JP2011198162A - Representative notation extracting device, method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To extract a suitable representative notation from a given group of character strings.SOLUTION: A plurality of given character strings are divided into tokens respectively. On all token arrays in which one or more tokens are directly connected to constitute all or a part of the character strings, the number of character strings including the token array out of the plurality of character strings is counted. The token array in which the counted number of character strings is a predetermined threshold or more, that is, the token array excluding the token array included in the other long token array out of the token array, is selected. The selected token array is output as the extracted result.

Description

  The present invention relates to character string processing.

  In recent years, with the progress of information processing technology and communication technology, various web services provided using a server computer on the web (WWW) are rapidly spreading. As important technical fields that support web services, there are processing of information on character strings such as product names and processing of information on facilities such as stores.

  Regarding character strings, in various services such as electronic commerce, many character strings such as product names are accumulated and used for keyword extraction and search. In this type of field, a proposal is made to measure the frequency of words in a character string, automatically convert the frequency to a word “weight” and automatically extract keywords (for example, see Patent Document 1), or dependency analysis. There is a proposal of keyword extraction using, for example (see Patent Document 2). These conventional technologies are suitable for extracting a characteristic keyword by giving one content such as an article.

  In addition, information on facilities such as stores is obtained by associating facility names and related information with positions such as latitude and longitude, and is called position information, facility information, POI (Point Of Interest), etc. It will be referred to as “POI”. As an example of utilizing POI, a navigation service, portal site service, and other websites accept keywords such as location and type of business, search for POI, and display a web screen in the form of a map or a list. Yes (see Non-Patent Document 1, for example). In this type of service, a POI search result is displayed from a map database prestored with facility information (referred to as “POI”) representing a store or the like using the input location and keyword as a query (search request). .

Japanese Patent Laid-Open No. 08-30627 JP 2004-240576 A JP 2002-340604 A

Yahoo Japan Corporation, "Yahoo! Local Information", [online], [Search February 5, 2010], Internet <URL: http://local.yahoo.co.jp/>

  Regarding the processing of character strings, the problems that existed in the prior art (for example, Patent Documents 1 and 2) are character strings that express them in common terms from a group of similar character strings related to the same object (this application) Then, we will call it “representative notation”), and there is no such extraction technique for the potential demand to extract.

For example, in a portal electronic commerce website (hereinafter referred to as “portal EC site”) called a tenant format (mall format) or the like, even if the product is the same in the JAN code that uniquely identifies the product, a plurality of store openings When a store (shop, store) sells, a product name registered as a character string may differ depending on the store. For this reason, in order to implement a plan that introduces a plurality of stores that handle a specific same product or to perform a search without omission, it is desirable to extract a representative notation suitable for expressing the same product. However, performing such extraction manually has a heavy load, and it has been more difficult for portal EC sites with a large number of products handled.
In view of the above problems, an object of the present invention is to provide a technique for extracting an appropriate representative notation from a given character string group.

  On the other hand, with respect to POI processing, a problem that has existed in the prior art (for example, Non-Patent Document 1) is that the range of facilities constituting a complex facility cannot be easily specified in the map database.

  As a premise, first, in the service as described above for searching for POI (for example, Non-Patent Document 1), if the input location is a facility name such as a building name, the tenant in the building is used as a keyword. The intention is to search for a match. However, in such a case, if the facility name and the tenant in the facility are not properly associated in advance in the database, it will not be possible to obtain an appropriate search result, for example, only the facility name will be displayed in the store name. There was a problem.

  In connection with this problem, there is also a proposal to display a child POI list when a parent-child relationship is set in the POI, a list of child POIs is linked to the parent POI, and the parent POI is selected in the destination setting of navigation. However, this proposal does not disclose a specific method for associating a parent POI and a child POI, and it is a burden on a service provider or the like to perform such association manually.

In particular, in recent years, there are many cases where a plurality of facilities such as buildings and areas are integrated with each other to form a complex facility such as a large-scale commercial facility such as “XX Town” and “XX Hills”. Buildings and areas in this type of complex are easy to navigate. For this reason, in order to present an appropriate search result in a search for a tenant or the like, a plurality of facilities included in a complex facility must be associated with each other on the database. However, such a complex facility has a large number of POIs, and there is a problem that it is a heavy burden for a service provider to perform association manually.
In view of the above problems, an object of the present invention is to easily specify the range of facilities constituting a complex facility in a map database.

  Based on the above object, one aspect (A1) of the present invention is a representative notation extraction apparatus, in which a plurality of given character strings are divided into tokens, and one or more tokens are directly connected. Counting means for counting the number of character strings including the token string in the plurality of character strings, and the counted number of character strings, for all token strings constituting all or part of the character string A token selection means for selecting a token string that is a token string that is greater than or equal to a predetermined threshold value, and excluding the token string included in another longer token string, and the selected token string; And a result output means for outputting as an extraction result.

