Search device, search method, search program, and recording medium

Patent metadata

Field	Value
Publication number	US-11182605-B2
Application number	US-201916532762-A
Country	US
Kind code	B2
Filing date	Aug 6, 2019
Priority date	Nov 2, 2018
Publication date	Nov 23, 2021
Grant date	Nov 23, 2021

Abstract

A search device identifies names of POI from a document group having not been determined. A storage unit that stores a POI presence/absence learning model having learned contexts relating to presence/absence of POI, a POI state learning model having learned contexts relating to states of POI, and a POI name learning model having learned features relating to names of POI, an acceptance unit that accepts a first document group that is a determination target, first and second determination units and an identifying unit that identifies a name of a POI using the POI name learning model from each document of a third document group for which information relating to states of POI is determined by the second determination unit in a second document group are included.

First claim

What is claimed is:

1. A search device comprising:
a storage unit that stores a Point(s) of Interest (POI) presence/absence learning model having learned contexts relating to presence/absence of POI, a POI state learning model having learned contexts relating to states of POI, and a POI name learning model having learned features relating to names of POI;
an acceptance device that accepts a first document group that is a determination target; and
a central processing unit that:
determines presence/absence of information relating to POI using the POI presence/absence learning model for each document of the first document group;
determines information relating to states of POI using the POI state learning model for each document of a second document group for which it is determined by the central processing unit that a POI is present in the first document group; and
identifies a name of a POI using the POI name learning model from each document of a third document group for which information relating to states of POI is determined by the central processing unit in the second document group, wherein each document in the third document group is a document having been screened in two stages in an order of determining presence/absence of information relating to POI and determining information relating to states of POI.

2. The search device according to claim 1 , wherein the POI presence/absence learning model is generated based on a word feature vector model generated from documents used for learning features of words and a plurality of document groups for which the presence/absence of POI has been determined.

3. The search device according to claim 2 , wherein the POI presence/absence learning model has learned determination of the presence/absence of POI using a predetermined algorithm based on feature vectors of documents extracted from a plurality of document groups for which the presence/absence of POI has been determined and feature vectors represented in the word feature vector model.

4. The search device according to claim 2 , wherein the POI state learning model is generated based on the word feature vector model and a plurality of document groups for which information relating to states of POI has been determined.

5. The search device according to claim 4 , wherein the POI state learning model has learned determination of information relating to states of POI using a predetermined algorithm based on feature vectors of documents extracted from a plurality of document groups for which the information relating to states of POI has been determined and feature vectors represented in the word feature vector model.

6. The search device according to claim 2 , wherein the POI name learning model is generated based on the word feature vector model and a plurality of document groups for which names of POI have been identified.

7. The search device according to claim 6 , wherein the POI name learning model has learned identification of names of POI using a predetermined algorithm based on feature vectors of documents extracted from a plurality of document groups for which names of POI have been determined and feature vectors represented in the word feature vector model.

8. A search method executed using a computer able to access a storage unit storing a Point(s) of Interest (POI) presence/absence learning model having learned contexts relating to presence/absence of POI, a POI state learning model having learned contexts relating to states of POI, and a POI name learning model having learned features relating to names of POI, the search method comprising:
an accepting step of accepting a first document group that is a determination target;
a first determination step of determining presence/absence of information relating to POI using the POI presence/absence learning model for each document of the first document group;
a second determination step of determining information relating to states of POI using the POI state learning model for each document of a second document group for which it is determined in the first determination step that a POI is present in the first document group; and
an identifying step of identifying a name of a POI using the POI name learning model from each document of a third document group for which information relating to states of POI is determined in the second determination step in the second document group, wherein each document in the third document group is a document having been screened in two stages in an order of determining presence/absence of information relating to POI and determining information relating to states of POI.

9. The search method according to claim 8 , wherein the POI presence/absence learning model is generated based on a word feature vector model generated from documents used for learning features of words and a plurality of document groups for which the presence/absence of POI has been determined.

10. The search method according to claim 9 , wherein the POI presence/absence learning model has learned determination of the presence/absence of POI using a predetermined algorithm based on feature vectors of documents extracted from a plurality of document groups for which the presence/absence of POI has been determined and feature vectors represented in the word feature vector model.

11. The search method according to claim 9 , wherein the POI state learning model is generated based on the word feature vector model and a plurality of document groups for which information relating to states of POI has been determined.

12. The search method according to claim 11 , wherein the POI state learning model has learned determination of information relating to states of POI using a predetermined algorithm based on feature vectors of documents extracted from a plurality of document groups for which the information relating to states of POI has been determined and feature vectors represented in the word feature vector model.

13. The search method according to claim 9 , wherein the POI name learning model is generated based on the word feature vector model and a plurality of document groups for which names of POI have been identified.

14. The search method according to claim 13 , wherein the POI name learning model has learned identification of names of POI using a predetermined algorithm based on feature vectors of documents extracted from a plurality of document groups for which names of POI have been determined and feature vectors represented in the word feature vector model.

15. A non-transitory computer readable medium comprising a search program that causes a computer able to access a storage function of storing a Point(s) of Interest (POI) presence/absence learning model having learned contexts relating to presence/absence of POI, a POI state learning model having learned contexts relating to states of POI, and a POI name learning model having learned features relating to names of POI to execute:
an acceptance function of accepting a first document group that is a determination target;
a first determination function of determining presence/absence of information relating to POI using the POI presence/absence learning model for each document of the first document group;
a second determination function of determining information relating to states of POI using the POI state learning model for each document of a second document group for which it is determined by the first determination step that a POI is present in the first document group; and
an identifying function of identifying a name of a POI using the POI name learning model from each document of a third document group for wh

Assignees

• Toyota Mapmaster Incorporated
• Toyota Motor Co Ltd

Inventors

• Deguchi Satoru
• Nakanishi Kenta
• Jin Xin

Classifications

• G06N20/20 Primary

  Ensemble learning · CPC title

• G06V30/414

  Extracting the geometrical structure, e.g. layout tree; Block segmentation, e.g. bounding boxes for graphics or text · CPC title

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06V20/62

  Text, e.g. of license plates, overlay texts or captions on TV images · CPC title

• G06F16/374

  Thesaurus · CPC title