JP2008015981A - Device and method for analyzing location, its program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To automatically select only one location and output it even if there exist two or more location candidates corresponding to each of a plurality of inputted address expressions. <P>SOLUTION: A location candidate extraction part 41 searches a database 1 registered with only address information, or the name of a facility that exists at a specified address and its address information for every record using an inputted address expression as a key, and extracts the address information of a record included with the address expression as a location candidate. The above-described processing is repeated by the number of address expressions. A combination generation part 42 generates all combinations between a location candidate of an address expression and one of another address expression from location candidates corresponding to each of the plurality of address expressions. A distance calculation part 7 calculates a distance between location candidates in all of the combinations, and then determines a location candidate which belongs to a combination that has a shorter distance between location candidates as the location. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a location which is information (for example, address information, longitude / latitude information, etc.) indicating a location corresponding to the address expression from an address expression consisting of part of the address information or the name of a facility existing at a predetermined address. Related to technology.

Conventionally, as a system for finding the location corresponding to this from the address expression, the address information input is a database in which only the address information or the name of the facility existing at the predetermined address and the address information is registered for each record. Is used as a key, and the address information of a record including the corresponding address expression is output as a location (see Patent Document 1).
JP 2001-134579 A

  By the way, there are not a few address expressions that have two or more corresponding locations. For example, there are locations in Tokyo and Osaka that correspond to address expressions consisting of part of the address information “Nihonbashi”, and locations that correspond to address expressions consisting of part of the address information “Suehirocho” are all over Japan. There are also 126 locations.

  In this case, the above-described conventional system only displays a list of all the corresponding locations as candidates, and after that, the user must make a selection based on his / her own judgment. There has been a problem that it is difficult or time consuming to determine the location, that is, the target location.

  An object of the present invention is to make it possible to automatically select and output only one of them, even when there are two or more locations corresponding to a plurality of input address expressions. .

  In the present invention, in order to solve the above-mentioned problem, each of a plurality of input address expressions is referred to by referring to a database in which the address name only or the name of a facility existing at a predetermined address and the address information are registered for each record. Output address information of all records including the address expression as address candidates, and output all combinations of the address candidates of one address expression and the address candidates of other address expressions. The distance between the address candidates in all combinations of the address candidates of the address expression and the address candidates of the other address expressions is obtained and output, and the output among the address candidates corresponding to each of the plurality of output address expressions is output. A location candidate belonging to a combination with a smaller distance between the location candidates is determined as the location.

  According to the present invention, by determining each candidate having a smaller distance among the candidate locations corresponding to each of the plurality of input address expressions, the locations corresponding to the plurality of input address expressions are determined. Even when there are two or more of them, only one of them can be automatically and appropriately selected and output.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

<Embodiment 1>
FIG. 1 shows an example of an embodiment of a location analysis apparatus according to the present invention. In the figure, 1 is a database, 2 is an address expression extraction unit, 3 is an address expression storage unit, 4 is a location candidate generation unit, A location candidate storage unit, 6 is a combination storage unit, 7 is a distance calculation unit, 8 is a location determination unit, and 9 is an address expression-location association unit.

  The database 1 is at least one database in which address information and its latitude / longitude information or the name of a facility existing at a predetermined address and its address information and latitude / longitude information are registered for each record. It consists of a store database 12, a station name database 13, and a landmark database 14.

  As shown in FIG. 2, the address database 11 stores address information and its latitude / longitude information, here, address information of all of Japan and its latitude / longitude information for each record. The part where the address information is registered is called an address field, and the part where the latitude / longitude information is registered is called a latitude / longitude field. Here, the address database 11 includes addresses including addresses and issues existing all over Japan as records (for example, “Nihonbashi 1-chome xx-1 No. 1”, “Nihonbashi 1-chome xx-2 Chuo-ku, Tokyo”). ”, Etc.) along with the latitude / longitude information, or addresses up to the middle of the street, such as“ Chuo-ku, Tokyo ”or“ Chuo-ku, Tokyo ” “Nihonbashi 1-chome” and “Chuo-ku, Tokyo”) may also be registered along with the latitude and longitude information.

  As shown in FIG. 3, the store database 12 is a facility in which a facility existing at a predetermined address, here a store name (store name), its address information, and latitude / longitude information is registered for each record. In the record, the part where the store name is registered is called the store name field, the part where the address information is registered is called the address field, and the part where the latitude / longitude information is registered is called the latitude / longitude field.

