a case where position information cannot be included.

  In addition, when all the position information existing in the text is handled equally, there is a risk that the accuracy of the snippet may decrease due to the place name that is noisy or the place name that is ambiguous.

  An object of the present invention is to generate a search result snippet in consideration of a range that a place name means, which can generate a snippet applied to a user's request using position information existing in text in view of the above problems An apparatus, a generation method, a generation program, and a recording medium on which the generation program is recorded.

  In the present invention, the importance of the place name in the text is calculated using the latitude and longitude of the place name included in the text and the implication width, and a snippet is generated using the importance.

That is, the search result snippet generation device according to claim 1 is specified by specifying a search keyword and position information for specifying the content of the text from a text set existing inside the computer or accessible via a computer network. In a search result snippet generation device that obtains a snippet showing an outline of each search result when searching for text that includes a search keyword and related to a specific position, place names existing in the text of the text set are obtained. Geographic index means for extracting and assigning latitude and longitude to the extracted place name, place name implication range specifying means for specifying the area implied by each place name, and latitude and longitude of the place name assigned by the geographical index means and the place name Information about the size implied by the place name specified by the implication range specifying means Based on, by using an equation information measuring is great importance to narrower, and place name importance degree determination unit that performs processing for calculating the importance of the place names in the text, given search terms, and Based on the importance of the place name, the sentence, which is a partial document in the text, is evaluated with the sum of the importance depending on the search keyword and the importance of the place name , and the sentence considered important is extracted and output as a snippet. A sentence evaluation means for processing, and the importance of the place name in the text searched based on the designated search keyword and position information is obtained by inquiring to the place name importance degree determination means, and the obtained place name severity, and the search keyword was sent to the sentence evaluating means, snippets inquiry reception hand for accepting a snippet output from the sentence evaluation means It is characterized by comprising and.

The search result snippet generation device according to claim 2 is the search result snippet generation device according to claim 1, wherein the importance of the place name transmitted from the snippet inquiry reception unit to the sentence evaluation unit is close to the specified position information and is a fixed distance. It is characterized by the importance of one or more place names existing within .

The search result snippet generation method according to claim 3 is specified by specifying a search keyword and position information for specifying text contents from a text set existing inside a computer or accessible via a computer network. In a search result snippet generation method for obtaining a snippet indicating an outline of each search result when searching for text including a search keyword and related to a specific position, a geographic index means includes a text index in the text set. Geographic index step of extracting existing place names and assigning latitude and longitude to the extracted place names, place name entailment range specifying means for specifying the area name entailment range specifying step, and location name importance level determination Means of the place name assigned by the geographical index step Based on the information of the size of a place name specified by longitude and the place name implies range identification step implies, using the formula information measuring is great importance to narrower, the importance of the place names in the text A place name importance determining step for performing a process of calculating the place name, and a snippet inquiry accepting means for determining the importance of the place name in the text searched based on the designated search keyword and position information, A step in which the snippet inquiry accepting means sends the importance of the obtained place name and a search keyword to the sentence evaluating means, and the sentence evaluating means sends the place name sent from the snippet inquiry accepting means. the importance of, and on the basis of the search keyword, the sentence is a partial document in the text, depending on the search keyword It was evaluated with the sum of the importance of importance and place names, and sentences evaluation step for performing a process for outputting as snippets extracted sentences are considered to be important, snippets inquiry reception unit, output from the sentence evaluation means And a step of receiving a snippet.

The search result snippet generation method according to claim 4 is the search result snippet generation method according to claim 3 , wherein the importance of the place name transmitted from the snippet inquiry reception means to the sentence evaluation means is close to the specified position information, and is a fixed distance. It is characterized by the importance of one or more place names existing within .

A search result snippet generation program according to claim 5 is a program that causes a computer to function as each means according to claim 1 or 2 .

A recording medium according to claim 6 is a computer-readable recording medium in which the search result snippet generation program according to claim 5 is recorded.

  With the above configuration, it is possible to generate a snippet that takes into account the location specified by the user, and can assist the user in selecting a desired text from the search result.

According to the first to sixth aspects of the present invention, the importance of the place name appearing in the document is estimated based on the width implied by the place name and the position information given as the search condition, and the importance is calculated. By extracting the search result snippet, it is possible to generate a snippet that takes into account the specified location, for example, the location specified by the user, and assists the user in selecting the desired text from the search result it can.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings, but the present invention is not limited to the following embodiments. FIG. 1 shows an example of the configuration of a search system using the snippet generation device of the present invention.

