INFORMATION RETRIEVAL APPARATUS, INFORMATIN RETRIEVAL SYSTEM, AND INFORMATION RETRIEVAL METHOD

• US 20090234847A1
• Filed: 02/10/2009
• Published: 09/17/2009
• Est. Priority Date: 03/11/2008
• Status: Active Grant

• Current Assignee: Clarion Co., Ltd. (Faurecia SA)
• Original Assignee: Xanavi Informatics Comporation
• Inventors: Takahashi, Hisashi, HOMMA, Takeshi, Kokubo, Hiroaki
• Granted Patent: US 8,073,845 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G01C 21/3611 using character input or menus, e.g. menus of POIs character input methods in general G06F3/0233

Abstract

Provided is an information retrieval method including: retrieving, by a computer, a name including input characters from a database for storing the name, an attribute word associated with the name, and a degree of relevance between the name and the attribute word; outputting the retrieved name as a candidate name; and extracting an attribute word associated with the candidate name, the extracting including: calculating a degree of independency indicating a degree of difference between the extracted attribute words, a degree of coverage indicating an extent to which the combination of the extracted attribute words covers the candidate names, and a degree of equality of a number of corresponding candidate names for each attribute word; and calculating a score of the combination of the attribute words based on at least one of the independency, the coverage and the equality to output the combinations of the attribute words to an output unit.

Claims

1. An information retrieval apparatus, comprising:

an input unit for inputting characters;

a database for storing a name, an attribute word associated with the name, and a degree of relevance indicating a degree of relevance between the name and the attribute word;

a name retrieval unit for retrieving a name including the input characters from the database so as to output the retrieved name as a candidate name;

an attribute word generating unit for extracting an attribute word associated with the candidate name output from the name retrieval unit from the database; and

an output unit for displaying the attribute word extracted by the attribute word generating unit and the candidate name from the name retrieval unit, wherein;

the attribute word generating unit is configured to;

obtain a degree of relevance of the candidate name associated with the attribute word from the database with respect to a combination of the extracted attribute words;

calculate a degree of independency indicating a degree of difference between the extracted attribute words;

calculate a degree of coverage indicating an extent to which the combination of the extracted attribute words covers the candidate names; and

calculate a degree of equality indicating uniformity of a number of corresponding candidate names for each attribute word;

the attribute word generating unit comprises a first score calculating unit for calculating a score of the combination of the attribute words based on at least one of the calculated degree of independency, the calculated degree of coverage and the calculated degree of equality; and

the attribute word generating unit outputs the combinations of the attribute words to the output unit in a descending order of the score.

2. The information retrieval apparatus according to claim 1, wherein the attribute word stored in the database comprises a morpheme constituting a part of the candidate name.

3. The information retrieval apparatus according to claim 1, wherein:

the attribute word generating unit is further configured to;

generate an attribute indicating an irrelevance with the combination of the extracted attribute words as a second attribute word so as to calculate a degree of relevance between the second attribute word and the candidate name;

calculate a degree of independency indicating the degree of difference between the candidate names of the extracted attribute words based on the degree of relevance between the combination of the extracted attribute words and the candidate name as well as the degree of relevance between the second attribute word and the candidate name;

calculate a degree of coverage indicating an extent to which the combination of the extracted attribute words covers the candidate names based on the degree of relevance between the combination of the extracted attribute words and the candidate name as well as the degree of relevance between the second attribute word and the candidate name; and

calculate a degree of equality indicating the uniformity of the number of corresponding candidate names between the attribute words based on the degree of relevance between the combination of the extracted attribute words and the candidate name as well as the degree of relevance between the second attribute word and the candidate name;

the attribute word generating unit comprises a second score calculating unit for calculating a score of the combination of the attribute words based on at least one of the calculated degree of independency, the calculated degree of coverage and the calculated degree of equality; and

the attribute word generating unit outputs the combinations of the attribute words to the output unit in the descending order of the score calculated by the second score calculating unit.

4. The information retrieval apparatus according to claim 3, wherein the attribute word generating unit compares the score calculated by the first score calculating unit with the score calculated by the second score calculating unit, so as to output the combination of the attribute words having a higher score to the output unit.

5. The information retrieval apparatus according to claim 1, wherein:

the attribute words stored in the database are classified according to a type of the attribute word; and

the attribute word generating unit calculates the score of the combination of the attribute words for each type of the attribute word, so as to output the combination of the attribute words having a maximum score to the output unit.