  Another aspect (A5) that captures the above aspect as a method is a representative notation extraction method, in which a plurality of given character strings are each divided into tokens, and one or more tokens are directly connected. For all token strings that constitute all or part of the character string, a counting step for counting the number of character strings including the token string in the plurality of character strings, and the counted number of character strings is predetermined. A token selection step of selecting a token string that is greater than or equal to the threshold value of the token string, and excluding the token string included in the other longer token string from the token string, and extracting the selected token string A result output step of outputting as a result is executed by a computer.

  The representative notation extraction program, which is another aspect (A6) obtained by capturing each aspect as a computer program, is a computer program for controlling a computer, and divides a plurality of given character strings into tokens. And a count for counting the number of character strings that include the token string in the plurality of character strings, for all token strings that form one or more of the character strings by directly connecting one or more of the tokens And a token sequence in which the number of character strings counted is a token sequence that is greater than or equal to a predetermined threshold value, and excluding token sequences included in other longer token sequences in the token sequence. A token selection step and a result output step for outputting the selected token string as an extraction result. Characterized in that to execute the Yuta.

  As described above, by selecting the longest token string included in a predetermined number of character strings from tokens obtained by dividing the character string group given as corresponding to the same product etc. Since the greatest common divisor character string that is most commonly used in the column group is obtained, an appropriate representative notation expressing the character string group can be extracted.

  In another aspect (A2) of the present invention, in any one of the above aspects, a character string including the largest number of selected token strings is selected as a reference character string from the plurality of given character strings, and the reference character It is characterized by having a rearrangement means for rearranging the selected token string in accordance with the order of the tokens in the string.

  Thus, by sorting the selected tokens into the notation that contains the most selected tokens, that is, in the order of the exhaustive notation, the appropriate consideration that was taken into account when the exhaustive notation was made The order of expression is reflected. In particular, even when the notation is a product name, the expression order takes into account the type and characteristics of the product, and an appropriate representative notation that is easy to understand is obtained.

  In another aspect (A3) of the present invention, in any one of the aspects described above, there is provided a duplication deletion unit that deletes tokens that overlap each other among the token strings that constitute the selected token string. Features.

  In this way, by deleting duplication in token units from the selected token string, duplication can be eliminated for the representative notation in which token strings are arranged, and an appropriate representative notation without waste can be obtained. In particular, even when the character string is a product name, it becomes a concise expression that eliminates duplication, and an appropriate representative product name that can be easily understood by anyone can be obtained.

  According to another aspect (A4) of the present invention, in any one of the above aspects, a map data storage unit storing map data including a facility polygon and a POI, and a facility polygon is selected from the map data storage unit A polygon selection means; a POI selection means for selecting a POI associated with the selected facility polygon from the map data storage means; and a facility name candidate for each facility polygon by analyzing the name of the selected POI. A representative notation extraction device having candidate extraction means for extracting, wherein the dividing means divides each POI name into a plurality of given character strings into tokens, and each divided token is The counting means performs the counting, the token selecting means performs the selection of a token string, and the result output means. The selected token string is output as a facility name candidate extraction result, and the representative notation extraction device further stores the output facility name candidate as the name of the selected facility polygon in the map data storage means. It is characterized by having map data update means to be made.

  In this way, a polygon representing a building or the like is represented by dividing the token into a token group, selecting a token string based on the number of character strings, etc., using the name of the POI having a relation such as the position included in the polygon as a character string group. As a notation, facility name candidates such as building names can be easily extracted without depending on human resources. The map data can be updated efficiently by using the facility name candidate as the name corresponding to the polygon.

  Other aspect groups suitable for combination with the above-described aspect group of the present invention are shown below.

  That is, one mode (B1) is a map data processing device, which includes a map data storage unit storing map data including facility polygons and POIs, and a plurality of maps located in the vicinity of each other from the map data storage unit. The polygon selection means for selecting the facility polygon, the POI selection means for selecting the POI related to the selected facility polygon from the map data storage means, and the name of the POI selected as related to the facility polygon And a candidate extracting means for extracting facility name candidates for each facility polygon and, based on the commonality of the extracted facility name candidates, all or some of the selected facility polygons are mutually And an associating means for associating.

  Another aspect (B5) that captures the above aspect as a method is a map data generation method, in which a plurality of facilities located in the vicinity of each other from map data storage means storing map data including facility polygons and POIs. A polygon selection step for selecting a polygon, a POI selection step for selecting a POI associated with the selected facility polygon from the map data storage means, and an analysis of the name of the POI selected as associated with the facility polygon Then, a candidate extracting step for extracting facility name candidates for each facility polygon and all or some of the selected facility polygons are associated with each other based on the commonality of the extracted facility name candidates. The association step is executed by a computer.