  As shown in FIG. 4, the station name database 13 is a facility in which a facility existing at a predetermined address, in this case, a station name (station name), its address information, and latitude / longitude information is registered for each record. Of these, a part where a station name is registered is called a station name field, a part where address information is registered is called an address field, and a part where latitude / longitude information is registered is called a latitude / longitude field.

  As shown in FIG. 5, the landmark database 14 is a facility in which a facility existing at a predetermined address, in this case, a landmark name (landmark name), its address information, and latitude / longitude information is registered for each record. In each record, the part where the landmark name is registered is the landmark name field, the part where the address information is registered is the address field, and the part where the latitude / longitude information is registered is the latitude / longitude field. Call.

  The address expression extraction unit 2 reads address information (text file) read from a storage medium (not shown) or input from another device or the like via a communication medium (address expression included in the document information (text file)). Usually, a plurality of address expressions) are extracted using a well-known unique expression extraction technique and output to the address expression storage unit 3 and the location candidate generation unit 4.

  Note that the operator may directly input a plurality of address expressions included in the document information from a keyboard or the like (not shown). In this case, the address expression extraction unit 2 is not necessary.

  The address expression storage unit 3 temporarily stores a plurality of address expressions output from the address expression extraction unit 2.

  The location candidate generation unit 4 refers to the database 1 and outputs the address information of all records including the address representation for each of the plurality of address representations output from the address representation extraction unit 2 as location candidates. This outputs all combinations of address candidates for address expressions and address candidates for other address expressions. More specifically, the address candidate extractor 41 and a combination generator 42 are included.

  The address candidate extraction unit 41 searches the database 1 using the input address expression as a key, extracts address information and latitude / longitude information of all records including the corresponding address expression as address candidates, and uses this as the address expression. In more detail, the database selection unit 411 and the location candidate acquisition unit 412 are configured.

  The database selection unit 411 stores and holds in advance a database selection table describing character string rules and corresponding database names and field names as shown in FIG. The addressed character string is compared with each character string rule described in the table, and if there is a matching character string rule, the database name and field name corresponding to the matching character string rule are read from the table. The selected database name and field name are output to the location candidate acquisition unit 412 and this is repeated for the number of address expressions.

  Here, the character string rule is to determine that the input address expression is “store name” when the character “store” exists at the end of the input address expression character string. This is for determining that the input address expression is “station name” when the character “station” exists at the end of the input address expression character string. Accordingly, by referring to the database selection table describing the character string rule and the corresponding database name and field name as shown in FIG. 6, the database name and field name corresponding to the input address expression are obtained. It is possible to select and output.

  The database selection unit 411 is used to reduce the number of databases and fields to be searched when extracting location candidates. This database selection unit 411 is not an indispensable component, and requires a large amount of calculation. You may make it search with respect to all the databases and all the fields by using the input address expression as a key.

  The address candidate acquisition unit 412 extracts a plurality of address expressions extracted and temporarily stored in the address expression storage unit 3 as described above, and database names and field names corresponding to each address expression selected and output by the database selection unit 411. Is input, and the field expression is searched for the database field corresponding to the selected database name and field name in the database 1 as the key, and the address information and latitude / longitude information of the record including the corresponding address expression are located. This is acquired as a candidate, and this is repeated as many times as the number of address expressions, and is output to the combination generation unit 42 and the location candidate storage unit 5.

  The combination generation unit 42 receives as input the location candidates corresponding to each of the plurality of address expressions extracted by the location candidate extraction unit 41, and all combinations of the location candidates of one address expression and the location candidates of other address expressions. Generated and output to the combination storage unit 6 and the distance calculation unit 7.

  The location candidate storage unit 5 temporarily stores location candidates corresponding to each of a plurality of address expressions output from the location candidate generation unit 4 (location candidate acquisition unit 412 of the location candidate extraction unit 41). Further, the combination storage unit 6 temporarily stores all combinations of one address expression location candidate and another address expression location candidate output from the location candidate generation unit 4 (the combination generation unit 42). .

The distance calculation unit 7 obtains and outputs the distance between the location candidates in all combinations of the location candidate of one address expression output from the location candidate generation unit 4 and the location candidate of another address expression. Is a location candidate corresponding to each of the plurality of address expressions output from the location candidate generation unit 4 (location candidate extraction unit 41) and stored in the location candidate storage unit 5, and a combination of the location candidate generation unit 4 (the combination generation unit 42) ) And the location candidates of one address expression and all combinations of the address candidates of other address expressions are input, and the distance D between the location candidates in one combination is the latitude information N of each location candidate. ₁ and N ₂ and longitude information E ₁ and E _2, that is, D = {(N ₁ −N ₂ ) ² + (E ₁ −E ₂ ) ² } ^1/2
This is calculated in the same manner for each combination and output to the location determination unit 8.