  In FIG. 1, the search system according to the present embodiment includes a text database 010, a geographic indexer 020, a location information database 030, a place name implication range specifying unit 100, a place name database 110, a snippet inquiry receiving unit 210, a place name importance degree determining unit 220, A sentence evaluation unit 230, a search engine 310, and a search application 320 are provided.

  Each of the above units is realized by a computer, for example, and each function will be described below.

  A text DB (database) 010 is a database in which text information in a computer and text information accessible via a network are collected. In addition to the information of the text itself, an index structure for returning a document including a specified keyword when accessed from the search engine 310 is also held.

  The geographic indexer 020 functions as the geographic index means of the present invention, accepts text information as an input, extracts a place name included in the text, and latitude and longitude of the position in the real world indicated by the extracted place name and normalization The place name (normalized place name starting from the prefecture name) is output.

  The position information DB (database) 030 is a database that stores, in each document, what place name appears at which position in the text and the latitude and longitude indicated by the place name. An example of the contents of the database is shown in FIG.

  The place name implication range specifying unit 100 functions as place name implication range specifying means of the present invention, specifies the area implied by the place name, that is, the range indicated by the place name, and registers it in the place name database 110. As the range indicated by the place name, latitude width and longitude width are conceivable. This data can be constructed based on block level position reference information distributed by the Ministry of Land, Infrastructure, Transport and Tourism (see Non-Patent Document 4).

  For example, when acquiring the range indicated by “Shibuya-ku, Tokyo,” the block level position reference information is acquired for the block starting from “Shibuya-ku, Tokyo”, and the latitude width and longitude in the block data are acquired. By obtaining the width, the range is indicated by the place name.

  The place name DB (database) 110 ”is a database for managing data created by the place name implication range specifying unit 100. An example of the contents of the database is shown in FIG.

  The snippet inquiry reception unit 210 functions as the snippet inquiry reception unit of the present invention, acquires the search keyword and the ID of the document for generating the snippet from the search engine 310, and accesses the place name importance determination unit 220 and the sentence evaluation unit 230. The sentence evaluation unit 230 returns the generated snippet to the snippet inquiry reception unit 210.

  Also, the place name importance determining unit 220 is accessed based on the document ID, and the place name information with importance (the place name and the importance of each place name in the document) included in the document is acquired.

  The sentence evaluation unit 230 is accessed based on the search keyword, document ID, and place name information, and a snippet is acquired.

  The place name importance determining unit 220 functions as a place name importance determining unit of the present invention, acquires a document ID from the snippet inquiry receiving unit 210, accesses the position information DB 030 and the place name DB 110, and makes a snippet inquiry for place name information with importance. Return to the reception unit 210.

  Further, the location information DB 030 is accessed based on the document ID, and the location name information (location name word, location name canonical name, location in the document, latitude, longitude) included in the document is acquired.

  Further, the place name DB 110 is accessed based on the place name canonical name of the place name information acquired above, and the width (latitude width, longitude width) implied by the place name canonical name is obtained.

  The place name importance determination unit 220 calculates the importance of each place name based on the latitude width and longitude width obtained above. As an example of the importance calculation method, the importance expressed by the following equation (1) can be considered as a method based on the assumption that a place name indicating a more specialized place is important.

Weight _{d, g} = 1 / max (latitude width, longitude width) (1)
Here, Weight _{d, g} indicates the importance of the place name g in the document d, and max () is a function that returns the maximum value among the input values.
In addition, there is a possibility that the location information DB 030 includes location name information that causes noise due to an analysis error caused by the geographic indexer 020 or the presence of a location name that is not closely related to the subject that appears in the document. As a countermeasure to this problem, it is conceivable that the place names included in each document are grouped with place names existing within a certain distance, and only the place names of the group to which the most place names belong are used. In this case, the group selection method may consider the location information specified by the user and the proximity of each group.

  The sentence evaluation unit 230 functions as a sentence evaluation unit of the present invention, acquires a search keyword, a document ID, and place name information with importance from the snippet inquiry reception unit 210, and is considered to be important in a specified document. Is extracted and returned to the snippet inquiry reception unit 210.

  First, the text DB 010 is accessed to acquire a document, and the position where the input search keyword appears in the document is specified.

  Based on the place name information with importance calculated by the place name importance degree determination unit 220 and the appearance position of the search keyword, a partial document having a high degree of importance is acquired from the partial documents in the document. The following formula (2) is given as an example of the importance of each partial document.