  A map data generation program according to another aspect (B6) that captures the above aspect as a computer program is a computer program for controlling a computer, and stores map data including facility polygons and POIs. Means for selecting a plurality of facility polygons located in the vicinity of each other, a POI selection step for selecting a POI associated with the selected facility polygon from the map data storage unit, and a facility polygon. A candidate extracting step of analyzing the name of the POI selected as an object to extract a facility name candidate for each facility polygon, and a plurality of facilities selected based on the commonality of the extracted facility name candidates An association step for associating all or part of the polygons with each other; And characterized by causing a computer to execute the.

  In this way, for each facility polygon that is a facility polygon representing a building or the like and close to each other, the name of the POI associated with the facility polygon is analyzed to extract facility name candidates of the facility polygon, and based on the commonality of the facility name candidates This makes it easy to correlate facility polygons included in the same large-scale commercial facility.

  According to another aspect (B3) of the present invention, in any one of the above aspects, the character string is divided by a dividing unit that divides a plurality of given character strings into tokens, and the one or more tokens are directly connected to each other. Or, for all token strings constituting a part, the counting means for counting the number of character strings including the token string in the plurality of character strings, and the counted number of character strings is equal to or greater than a predetermined threshold Token selection means for selecting a token string that is a large token string and excluding the token string included in another longer token string from the token string, and a result of outputting the selected token string as an extraction result Output means.

  In this way, by selecting the longest token string included in a predetermined number of character strings from the tokens obtained by dividing the name of POI that is a given character string group by character type, etc. Since the most common divisor notation used most commonly is obtained, an appropriate representative notation such as a large-scale commercial facility name expressing the character string group can be extracted.

  According to another aspect (B2) of the present invention, in any one of the aspects described above, the facility name candidate related to the commonality that is the basis for the association by the association unit corresponds to the plurality of associated facility polygons. The map data update means updates the map data stored in the map data storage means as a name to be used.

  In another aspect (B4) of the present invention, in any one of the above aspects, the token output as an extraction result by the result output unit based on a plurality of POI names related to the associated facility polygons Map data updating means for updating the map data stored in the map data storage means with a column as a name corresponding to the plurality of associated facility polygons.

  As described above, a common facility name candidate that is the basis of association between facility polygons, or a token string extracted from the POI name associated with the associated facility polygon corresponds to the associated facility polygon group. By updating map data as facility names, etc., there is no burden of manually assigning names of large-scale commercial facilities, etc., and customary and de facto nicknames that refer to specific building groups and regions are also set appropriately It becomes possible to do.

  It should be noted that a category (method for the apparatus, program for the method, etc.) different from each of the above-described modes and more specific modes described below are also included in the present invention. In the case of the different categories, “means” is appropriately replaced with “structure” or the like according to the category, such as “step”.

  According to the present invention, it is possible to provide a technique for extracting an appropriate representative notation from a given character string group.

The functional block diagram which shows the structure of 1st Embodiment of this invention. The figure which illustrates the information (data) used by 1st Embodiment of this invention. The flowchart which shows the process sequence in 1st Embodiment of this invention. The functional block diagram which shows the structure of 2nd Embodiment of this invention. The flowchart which shows the process sequence in 2nd Embodiment of this invention. The functional block diagram which shows the structure of 3rd Embodiment of this invention. The flowchart which shows the process sequence in 3rd Embodiment of this invention. The figure which illustrates the information (data) used by 3rd Embodiment of this invention. The conceptual diagram which shows on the map the state in which the facility polygon and POI are selected in 3rd Embodiment of this invention. The conceptual diagram showing the correspondence of the selected facility polygon and POI in 3rd Embodiment of this invention. The conceptual diagram which shows the state of the correlation between facility polygons on a map in 3rd Embodiment of this invention. In 3rd Embodiment of this invention, the conceptual diagram showing the state with which some selected facility polygons are linked | related (grouped). The functional block diagram which shows the structure of 4th Embodiment of this invention. The flowchart which shows the process sequence in 4th Embodiment of this invention.

  Next, a plurality of modes for carrying out the present invention (each referred to as “embodiment”) will be described with reference to the drawings. It should be noted that assumptions common to those already described in the background art and problems are omitted as appropriate.

[1. Configuration of First Embodiment]
The first embodiment relates to a representative notation extraction device (hereinafter referred to as “this device”) 1 shown in FIG. 1 (configuration diagram). The present apparatus 1 has a general computer configuration of at least an arithmetic control unit 5 such as a CPU, a storage device 6 such as an external storage device (HDD or the like) or a main memory, and a communication means 7 (not shown). LAN adapter, etc.). In the apparatus 1, a predetermined computer program (not shown) stored (installed) in advance in the storage device 6 controls the arithmetic control unit 5, whereby elements (11, 12, 14, etc.) shown in FIG. Etc.).