  As will be described later, the location determination unit 8 only needs to be able to determine the magnitude of the distance (value) between location candidates in all combinations of location candidates of one address expression and location candidates of other address expressions. That is, since an absolute distance value is not required, calculation of the route (1/2 power) portion in the above formula may be omitted. As for this formula, a known formula for calculating the distance between two points using latitude information and longitude information, for example, “formula for calculating geodesic line length by converting latitude and longitude into plane rectangular coordinates” (BRBowring , "TOTAL INVERSE SOLUTIONS FOR THE GEODESIC AND GREAT ELLIPTIC," Survey Review, Vol.33, No.261, July 1996, pp.461-476 (http://vldb.gsi.go.jp/sokuchi/surveycalc/algorithm /) See) etc. may be used.

  The location determination unit 8 selects, from among the location candidates corresponding to each of the plurality of address expressions output from the location candidate generation unit 4, location candidates belonging to a combination with a smaller distance between the location candidates output from the distance calculation unit 7. More specifically, each location is determined, and more specifically, location candidates corresponding to each of a plurality of address expressions output from the location candidate generation unit 4 (location candidate extraction unit 41) and stored in the location candidate storage unit 5, and location candidates All combinations of one address expression location candidate and another address expression location candidate output from the generation unit 4 (the combination generation unit 42) and stored in the combination storage unit 6 are calculated by the distance calculation unit 7. The distance between location candidates in each combination is input, and among the location candidates corresponding to each of a plurality of address expressions, physical ( Respectively determine the location candidates physical) distance belongs to a smaller combination on the location (address information and the latitude and longitude information), address representation - output on the location association unit 9.

  The address expression-location associating unit 9 outputs the plurality of address expressions extracted as described above in association with the addresses corresponding to each address expression determined by the location determining unit 8, and more specifically, as described above. The plurality of address expressions extracted as described above and temporarily stored in the address expression storage unit 3 and the location corresponding to each address expression output from the location determination unit 8 are input, and the plurality of address expressions and their addresses (address information and Latitude / longitude information) are output in association with each other.

  In addition, when it is not necessary to output the address expression and the address in association with each other, and it is sufficient to output only the address corresponding to the address expression, the address determined by the address determining unit 8 is directly selected from the address determining unit 8. What is necessary is just to make it output to the outside, and the address expression-address matching part 9 is unnecessary in this case.

  FIG. 7 shows the flow of processing in the location analysis apparatus of the present invention. Hereinafter, the operation will be described in detail with an example.

  The address expression extraction unit 2 extracts and outputs (s1) a plurality of address expressions from the document information input as described above, and is stored in the address expression storage unit 3 and input to the database selection unit 411. . The character string of the address expression input to the database selection unit 411 is compared with the character string rule in the database selection table stored and held in advance in the database selection unit 411, and corresponds to the matched character string rule. The database name and field name are selected, and the database name and field name are output to the location candidate acquisition unit 412. This process is repeated for the number of address expressions (s2).

  For example, if the input address expression is “Nihonbashi”, “address database, address field” is output as shown in FIG. If the input address expression is “Tokyo Tower”, “Landmark database, landmark name field” is output as shown in FIG. When the input address expression is “NTT ... store”, “store database, store name field” is output as shown in FIG. If the input address expression is “Tokyo station”, “station name database, station name field” is output as shown in FIG.

  Next, in the address candidate acquisition unit 412, the address expression stored in the address expression storage unit 3 is read and input, and the database name and field name selected and output by the database selection unit 411 are input. Then, the database field corresponding to the selected database name and field name is searched from the database 1 using the address expression as a key, and the address information and latitude / longitude information of the record including the corresponding address expression are the location candidates. And is output to the combination generation unit 42 and output to and stored in the location candidate storage unit 5, and this process is repeated for the number of address expressions (s3).

  For example, when the address expression is “Nihonbashi”, the selected database name is “address DB (database)”, and the selected field name is “address field”, as shown in FIG. "35.677, 139.776" and "Nihonbashi, Naniwa-ku, Osaka, 34,656, 135.509" are acquired and output as location candidates. Further, when the address expression is “A”, the selected database name is “... DB”, and the selected field name is “... Field”, “A1, 34.123, 138.654” and “A2” as shown in FIG. , 38.324, 140.002 "are acquired and output as location candidates, and the address expression is" B ", the selected database name is" ... DB ", and the selected field name is" ... field " As shown in FIG. 9, “B1,32.832,134.843” and “B2,34.534,135.809” are acquired and output as location candidates (“A”, “B”, “A1”, “A2”) , “B1” and “B2” are expressions in which detailed contents are omitted, and are actually character strings similar to “Nihonbashi” and “Nihonbashi, Chuo-ku, Tokyo”.

