Location expression detection device and computer readable medium

• US 20090112537A1
• Filed: 09/09/2008
• Published: 04/30/2009
• Est. Priority Date: 10/29/2007
• Status: Abandoned Application

• Current Assignee: OKI Electric Industry Company Limited
• Original Assignee: OKI Electric Industry Company Limited
• Inventors: Okumura, Koji
• Application Number:
  US12/230,977
• Publication Number:
  US 20090112537A1
• Time in Patent Office: Days
• US Class Current: 703/3
• CPC Class Codes: G06F 16/30 of unstructured textual data document management systems G06F16/93
• CPC Class Codes: G06F 40/295 Named entity recognition

Abstract

A location expression detection device extracts a location expression contained in an object text and determines one of the candidate places when the extracted location expression has a plurality of candidate places. A location expression information storage section stores correspondence of location expressions against locations. A location expression extraction section extracts a location expression from an object text and determines whether the extracted location expression has plural candidate places with reference to stored contents. An object text characteristic word extraction section extracts a characteristic word in the object text. A compared text search section searches for a compared text by using at least one of the plural candidate places and/or the characteristic word. A location expression identification section determines one of the candidate places based on information related to the searched compared text.

Claims

1. A location expression detection device for extracting a location expression contained in an object text, and when the extracted location expression has a plurality of candidate places, determining one of the candidate places, the location expression detection device comprising:

a location expression information storage section that stores correspondence of location expressions that can exist in various texts against locations that the location expressions can specify;

a location expression extraction section that extracts a location expression from an object text and determines whether the extracted location expression has a plurality of candidate places by referring to stored contents of the location expression information storage section;

an object text characteristic word extraction section that extracts a characteristic word in the object text;

a compared text search section that searches for a compared text which acts as a compared object to the object text by using at least one of the plurality of candidate places and/or the characteristic word; and

a location expression identification section that, when the extracted location expression has the plurality of candidate places, determines one of the candidate places based on information related to the searched compared text,

wherein the object text characteristic word extraction section, the compared text search section, and the location expression identification section function at least when the extracted location expression has the plurality of candidate places.

2. The location expression detection device according to claim 1, wherein the compared text search section further searches for a compared text which includes an unambiguous location expression relating to each of the plurality of candidate places, and

the location expression identification section further comprises:

a compared text characteristic word extraction section that extracts a characteristic word in the searched compared text; and

a characteristic word matching degree determination and output section that identifies a location of the location expression included in the object text based on an unambiguous location notation in the compared text with the highest degree of matching of the extracted characteristic word to the characteristic word in the object text.

3. The location expression detection device according to claim 1, wherein the compared text search section further acquires, for each of the plurality of candidate places,

a first number of cases, which is the number of cases of compared text which includes an unambiguous location notation relating to each of the candidate places, and

a second number of cases, which is the number of cases of compared text which includes both the unambiguous location notation relating to each of the candidate places and the characteristic word of the object text;

and

the location expression identification section further comprises;

a characteristic word content rate computation section that obtains, for each of the plural candidate places, a characteristic word content rate which reflects a rate of compared texts among a plurality of compared texts which include the characteristic word, based on the first number of cases and on the second number of cases; and

a characteristic word content rate determination and output section that identifies the candidate place with the highest characteristic word content rate as the location of the location expression included in the object text.

4. The location expression detection device according to claim 1, wherein:

the location expression information storage section further stores coordinate information of the location which the location expression can specify;

the location expression extraction section further acquires the coordinate information of the plurality of candidate places;

the compared text search section further acquires a compared text which includes the characteristic word in the object text; and

the location expression identification section further comprises;

a location expression within compared text extraction section that obtains an unambiguous location expression in the compared text and coordinate information of the unambiguous location expression by referring to the stored contents in the location expression information storage section;

a distance computation section that obtains a distance between coordinate information of each of the plurality of candidate places and coordinate information relating to the location information of the compared text; and

a distance determination output section that identifies the candidate place with the shortest distance as the location of the location expression included in the object text.

5. A computer readable medium storing a location expression detection program causing a computer to execute a process for extracting a location expression contained in object text, and when the extracted location expression has a plurality of candidate places, determining one of the candidate places, the processing comprising:

storing correspondence of location expressions that can exist in various texts against locations that the location expressions can specify;

extracting a location expression from an object text and determining whether the extracted location expression has a plurality of candidate places by referring to stored contents;

extracting a characteristic word in the object text;

searching for compared text which acts as a compared object to the object text, using the plurality of candidate places and/or the characteristic words; and

determining one of the candidate places based on information related to the searched compared text when the extracted location expression has the plurality of candidate places.