  Among the elements realized by the computer program as described above, the information storage means is a storage device 6 in which various databases (also referred to as “DB”), files, variables such as arrays, various stacks and registers, system settings, etc. It can be realized in any format such as a value. Among such storage means, the character string storage means 11 corresponds to a certain JAN code on a portal EC site in a tenant format (mall format) as illustrated in FIG. 2 as a basis for extracting representative notation. And a plurality of character strings that are registered product names. Each means other than the storage means is a processing means for realizing and executing the following information processing functions and operations.

[2. Operation of First Embodiment]
Using this apparatus 1 configured as described above, the administrator of the portal EC site handles the same specific product with a different product name depending on the store as illustrated in FIG. An example of processing in the case of obtaining a representative notation in order to provide a planning corner that introduces a plurality of stores across is shown below. In this case, when the administrator selects the five character strings shown in FIG. 2 on an appropriate user interface screen or the like, the character string designating unit 12 accepts such an operation, and each selected character string is selected. Is given to the dividing means 14. The plurality of character strings selected here are also referred to as “character string groups”.

  The character string designating unit 12 is not limited to accepting a selection operation from a user such as an administrator. For example, the character string designating unit 12 is based on a database in which a character string group is registered by each mail order company for each of many products. The character string group for each product may be selected one after another, and the procedure for extracting the representative notation from each character string group may be controlled as follows. In any case, the flowchart of FIG. 3 shows a processing procedure for extracting a representative notation from a given character string group as follows.

[2-1. (Division into tokens)
That is, first, the dividing unit 14 divides each given character string into tokens (step S14). A “token” is a unit obtained by dividing a character string according to a predetermined standard. As a method of dividing the character string into tokens, it may be freely selected such as dividing each character group such as a character type or dividing by a morpheme.

For example, a character string group given as a basis for extracting representative notation is

Earth Defense Force Mobile Suit AB100
Earth Defense Force Mobile Suit "AB100"
Earth Defense Force AB100
Defense Army AB100 Mobile Suit Defense Army AB100

And call them as the original character string (character string group). In addition, the brackets “” of the symbols shall also constitute a character string, and when describing the symbols in the character string, they are indicated by double brackets, such as ““ ”and“ “”.

The result of dividing these individual character strings into tokens for each type of character group, that is, kanji, katakana, alphabet, number, or symbol is as follows. The dividing position is indicated by “/” (half-width slash).

Earth Defense Force / Mobile Suit / AB / 100
Earth Defense Force / Mobile Suit / "/ AB / 100 /"
Earth Defense Force / AB / 100
Defense Force / AB / 100 / Mobile Suit Defense Force / AB / 100

Furthermore, it may be further divided by tokens contained in other character strings, and if "Earth Defense Force" is divided by "Defense Army"

Earth / Defense Force / Mobile Suit / AB / 100
Earth / Defense Force / Mobile Suit / "/ AB / 100 /"
Earth / Defense Force / AB / 100
Defense Force / AB / 100 / Mobile Suit Defense Force / AB / 100

It becomes.

[2-2. (Count of the number of strings)
Subsequently, the counting means 15 is arranged for all token strings that constitute one or more of the character strings in which one or more tokens are directly linked (limited to those that are also directly linked in the original character string). Among the plurality of character strings, the number of character strings including the token string is counted (step S15). This applies so-called “N-gram” with the application unit as a token.

Specifically, all possible token strings are created for each character string as “all token strings in which one or more tokens are directly connected to constitute all or part of the character string”. For example, from the string "Defense Army / AB / 100"

"Defense Army"
"AB"
"100"
"Defense Army / AB"
"AB / 100"
Defense Force / AB / 100

6 types of token sequences are generated. Note that a token string such as “Defense Army / 100” (excluding the intermediate token “AB” in between) that is not directly connected in the original character string is not created in this example.

  In this way, for every token string created from each character string, the number of times it appears in all the original character strings, that is, how many character strings are included is counted.

[2-3. (Token selection)
Further, the token selection means 16 is a token string in which the counted number of character strings is equal to or larger than a predetermined threshold value or larger than the threshold value, and includes a token string included in another longer token string in the token string. The removed token string (hereinafter referred to as “longest token string”) is selected, and the others are cut off (step S16). For example, in the above example, the longest token string that appears in 50% or more of the original character string group is

Earth / Defense Force (3)
Defense Force (5)
Earth (3)
Defense Force / AB / 100 (3)
AB / 100 (5)
Mobile suit (3)

(The number in parentheses is the number of strings including the token string.)