  The above processing is “location candidate extraction processing”.

  Next, in the combination generation unit 42, when address candidates (address information and latitude / longitude information) corresponding to each of the plurality of address expressions extracted and output by the address candidate extraction unit 41 are input, one address expression All combinations of address candidates (address information) and address candidates (address information) of other address expressions are generated, output to the combination storage unit 6, stored, and output to the distance calculation unit 7. (S4).

  For example, location candidates “Nipponbashi, Chuo-ku, Tokyo, 35.677, 139.776” and “Nihonbashi, Naniwa-ku, Osaka, 34.656, 135.509”, address candidates “A1, 34.123, 138.654”, address representation “A”. ”And“ A2, 38.324, 140.002 ”, and address candidates“ B1,32.832,134.843 ”and“ B2,34.534,135.809 ”of the address expression“ B ”are input as shown in FIG. Twelve combinations of “A1,” “Tokyo,..., A2,” “Tokyo,..., B1,” “A2, B2” are output.

  The above processing is “location candidate generation processing”.

  Next, in the distance calculation unit 7, location candidates (address information and latitude / longitude information) corresponding to each of the plurality of address expressions stored in the location candidate storage unit 5 are read out and input, and a combination is generated. When all combinations of the address candidate (address information) of one address expression generated and output by the unit 42 and the address candidate (address information) of another address expression are input, the address candidates in the combination The distance between them is calculated as described above, these are performed in the same manner for all combinations, and are output to the location determination unit 8 (s5).

For example, for the combination “Tokyo ..., A1”, the latitude / longitude information of the location candidate including the address information “Tokyo ...” is “35.677, 139.776”, and the latitude / longitude of the location candidate including the address information “A1” Since the longitude information is “34.123, 138.654”
D = {(35.677−34.123) ² + (139.776−138.654) ² } ^1/2
= 1.916715941
Is calculated. The other combinations are similarly calculated, and the distance between the location candidates corresponding to each combination is output, for example, as shown in FIG.

  Next, in the location determination unit 8, location candidates corresponding to each of the plurality of address representations stored in the location candidate storage unit 5 are read out and input, and the location of one address representation stored in the combination storage unit 6 When all combinations of candidates and location candidates of other address expressions are read out and input, and distances between location candidates in each combination calculated and output by the distance calculation unit 7 are input, Among the address candidates corresponding to each address expression, each address candidate belonging to a combination with a smaller distance between the address candidates is determined as the address (address information and latitude / longitude information), and the address expression-location associating unit 9 (S6).

  FIG. 12 shows details of the processing flow in the location determination unit 8, and FIG. 13 shows an example of processing contents of the location determination unit 8. The operation will be described in detail below according to these.

  First, location candidates of multiple address expressions, all combinations of location candidates, and distances corresponding to each combination, for example, six location candidates “Nihonbashi Chuo-ku, Tokyo, 35.677, 139.776”, “Nihonbashi, Naniwa-ku, Osaka, Osaka, Japan, 34.656, 135.509 "," A1, 34.123, 138.654 ", ..." B2, 34.534, 135.809 "and 12 combinations" Tokyo ..., A1 "," Tokyo ..., A2 "," Tokyo ..., B1 " , ... “A2, B2” and the distances “1.916715941”, “2.656630385”, “5.694603937”,... “5.652021674” of each combination are input (s11), and the minimum distance among the input distances of each combination , Here, “0.323857994” is extracted (s12).

  Next, a combination of address candidates corresponding to the minimum distance from among the input address candidates of a plurality of address expressions, which is the same as any one of the address candidates included in “Osaka Prefecture…, B2” in this case. Address candidates including address candidates, here address candidates “Nihonbashi” address candidates including address candidate “Osaka Prefecture…” “Nihonbashi, Chuo-ku, Tokyo, 35.677, 139.776” and “Nihonbashi, Naniwa-ku, Osaka, 34,656, Address candidates “B1,32.832,134.843” and “B2,34.534,135.809” of the address expression “B” including the address candidate “B2” are searched (S13).