I _{d, s, q, g} = FreqOfKeyword (d, s, q) + Weighted FreqOfGeord (d, s, g) (2)
Here, I _{d, s, q, g} is the importance of the partial document s in the document d searched with the search keyword q and the place name g, and FreqOfKeyword (d, s, q) is the search keyword q of the document d. A function indicating the number of appearances in the partial document s, Weighted FreqOfGeord (d, s, g) is a function for adding all the weights of the place names g appearing in the partial document s of the document d.

  The function WeightedFreqOfGeoword (d, s, g) in equation (2) is

It is. Here, G _s represents a set of place names included in the sentence s, and Weight _{d, g} represents the importance of the place name g in the document d as described above.

  The search engine 310 receives a search keyword and position information from the search application 320, accesses the text DB 010 and the position information DB 030, acquires a search result document, and queries the snippet query reception unit 210 for a document to be presented as the search result. The snippet from the snippet inquiry reception unit 210 is returned to the search application 320 as a search result.

  The search application 320 is an interface with the user, receives input of search conditions (search keywords, position information) from the user, accesses the search engine 310, and presents the obtained search results to the user.

  The location information DB 030 and the place name DB 110 are created by the geo indexer 020 and the place name implication range specifying unit 100 as pre-processing, respectively.

  Of the processing processes of the present invention, a flow for search processing is shown in FIG. In FIG. 4, first, in step S <b> 1, a search keyword and position information are input from the user through the search application 320. Next, in step S <b> 2, the search application 320 transmits the search keyword and position information to the search engine 310. In step S3, the search engine 310 accesses the text DB 010 and the position information DB 030 to obtain a search result document.

  In step S <b> 4, the search engine 310 transmits the obtained search result document and the search keyword to the snippet inquiry reception unit 210. In step S <b> 5, the snippet inquiry reception unit 210 transmits the document ID to the place name importance level determination unit 220.

  Next, in step S6, the place name importance determining unit 220 accesses the position information DB 030 and acquires the position information held by the document with the document ID. Next, in step S7, the place name importance level determination unit 220 accesses the place name DB 110 and obtains the size implied by each place name included in the position information.

  Next, in step S8, the place name importance determining unit 220 calculates the importance of each place name based on the width implied by the obtained place name. Next, in step S <b> 9, the place name importance level determination unit 220 returns the position information with importance level to the snippet inquiry reception unit 210. In step S <b> 10, the snippet inquiry reception unit 210 transmits the search keyword, document ID, and position information with importance to the sentence evaluation unit 230.

  Next, in step S11, the sentence evaluation unit 230 accesses the text DB 010 based on the document ID, and acquires a document with the document ID. Next, in step S12, the sentence evaluation unit 230 specifies where in the document the search keyword appears.

  In step S13, the sentence evaluation unit 230 calculates the importance of the partial document from the appearance position of the search keyword, the appearance position of the place name, and the importance of the place name. Next, in step S14, it is determined whether or not the importance level has been calculated for all the partial documents. If it has been calculated, in step S15, the sentence evaluation unit 230 uses the highest importance partial document as a snippet as a snippet inquiry reception unit. To 210.

  Next, in step S16, it is determined whether or not snippets have been acquired for all search results. If acquired, the snippet inquiry reception unit 210 transmits all snippets to the search engine 310 in step S17.

  In step S <b> 18, the search engine 310 adds a snippet to the search result and transmits it to the search application 320. Next, in step S19, the search application 320 formats the received search result and presents it to the user.

  If it has not been calculated in the determination in step S14, the process returns to step S13. If it has not been acquired in the determination in step S16, the process returns to step S4.

  The search result snippet generation method that takes into account the scope of the place name of the present invention executes, for example, the functions and processes described in FIGS.

  In addition, a program for causing a computer to execute the search result snippet generation method of the present invention is constructed.

  Also, the present invention can be realized by configuring a part or all of the functions of each means in the search result snippet generation device of the present embodiment by a computer program and executing the program using the computer. It goes without saying that the procedure in the search result snippet generation method of the embodiment can be configured by a computer program and the program can be executed by the computer, and the computer can read the program for realizing the function on the computer. Recording media such as FD (Floppy (registered trademark) Disk), MO (Magneto-Optical disk), ROM (Read Only Memory), memory card, CD (Compact Disk) -ROM, DVD (Dig) It is possible to record and save or distribute it in (Versatile Disk) -ROM, CD-R, CD-RW, HDD, removable disk or the like. It is also possible to provide the above program through a network such as the Internet or electronic mail.