Of these, “Defense Army” and “Earth” are included in the longer token string “Earth / Defense Force”, and “AB / 100” is also included in “Defense Force / AB / 100”. (Excluded items are indicated by a cross),

Earth / Defense Force (3)
× Defense Force (5)
× Earth (3)
Defense Force / AB / 100 (3)
× AB / 100 (5)
Mobile suit (3)

The remaining longest token sequence is

Earth / Defense Force (3)
Defense Force / AB / 100 (3)
Mobile suit (3)

These are the token sequences that currently remain.

[2-4. (Duplicate removal)
Further, the duplication deletion unit 17 deletes the selected token strings that overlap each other among the longest token strings among the longest token strings remaining as described above (step S17). For example, the above

Earth / Defense ForceDefense Force / AB / 100
Mobile suit

Then, if you delete the duplicate “Defense Army” and cut it out as an independent token,

Earth Defense Force
AB / 100
Mobile suit

These are the token sequences that are currently selected.

[2-5. (Sort and output results)
Next, the rearranging means 18 selects a character string including the most currently selected token string as a reference character string from the given character string group, and matches the order of the tokens in the reference character string. The selected token string is rearranged (step S18). Here, as a criterion for selecting a character string that includes the most token tokens that are currently selected, a character string that includes the most token tokens that are currently selected may be used as a reference character string. The standard of For example, here, the token string that includes the currently selected token string and that includes the maximum number of character strings (3, 5, etc.) including the respective token strings is represented by the token string “ It is assumed that the character string is “most contained” and is selected as the reference character string.

For example,

Earth (3)
Defense Force (5)
AB / 100 (5)
Mobile suit (3)

Based on the above, for each given string, calculate the total number of strings including the token string that the string contains,

Earth / Defense Force / Mobile Suit / AB / 100: 3 + 5 + 3 + 5 = 16
Earth / Defense Force / Mobile Suit / "/ AB100 /": 3 + 5 + 3 + 5 = 16
Earth / Defense Force / AB / 100: 3 + 5 + 5 = 13
Defense Force / AB / 100 / Mobile Suit: 5 + 5 + 3 = 13
Defense Force / AB / 100: 5 + 5 = 10

It becomes.

Here, “Earth / Defense Force / Mobile Suit / AB / 100” and “Earth / Defense Force / Mobile Suit /“ / AB100 / ”” happen to be the same with the maximum number of strings of 16 times. However, regardless of which is selected, the following result is obtained. That is, according to the order of appearance in this character string, if the token string currently selected as a result of the process up to the deletion of duplicates (step S17) is rearranged,

`` Earth Defense Force Mobile Suit AB 100 ''

It becomes. Finally, the result output means 19 uses this rearranged token string as an extraction result.

"Earth Defense Force Mobile Suit AB100"

(Step S19).

[3. Effects of the first embodiment]
As described above, in the first embodiment, a longest token string included in a predetermined number of character strings is obtained from tokens obtained by dividing a character string group given as corresponding to the same product or the like by character type or the like. By selecting, the greatest common divisor character string used most commonly in the character string group is obtained, so that an appropriate representative notation expressing the character string group can be extracted.

  Further, in the first embodiment, the selected tokens are rearranged in the notation including the most selected tokens, that is, in the order of the exhaustive notation (FIG. 3, step S18), thereby providing the exhaustive notation. Appropriate expression order is taken into account when is created. In particular, even when the notation is a product name, the expression order takes into account the type and characteristics of the product, and an appropriate representative notation that is easy to understand is obtained.

  Furthermore, in the first embodiment, by deleting duplication in token units from the selected token string (step S17), duplication can be eliminated for the representative notation in which token strings are arranged, and an appropriate representative without waste. The notation is obtained. In particular, even when the character string is a product name, it becomes a concise expression that eliminates duplication, and an appropriate representative product name that can be easily understood by anyone can be obtained.

[4. Second Embodiment]
An example of obtaining a facility name candidate from a POI name by applying the representative notation extraction device shown in the first embodiment will be described as a second embodiment. As shown in FIG. 4, the representative notation extraction device 2 of the second embodiment has means 21 to 24 and 26 for realizing and executing the effects described later regarding map data processing (left side of FIG. 4), of which As a specific configuration content of the candidate extraction unit 24, a configuration according to the first embodiment (inside the alternate long and short dash line in FIG. 4; reference numerals 14 to 19) is provided.

  In the second embodiment, the candidate extraction means 24 is given the POI group selected by the POI selection means 23, so that the character string storage means 11 and the character string designation shown in the first embodiment (FIG. 1). The means 12 is not essential, but it goes without saying that a work area (not shown) is appropriately used in the processing. Subsequently, the processing procedure in the second embodiment is shown in the flowchart of FIG.