  Next, among the address candidates of the searched address expression, the address candidates not included in the combination of address candidates corresponding to the minimum distance, here, the address candidate of the address expression “Nihonbashi” “Nihonbashi, Chuo-ku, Tokyo, 35.677, 139.776 ”and the address candidate“ B1,32.832,134.843 ”of the address expression“ B ”are deleted (s14), and all the combinations of the address candidates and the distances corresponding to each combination are deleted. A combination of address candidates including the address candidates (address information thereof) and distances corresponding thereto, here “Tokyo…, A1” and “1.916715941”, “Tokyo ..., A2” and “2.656630385”,… “A2 , B1 ”and“ 7.535074319 ”are deleted (s15). In FIG. 13, deletion is expressed by double strikethrough.

  Next, it is determined whether or not there is one address candidate for all address expressions (s16). Here, there are two address candidates for address expression "A", that is, "A1, 34.123, 138.654" and "A2," 38.324, 140.002 ”, the process returns to the process of s12, and the minimum distance from the distances of the combinations that remain without being deleted (however, the distances extracted so far, here,“ 0.323857994 ”is excluded)). That is, “2.874534049” is extracted.

  Then, by the process of s13, as described above, from among the plurality of input address candidates of the address expression, a combination of address candidates corresponding to the minimum distance, here, any address included in “A1, B2” Address candidates of address expressions including the same as the candidates, here are address candidates “A1, 34.123, 138.654” and “A2, 38.324, 140.002” of address expressions “A” including the address candidate “A1”, and s14 The address candidates that are not included in the combination of address candidates corresponding to the minimum distance among the address candidates of the address expression searched by the processing of the address expression “A” in this case, “A2, 38.324, 140.002” ”Is deleted, and a combination of location candidates including the deleted location candidate (address information thereof) and the corresponding distance, here“ Osaka , A2 ”and“ 5.80010974 ”and“ A2, B2 ”and“ 5.652021674 ”are deleted (actually, this is also performed for those including the location candidate“ B2 ”, but it has already been done. So the result will not change.)

  As a result, it is determined in the process of s16 that there is one address candidate for all address expressions, so that the address candidates for each address expression remaining without being deleted, “Nanbashi Nihonbashi, 34.656, 135.509”, “A1, 34.123, 138.654” and “B2, 34.534, 135.809” are output (s17).

  Finally, in the address expression-location association unit 9, a plurality of address expressions stored in the address expression storage unit 3 are read out and input, and each address expression determined and output by the location determination unit 8 is output. When a corresponding location is input, this address expression and location (address information) are output in association with each other (s7).

  For example, a plurality of address expressions are “Nihonbashi”, “A”, and “B”, and addresses corresponding to each address expression are “Nipponbashi, Naniwa-ku, Osaka, 34.656, 135.509”, “A1, 34.123, 138.654”, and “B2”. , 34.534, 135.809 ", as shown in FIG. 14," Nihonbashi: Nihonbashi, Naniwa-ku, Osaka, Osaka, 34.656, 135.509 "," A: A1, 34.123, 138.654 "and" B: B2, 34.534, 135.809 " Is output.

  As described above, according to this embodiment, even when there are two or more locations corresponding to a plurality of input address expressions, only one of them is automatically selected and output. It becomes possible to do.

  If a plurality of address expressions are input directly from a keyboard (not shown) by an operator or obtained from a database or a car navigation system, as described above, address expression extraction is performed. If the processing (s1) is not required, and it is not necessary to output the address expression and the location in association with each other, and only the location corresponding to the input address expression needs to be output, the location determination processing (s6) The determined location may be directly output to the outside. In this case, the address expression-location associating process (s7) is not necessary.

<Embodiment 2>
In the second embodiment of the present invention, as shown in FIG. 15, the address information divided for each address level in prefectures, municipalities, towns and villages, town large letters, and letters / chomes is displayed for each record together with the latitude / longitude information. Use the address database registered in. That is, for example, if the address information of one record in the address database described in the first embodiment is “Nihonbashi, Chuo-ku, Tokyo ...”, the address information of the same record in the present embodiment is “Tokyo / city, "Chuo / Ku, Nihonbashi, ..." (However, "," is a delimiter symbol, and "/" is the character string up to the next delimiter symbol "," may or may not be in the address expression. As shown in FIG. 5), the registered address is divided into each prefecture, municipality, town, village, town large character, letter / chome address level.

  Then, the address candidate generation unit 4 (specifically, the address candidate acquisition unit 412 of the address candidate extraction unit 41) uses the input address expression, here, the address expression consisting of a part of the address information as a key in the address field of the address database. , The address information of the record including the corresponding address expression is acquired, and the address information of the record satisfying the coincidence with the address expression is output together with the latitude / longitude information as a location candidate.