  Although the present invention has been specifically described above based on the embodiments, the present invention is not limited to the above-described embodiments and can be widely applied.

The block diagram which shows one embodiment of the snippet production | generation apparatus of this invention. Explanatory drawing which shows an example of position information DB in one embodiment of the snippet production | generation apparatus of this invention. Explanatory drawing which shows an example of place name DB in one embodiment of the snippet production | generation apparatus of this invention. The flowchart which shows one example of embodiment of the snippet production | generation method of this invention.

Explanation of symbols

  010 ... Text database, 020 ... Geographic indexer, 030 ... Location information database, 100 ... Place name implication range specifying part, 110 ... Place name database, 210 ... Snippet inquiry reception part, 220 ... Place name importance determination part, 230 ... Sentence evaluation part, 310 ... Search engine, 320 ... Search application.

Claims (6)

Search a text set that contains the specified search keyword and related to a specific location by specifying a search keyword and location information that specify the content of the text from a set of text that exists inside the computer or that can be accessed via a computer network In the search result snippet generation device that acquires a snippet showing an overview of each search result,
Geographic index means for extracting a place name existing in the text of the text set and assigning latitude and longitude to the extracted place name;
A place name entailment range specifying means for specifying an area implied by the place name;
Based on the latitude / longitude of the place name assigned by the geographical index means and the information on the area implied by the place name specified by the place name implication range specifying means , an expression having a greater importance as the area information becomes narrower with a place name importance degree determination unit that performs processing for calculating the importance of the place names in the text,
Based on the importance of the given search keyword and place name, the sentence that is a partial document in the text is evaluated with the sum of the importance depending on the search keyword and the importance of the place name. Sentence evaluation means for performing processing to extract and output as a snippet;
The importance of the place name in the text searched based on the designated search keyword and position information is obtained by inquiring the place name importance determining means , and the importance of the obtained place name and the search keyword are obtained. A search result snippet generation device that takes into account a range meant by a place name, comprising: a snippet inquiry reception unit that transmits to a sentence evaluation unit and receives a snippet output from the sentence evaluation unit. The importance of the place name transmitted by the snippet inquiry reception means to the sentence evaluation means is the importance of one or more place names that are close to the specified position information and exist within a certain distance.
An apparatus for generating a search result snippet in consideration of a range implied by a place name according to claim 1. Search a text set that contains the specified search keyword and related to a specific location by specifying a search keyword and location information that specify the content of the text from a set of text that exists inside the computer or that can be accessed via a computer network In the search result snippet generation method to obtain a snippet that shows an overview of each search result,
Geographic index means for extracting a place name existing in the text of the text set and assigning latitude and longitude to the extracted place name;
The place name implication range specifying means specifies a place name entailment range specifying step for specifying an area implied by the local name,
The place name importance degree determination means is configured such that, based on the latitude / longitude of the place name assigned by the geo index step and the area information implied by the place name specified by the place name implication range specifying step, the area information becomes narrower. A place name importance determination step for calculating the importance of the place name in the text using an expression that has a large importance ;
A snippet inquiry accepting unit inquires and acquires the importance of the place name in the text searched based on the designated search keyword and position information to the place name importance degree determining means;
A snippet inquiry accepting unit transmitting the importance of the acquired place name and a search keyword to the sentence evaluating unit;
Based on the importance of the place name transmitted from the snippet inquiry acceptance means and the search keyword , the sentence evaluation means determines the sentence that is a partial document in the text as the sum of the importance depending on the search keyword and the importance of the place name. a sentence evaluating step tHAT have evaluated, performs processing for output as snippets extracted sentences are considered to be important,
A method for generating a search result snippet in consideration of a range meant by a place name, characterized in that: a snippet inquiry receiving unit includes a step of receiving a snippet output from the sentence evaluating unit. The importance of the place name transmitted by the snippet inquiry reception means to the sentence evaluation means is the importance of one or more place names that are close to the specified position information and exist within a certain distance.
A method of generating a search result snippet in consideration of a range meant by a place name according to claim 3. A program for generating a search result snippet that allows a computer to function as each unit according to claim 1 or 2 and that takes into account a range that a place name means. 6. A computer-readable recording medium on which a search result snippet generation program that takes into consideration the scope of the place name according to claim 5 is recorded.

Images