  First, the map data storage means 21 stores map data including facility polygons and POIs in advance. Then, as shown in FIG. 5, first, the polygon selection means 22 selects facility polygons from the map data storage means 21 in response to a user's operation or the like or by automatic processing based on an appropriate standard (step S22). The POI selection means 23 selects the POI associated with the selected facility polygon from the map data storage means 21 (step S23). Here, “associated” is typically a case where the POI position coordinates are included in a polygon, but is not limited thereto, and the facility and the POI are preliminarily identified with each other's identification information (ID) or the like. Including the case where it is associated.

  Then, the candidate extracting unit 24 analyzes the name of the selected POI and extracts the facility name candidate of each facility polygon (step S24). The configuration is based on the first embodiment.

  That is, the dividing unit 14 divides each POI name into a plurality of given character strings into tokens (step S14), and the counting unit 15 performs the counting based on each divided token ( In step S15), the token selecting means 16 performs the selection of the token string (step S16). The means 17 and 18 conform to the first embodiment (steps S17 and S18), but may be omitted, and the description thereof is omitted here. Then, the result output unit 19 outputs the selected token string to the map data update unit 26 as a facility name candidate extraction result (step S19).

  The facility name candidate that is the extraction result output in this way is received by the map data updating unit 26 and stored in the map data storage unit 21 as the name of the selected facility polygon (step S26).

For example, the name of the POI included in the facility polygon of a building with no name set is

"Pastry XX Nerio Tower Koshigaya"
“XX Bank Nerio Tower Koshigaya Branch”
"Sushi XX Nerio Tower Koshigaya"
“○○ Shop Koshigaya Nerio Tower”

Then, “store” and “branch” behind the name are ignored as standard expressions, and as a result of processing using tokens as in the first embodiment, the facility name candidate is “Nerio Tower Koshigaya”.

  In this way, a polygon representing a building or the like is represented by dividing the token into a token group, selecting a token string based on the number of character strings, etc., using the name of the POI having a relation such as the position included in the polygon as a character string group. As a notation, facility name candidates such as building names can be easily extracted without depending on human resources. The map data can be updated efficiently by using the facility name candidate as the name corresponding to the polygon.

[5. Third Embodiment]
In the said 2nd Embodiment, the example which extracted the name of the single plant | facility from the name of POI was shown with the representative description extraction device according to 1st Embodiment. In contrast, the third embodiment shows a map data processing apparatus 3 (FIG. 6) that extracts the name of a complex facility (such as a large-scale commercial facility) including a plurality of elements such as buildings from the name of the POI. The technique for extracting the name of the facility in the third embodiment is not limited to the representative notation extraction device as shown in the first and second embodiments.

  Specifically, the map data processing device 3 according to the third embodiment controls the arithmetic control unit 5 including the storage device 6 and the communication means 7 with a predetermined computer program, thereby providing the following operational effects (FIG. Each processing means (FIG. 6) for realizing 7) is realized.

  First, map data storage means 31 stores map data including facility polygons and POIs in advance. For example, FIG. 8 shows an example of the map data storage means 31. The facility polygon database (FIG. 8 (1)) and the POI database (FIG. 8 (2)) are associated with each other by a polygon ID (shown by a broken line). This is an example.

  Then, the polygon selection unit 32 selects a plurality of facility polygons located near each other from the map data storage unit 31 by a user operation or automatic processing (step S32), and the POI selection unit 33 selects the selected facility A POI associated with the polygon is selected from the map data storage means 31 (step S33). Here, “related” to the facility polygon is typically a case where the POI position coordinates are included in the polygon, but not limited thereto, as illustrated in FIG. This includes the case where the POI is associated with identification information such as a polygon ID.

  As an example, consider a case where a user selects facility polygons of four buildings that are close to each other on map data. Here, each building for which the facility polygon is selected is called “tower building”, “east building”, “west building”, “another building”, and such names are set in advance in the facility polygon. There is no need. Also, consider a case where POI selection means 33 selects 12 POIs (POI1 to POI12) in total, as follows, for POIs previously associated with each selected facility polygon and polygon ID.

Tower Building POI1: “XX Bank Amor Town Branch”
POI2: “Amol Town Clinic”
POI3: “Amortown Post Office”

East Wing POI4: “Western confectionery XX Amor Town”
POI5: “XX Shop Amol Town”
POI6: “Japanese food ○○ Tei Amor Town”

West Building POI7: “Tempura XX”
POI8: “Super XX Amol Town”
POI9: “Amortown XX Event Hall”

Another building POI10: “XX Factory Higashimachi Building”
POI11: “Beauty salon XX Higashimachi Building”
POI12: “XX Ticket Higashimachi Building Branch”

  FIG. 9 is a conceptual diagram represented on a map and FIG. 10 shows the correspondence between facility polygons and POIs when the facility polygons of four buildings and 12 POIs are selected as described above.