  According to the present embodiment, for example, an address expression is configured such that a record including address information “Highland of England ...” is extracted from the input address expression “Iran” and is generated as a location candidate. It is less likely that incorrect location candidates are generated due to simple matching of character strings.

<Embodiment 3>
In the third embodiment of the present invention, distances between location candidates in all combinations of location candidates in the distance calculation unit 7 are calculated not from the latitude / longitude information but from the matching rate of the character strings constituting them. That is, all combinations of location candidates of one address expression and location candidates of other address expressions output from the location candidate generation unit 4 (the combination generation unit 42) are input, and between the location candidates in one combination The distance D is the number of characters length (add1) and length (add2) of the address information of each location candidate and the number of characters that match the address information of each location candidate. Calculate using length (add1, add2), ie D = {2 × match length (add1, add2)} / {length (add1) + length (add2)}
This is calculated in the same manner for each combination and output to the location determination unit 8.

  Then, the location determination unit 8 has a smaller distance between the location candidates output from the distance calculation unit 7 among the location candidates corresponding to each of the plurality of address expressions output from the location candidate generation unit 4. Location candidates belonging to a combination having a larger distance represented by the column matching rate are determined as locations, and are output to the address expression-location association unit 9.

  According to the present embodiment, the distance between the address candidates can be obtained from the character string of the address information of the address candidates, and the latitude / longitude information in each database can be made unnecessary.

<Embodiment 4>
In the fourth embodiment of the present invention, the location candidates corresponding to each of the plurality of address expressions in the location determination unit 8 are determined from the sum of the distances between the location candidates corresponding to all combinations of the address candidates of the plurality of address expressions. More specifically, the location candidate corresponding to each of the plurality of address expressions output from the location candidate generation unit 4 (location candidate extraction unit 41) and stored in the location candidate storage unit 5, and the location candidate generation unit 4 ( In the combination calculation unit 42), all combinations of one address expression location candidate and another address expression location candidate stored in the combination storage unit 6 and each combination calculated by the distance calculation unit 7 Using distances between location candidates as input, generate all combinations of location candidates for each address expression from the address candidates corresponding to each of multiple address expressions. , Out of the combinations of the address candidates for each address expression, output from the distance calculation unit 7 corresponding to each of the combinations of the address candidates for one address expression and the address candidates for other address expressions included in the combination. The location candidates belonging to the combination having the smallest sum of the distances between the location candidates are determined as locations corresponding to the plurality of address representations, and are output to the address representation-location association unit 9.

  FIG. 16 shows an example of processing contents in the location determination unit 8 of the present embodiment, and the operation will be described in detail below according to this.

  First, one address candidate is selected for each address expression from a plurality of input address candidates, and a set of all possible address candidates (address information thereof) is generated. For example, in FIG. 16, there are six location candidates “Nihonbashi, Chuo-ku, Tokyo, 35.677, 139.776”, “Nihonbashi, Naniwa-ku, Osaka, Osaka, 34.656, 135.509”, “A1, 34.123, 138.654”, “B2, 34.534, From “135.809”, 8 combinations of “Tokyo, A1, B1,” “Tokyo, A1, B2,” “Tokyo, A2, B1,” “Osaka, A2, B2” are generated. Is done.

  Next, in each set of all possible location candidates generated, a combination of a location candidate of one address expression and a location candidate of another address expression, which is a partial configuration, is combined with the location candidate generation unit 4 ( The combination generation unit 42) selects from all combinations of the address candidates of one address expression and the address candidates of other address expressions. Then, the sum of the distances between the location candidates of the selected combination is set as the distance of each set of all possible location candidates. For example, in FIG. 16, a combination of “Tokyo ..., A1”, “Tokyo ..., B1” and “A1, B1” is a partial configuration with respect to the set of “Tokyo ..., A1, B1”. The selected distance of “Tokyo ... A1, B1” is “11.63504974” which is the sum of the distances of “Tokyo ..., A1”, “Tokyo ..., B1” and “A1, B1”.

  Then, the minimum distance is extracted from the distances of all the possible combinations of possible location candidates, and the location candidates belonging to the set corresponding to the minimum distance are the locations corresponding to the plurality of address expressions. Determined and output. In the example of FIG. 16, the distance “6.3882375” is extracted, and the location candidates “Nipponbashi, Naniwa-ku, Osaka, 34.656, 135.509”, “Osaka Prefecture…, A1, B2”, which is a set corresponding to this distance, “A1, 34.123, 138.654” and “B2, 34.534, 135.809” are output as locations.