  And the candidate extraction means 34 extracts a facility name candidate for every facility polygon by analyzing the name of each POI selected corresponding to the facility polygon (step S34). The technique for performing this extraction is not limited to the representative notation extraction device as shown in the first and second embodiments, and language analysis and other algorithms may be freely selected. Thereafter, the associating unit 35 associates, or groups, all or some of the selected facility polygons with each other based on the commonality of the extracted facility name candidates (step S35).

  In the above example, for the tower building, the east building, and the west building, “Amortown”, which is included in many of the POI names, is a facility name candidate. Since the facility name candidates are common in this way, the association means 35 associates the tower building, the east building, and the west building as a complex facility (FIGS. 11 and 12). The location of the other building was close, but the facility name candidate based on the common part of the POI name is “Higashimachi Building”, which is different from the other 3 buildings and is not included in the complex facility. (FIGS. 11 and 12).

  Subsequently, the map data update unit 36 sets the facility name candidate “Amortown” related to the commonality that is the basis of the association by the association unit 35 to each of a plurality of associated facilities, that is, the tower building, the east building, and the west building. Map data such as facility information stored in the map data storage means 21 is updated as the name of the complex facility corresponding to the polygon (step S36).

  As described above, in the third embodiment, for each facility polygon that represents a building or the like and is close to each other, the name of the related POI is analyzed to extract facility name candidates for the facility polygon, and the facility polygon is extracted. Based on the commonality of name candidates, facility polygons included in the same large-scale commercial facility can be easily associated with each other.

  In the third embodiment, as described above, the map data is updated with the common facility name candidates that are the basis for the association between the facility polygons as the facility names corresponding to the associated facility polygon groups. (Step S36), it is possible to set appropriately a customary or de facto common name indicating a specific group or region of a building without the burden of manually assigning a name of a large-scale commercial facility or the like. .

[6. Fourth Embodiment]
The fourth embodiment is an example in which the configuration according to the representative notation extraction device (FIG. 1) in the first embodiment is applied to the extraction of facility name candidates in the map data processing device according to the third embodiment. That is, as shown in FIG. 13, the map data processing device 4 in the fourth embodiment has a configuration according to the third embodiment (FIG. 6), but the specific configuration of the candidate extracting means 44 is representative. The notation extraction device 41 is provided with means for realizing and executing the following operational effects. Note that the map data processing device 4 and the representative notation extraction device 41 may be realized as an integrated device, and details and meanings of each processing are the same as those described up to the third embodiment.

  Here, FIG. 14 is a flowchart showing a processing procedure in the fourth embodiment. That is, when a POI group corresponding to each selected facility polygon is selected as in the third embodiment (steps S32 and S33), the representative notation extracting device 41 corresponds to the facility polygon for each facility polygon. By analyzing the names of the POIs selected, facility name candidates are extracted as follows (step S44).

  That is, first, the dividing unit 14 divides each POI name related to the target facility polygon into tokens as a plurality of given character strings (step S14), and the counting unit 15 selects each character string. For each token sequence that consists of one or more tokens that are all or part of the string (limited to those that were also directly linked in the original string), The number of character strings is counted (step S15). Subsequently, the token selection means 16 excludes token strings that are included in other token strings in the token string, the token string having the counted number of character strings equal to or greater than a predetermined threshold value or greater than the threshold value. The token sequence (longest token sequence) is selected, and others are cut off (step S16).

  Further, the duplication deletion unit 17 deletes the selected token sequence that is duplicated among the longest token sequences among the longest token sequences remaining at this time (step S17). , Select the character string that contains the most currently selected token string from the given character string group as the reference character string, and select the currently selected token according to the order of each token in the reference character string. The columns are rearranged (step S18). The token sequence rearranged in this way is output to the association unit 35 and the map data update unit 46 by the result output unit 19 as an extraction result for each facility polygon (step S19).

  For each facility polygon, the above-described steps S14 to S19 are performed to extract facility name candidates. Finally, the association means 35 associates facility polygons with each other based on the facility name candidates (step S35). Further, the map data update means 46 updates the map data such as facility information (step S46), and this point is the same as in the third embodiment.

  However, in the fourth embodiment, the map data update means 46 outputs the token string output as the extraction result by the result output means 19 as described above based on the names of a plurality of POIs related to the associated facility polygons. Are stored in the map data storage means 21 as the names of the entire large-scale commercial facilities that are complex facilities corresponding to the facility polygons of the plurality of associated facilities, that is, the tower building, the east building, and the west building. The map data is updated (step S46).

  As described above, in the fourth embodiment, by selecting the longest token string included in a predetermined number of character strings from tokens obtained by dividing the name of a given character string group POI by character type or the like. Since the greatest common divisor notation used most commonly in the character string group is obtained, an appropriate representative notation such as a large-scale commercial facility name representing the character string group can be extracted.

  In the fourth embodiment, as described above, the token data extracted from the POI name related to the associated facility polygon is used as the facility name corresponding to the associated facility polygon group, and the map data is updated. By doing so (step S46), there is no burden of manually assigning names of large-scale commercial facilities, etc., and it is possible to appropriately set customary and de facto common names that refer to specific building groups and regions. It becomes.

  Note that the criteria and algorithm for extracting facility name candidates as criteria for grouping facility polygons and the criteria and algorithm for extracting facility name candidates to be given as names to grouped complex facilities may be different from each other. For example, the linguistic analysis as described in the third embodiment is used for grouping, and the name of the complex facility is separately changed from the POI in the group, as described in the first embodiment and the second embodiment. It can also be extracted with an algorithm using. In this way, it is possible to effectively use each algorithm according to characteristics according to the situation such as the stage of processing.

[7. Other embodiments]
In addition, said each embodiment is only an illustration, and this invention includes what is illustrated below and other embodiment other than that. For example, each element such as means is not limited to an arithmetic control unit of a computer, and may be realized by another information processing mechanism such as an electronic circuit based on wired logic or the like. Further, each configuration diagram, data diagram, flowchart diagram, and the like are merely examples, and the presence / absence of each element, its order, specific contents, and the like can be appropriately changed.

  Further, each of the devices (representative notation extraction device, map data processing device) of the present invention may be realized by using a plurality of devices such as servers, and each storage means is realized by a separate and independent server device or system. The configuration is also common. Depending on the function, the configuration can be flexibly changed, for example, by calling an external platform or the like with an API (application program interface) or network computing (so-called cloud or the like).

1, 2, 41 Representative notation extraction device 3, 4 Map data processing device 5 Arithmetic control unit 6 Storage device 7 Communication means 11 Character string storage means 12 Character string designation means 14 Division means 15 Count means 16 Token selection means 17 Duplicate deletion means 18 Sorting means 19 Result output means 21, 31 Map data storage means 22, 32 Polygon selecting means 23, 33 POI selecting means 24, 34, 44 Candidate extracting means 26, 36, 46 Map data updating means 35 Association means

Claims (6)

A dividing means for dividing a plurality of given character strings into tokens, and
Counting means for counting the number of character strings in which the token string is included in the plurality of character strings for all token strings that form one or more of the character strings by directly connecting one or more tokens;
Token selection means for selecting a token string in which the counted number of character strings is a token string that is greater than or equal to a predetermined threshold value and excluding a token string included in another longer token string. When,
A result output means for outputting the selected token string as an extraction result;
A representative notation extraction device characterized by comprising: A character string that includes the largest number of selected token strings among the given character strings is selected as a reference character string, and the selected token strings are arranged in accordance with the order of the tokens in the reference character string. The representative notation extraction device according to claim 1, further comprising rearranging means for changing. The representative notation extracting device according to claim 1, further comprising: a duplication deleting unit that deletes the tokens constituting the selected token string that overlap each other among the token strings. In the representative notation extraction device according to any one of claims 1 to 3,
Map data storage means for storing map data including facility polygons and POIs;
Polygon selection means for selecting facility polygons from the map data storage means;
POI selection means for selecting a POI associated with the selected facility polygon from the map data storage means;
Candidate extracting means for analyzing the name of the selected POI and extracting facility name candidates of each facility polygon;
A representative notation extraction device comprising:
The dividing means divides each POI name into a plurality of given character strings into tokens,
The counting means performs the counting based on each divided token, the token selecting means performs the selection of a token string,
The result output means outputs the selected token string as a facility name candidate extraction result,
The representative notation extracting device further includes map data updating means for storing the output facility name candidate in the map data storing means as the name of the selected facility polygon. A dividing step of dividing a plurality of given character strings into tokens;
A counting step of counting the number of character strings that include the token string in the plurality of character strings, for all token strings that form one or more of the character strings by directly connecting one or more tokens;
Token selection step of selecting a token string in which the number of counted character strings is a token string greater than or equal to a predetermined threshold value or larger and excluding a token string included in another longer token string. When,
A result output step for outputting the selected token string as an extraction result; and
A representative notation extraction method characterized in that a computer executes. A computer program for controlling a computer,
A dividing step of dividing a plurality of given character strings into tokens;
A counting step of counting the number of character strings that include the token string in the plurality of character strings, for all token strings that form one or more of the character strings by directly connecting one or more tokens;
A token selection step of selecting a token string in which the number of counted character strings is a token string greater than or equal to a predetermined threshold value and excluding a token string included in another longer token string in the token string When,
A result output step for outputting the selected token string as an extraction result; and
A representative notation extraction program for causing a computer to execute.

Images