  The present invention can also be realized by installing a program having the procedures shown in the flowcharts of FIGS. 7 and 12 via a medium or a communication line in a known computer having or connected to the database described above. It is.

The block diagram which shows an example of embodiment of the location analyzer of this invention Explanatory diagram showing an example of the address database Explanatory drawing which shows an example of a store database Explanatory diagram showing an example of the station name database An explanatory diagram showing an example of a landmark database Explanatory drawing which shows an example of a database selection table Flow chart of processing in the location analysis apparatus of the present invention Explanatory drawing which shows an example of the input / output value of a database selection part Explanatory drawing which shows an example of the input / output value of a location candidate acquisition part Explanatory drawing which shows an example of the input / output value of a combination production | generation part Explanatory drawing which shows an example of the input / output value of a distance calculation part Flow chart of processing in the location determination unit Explanatory drawing which shows an example of the processing content of a location determination part Explanatory drawing which shows an example of the input-output value of an address expression-address matching part Explanatory drawing showing another example of address database Explanatory drawing which shows the other example of the processing content of a location determination part.

Explanation of symbols

  1: database, 2: address expression extraction unit, 3: address expression storage unit, 4: location candidate generation unit, 5: location candidate storage unit, 6: combination storage unit, 7: distance calculation unit, 8: location determination unit, 9: Address expression-location association unit, 41: Location candidate extraction unit, 42: Combination generation unit, 411: Database selection unit, 412: Location candidate acquisition unit, s1: Address expression extraction process, s2: Database selection process, s3 : Location candidate acquisition processing, s4: combination generation processing, s5: distance calculation processing, s6: location determination processing, s7: address expression-location association processing.

Claims (18)

In an apparatus for obtaining a location which is information representing a place corresponding to the address expression from an address expression consisting of a part of the address information or the name of a facility existing at a predetermined address,
At least one database in which only the address information or the name of the facility existing at the predetermined address and the address information are registered for each record;
With reference to the database, for each of a plurality of input address expressions, the address information of all records including the address expression is output as the address candidates, and the address candidates of one address expression and the address candidates of other address expressions A candidate location generator that outputs all combinations of
A distance calculation unit that obtains and outputs the distance between the location candidates in all combinations of the location candidate of one address expression output from the location candidate generation unit and the location candidate of another address expression;
A location determination unit that determines a location candidate that belongs to a combination with a smaller distance between the location candidates output from the distance calculation unit among the location candidates corresponding to each of the plurality of address expressions output from the location candidate generation unit; A location analysis device characterized by comprising: The location candidate generator
Searching the database using the input address expression as a key, extracting address information of all records including the corresponding address expression as a location candidate, and repeating this as many times as the number of address expressions,
From a combination generation unit that generates and outputs all combinations of a location candidate of one address expression and a location candidate of another address expression from the address candidates corresponding to each of the plurality of address expressions extracted by the address candidate extraction unit The location analysis apparatus according to claim 1, wherein: Using at least one database in which the address information and its latitude / longitude information or the name of the facility existing at a given address and its address information and latitude / longitude information are registered for each record,
The address candidate generation unit outputs the address candidate including the latitude / longitude information together with the address information,
The location analysis unit according to claim 1, wherein the distance calculation unit calculates distances between location candidates in all combinations of location candidates output from the location candidate generation unit using the latitude / longitude information thereof. apparatus. The distance calculation part which calculates the distance between the location candidates in all the combinations of the location candidates output from the location candidate generation part from the matching rate of the character string which comprises them is provided. Location analysis device. The location determination department
The minimum distance is extracted from the distances between the address candidates of all combinations of the address candidates output from the distance calculation unit, and the address candidates corresponding to each of the plurality of address expressions output from the address candidate generation unit are extracted. The address candidate of the address expression including the same one of the address candidates included in the combination of address candidates corresponding to the minimum distance is searched, and the minimum distance is selected from the address candidates of the searched address expression. The location candidates that are not included in the combination of location candidates corresponding to, are deleted, and the deletion is performed from all the combinations of location candidates output from the location candidate generation unit and the distance calculation unit and the distances corresponding to each combination. Address combinations corresponding to all address expressions by deleting the combination of address candidates including the address candidates and the distance corresponding to the combination. Complement the location analyzer according to claim 1, wherein the determining the location corresponding to a plurality of address representations is repeated until one. All combinations of address candidates of each address expression are generated from the address candidates corresponding to each of the plurality of address expressions output from the address candidate generation unit, and among the combinations of the address candidates of each address expression, Multiple address candidates belonging to the combination with the smallest sum of the distances between the address candidates output from the distance calculation unit corresponding to each of the combinations of the address candidates of one address expression and the address candidates of other address expressions included The location analysis device according to claim 1, further comprising a location determination unit that determines a location corresponding to each of the address expressions. Use at least a database that registers address information separated for each address level in prefectures, cities, towns, towns, towns, large letters, and letters / chomes,
The address candidate generation unit outputs address information of a record satisfying a delimiter coincidence with the address expression among records including an address expression including a part of the input address information, as the address candidate. The location analysis apparatus according to any one of 1 to 6. The location analysis apparatus according to claim 1, wherein the plurality of address expressions are extracted from one sentence or one document. In a method for obtaining a location which is information representing a place corresponding to the address expression from an address expression consisting of part of the address information or the name of a facility existing at a predetermined address,
Using a computer with at least one database in which the address information only or the name of the facility existing at a given address and the address information are registered for each record,
The computer
With reference to the database, for each of a plurality of input address expressions, the address information of all records including the address expression is output as the address candidates, and the address candidates of one address expression and the address candidates of other address expressions Location candidate generation processing step for outputting all combinations with
A distance calculation processing step for obtaining and outputting the distance between the location candidates in all combinations of the location candidates of one address expression and the location candidates of other address expressions output in the location candidate generation processing step;
Location determination for determining a location candidate belonging to a combination having a smaller distance between the location candidates output in the distance calculation processing step from among the location candidates corresponding to each of the plurality of address expressions output in the location candidate generation processing step. And a processing step. The location candidate generation processing step
Searching the database using the input address expression as a key, extracting address information of all records including the corresponding address expression as address candidates, and repeating this as many times as the number of address expressions,
A combination generation processing step for generating and outputting all combinations of a location candidate of one address expression and a location candidate of another address expression from the address candidates corresponding to each of the plurality of address expressions extracted in the location candidate extraction processing step. The location analysis method according to claim 9, comprising: Using at least one database in which the address information and its latitude / longitude information or the name of the facility existing at a given address and its address information and latitude / longitude information are registered for each record,
The location candidate generation processing step outputs the location candidate including the latitude / longitude information together with the address information,
The distance calculation processing step calculates distances between location candidates in all combinations of location candidates output in the location candidate generation processing step using the latitude / longitude information thereof. Location analysis method. The distance calculation processing step of calculating distances between location candidates in all combinations of location candidates output in the location candidate generation processing step from a matching rate of character strings constituting them. Location analysis method for. The location determination process step
Extracting a minimum distance from the distances between the location candidates of all combinations of location candidates output in the distance calculation processing step;
Of the address candidates corresponding to each of the plurality of address expressions output in the address candidate generation processing step, an address expression including the same address candidate as one of the address candidates included in the combination of address candidates corresponding to the minimum distance Searching for potential locations,
Deleting a candidate address not included in the candidate address combination corresponding to the smallest distance from the candidate address expressions of the searched address expression;
Among all combinations of location candidates output in the location candidate generation processing step and the distance calculation processing step and distances corresponding to each combination, combinations of location candidates including the deleted location candidates and distances corresponding thereto are obtained. A step to delete,
10. The location analysis method according to claim 9, further comprising the step of repeating the same process until the number of location candidates corresponding to all address expressions becomes one, and determining the locations corresponding to each of the plurality of address expressions. All combinations of address candidates for each address expression are generated from the address candidates corresponding to each of the plurality of address expressions output in the address candidate generation processing step, and the combination among the combinations of the address candidates for each address expression The address candidates belonging to the combination with the smallest sum of the distances between the address candidates output in the distance calculation processing step corresponding to each of the address candidates of one address expression and the address candidates of other address expressions included in The location analysis method according to claim 9, further comprising a location determination processing step for determining a location corresponding to each of a plurality of address expressions. Use at least a database that registers address information separated for each address level in prefectures, cities, towns, towns, towns, large letters, and letters / chomes,
The address candidate generation processing step outputs address information of a record satisfying a delimiter coincidence with the address expression out of records including an address expression made up of a part of the input address information as the address candidate. Item 15. The location analysis method according to any one of Items 9 to 14. The location analysis method according to claim 9, wherein the plurality of address expressions are extracted from one sentence or one document.   A location analysis program for causing a computer to execute each processing step of the location analysis method according to claim 9.   18. A computer-readable recording medium on which the location analysis program according to claim 17 is recorded.

Images