6. The information retrieval apparatus according to claim 1, further comprising:

a second input unit for selecting the candidate name and the attribute word output to the output unit;

an entered name storing unit for storing the candidate name and the attribute word selected by the second input unit as well as the attribute word that has not been selected by the second input unit; and

an updating unit for degree of relevance for updating the degree of relevance stored in the database based on the candidate name and the attribute word stored in the entered name storing unit,

wherein the updating unit for degree of relevance increases the degree of relevance between the candidate name and the attribute word selected by the second input unit, and decreases the degree of relevance between the attribute word that has not been selected by the second input unit and the candidate name, among the candidate names and the attribute words stored in the entered name storing unit, so as to update a value of the degree of relevance stored in the database.

7. The information retrieval apparatus according to claim 1, wherein:

the database includes a table of numeric values assigned to one of each name and a classified group of a plurality of the names; and

the attribute word generating unit uses a value obtained by multiplying the numeric value and the degree of relevance corresponding to the candidate name, as a new degree of relevance.

8. The information retrieval apparatus according to claim 1, further comprising a communication unit for receiving the degree of relevance,

wherein the degree of relevance stored in the database is updated with the received degree of relevance.

9. The information retrieval apparatus according to claim 2, wherein the attribute word generating unit is further configured to:

calculate the attribute word having a high score;

retrieve the candidate name by character strings before and after the attribute word;

generate a character string in which the attribute word matches the candidate name for a longer part; and

output the generated character string as the attribute word to the output unit.

10. The information retrieval apparatus according to claim 2, wherein the attribute word stored in the database comprises:

a morpheme that is a part of the name; and

position information indicating a position of the morpheme in the name.

11. The information retrieval apparatus according to claim 1, further comprising:

a speech input unit for obtaining a speech signal; and

a speech recognizer for recognizing words from the obtained speech signal so as to output the recognized words,

wherein the words output from the speech recognizer are input to the input unit.

12. The information retrieval apparatus according to claim 1, wherein the attribute word generating unit is further configured to:

calculate a degree of understandability indicating a general level of the degree of relevance by the combination of the attribute words; and

calculate a score of the combination of the attribute words based on at least one of the calculated degree of independency, the calculated degree of coverage, the calculated degree of equality and the calculated degree of understandability.

13. The information retrieval apparatus according to claim 12, wherein the degree of understandability is an indicator indicating easiness of understanding a relationship between the attribute word and the name.

14. The information retrieval apparatus according to claim 1, wherein the degree of independency is an indicator of different attribute words between the extracted attribute words.

15. The information retrieval apparatus according to claim 1, wherein the degree of coverage is an indicator indicating what ratio of the candidate names can be selected by the combination of the extracted attribute words.

16. The information retrieval apparatus according to claim 1, wherein the degree of equality is an indicator indicating uniformity of the number of candidate names corresponding to the extracted attribute word.

17. An information retrieval system, comprising:

an information retrieval apparatus for retrieving a name including input characters from a database; and

a server coupled to the information retrieval apparatus via a network, wherein;

the information retrieval apparatus comprises;

an input unit for inputting characters;

a database for storing the name, an attribute word associated with the name, and a degree of relevance indicating a degree of relevance between the name and the attribute word;

a name retrieval unit for retrieving a name including the input characters from the database so as to output the retrieved name as a candidate name;

an attribute word generating unit for extracting an attribute word associated with the candidate name output from the name retrieval unit from the database;

an output unit for displaying the attribute word extracted by the attribute word generating unit and the candidate name from the name retrieval unit;

a communication unit for communicating with the server; and

an updating unit for degree of relevance for updating the degree of relevance stored in the database with a degree of relevance received from the server;

the attribute word generating unit is configured to;

obtain a degree of relevance of the candidate name associated with the attribute word from the database with respect to a combination of the extracted attribute words;

calculate a degree of independency indicating a degree of difference between the extracted attribute words;

calculate a degree of coverage indicating an extent to which the combination of the extracted attribute words covers of the candidate names; and

calculate a degree of equality indicating uniformity of a number of corresponding candidate names for each attribute word;

the attribute word generating unit comprises a first score calculating unit for calculating a score of the combination of the attribute words based on at least one of the calculated degree of independency, the calculated degree of coverage and the calculated degree of equality; and

the attribute word generating unit outputs the combinations of the attribute words to the output unit in a descending order of the score.

18. The information retrieval system according to claim 17, wherein the attribute word stored in the database comprises a morpheme constituting a part of the candidate name.

19. An information retrieval method of retrieving a name including characters input in a computer from a database, the method comprising:

retrieving, by the computer, the name including the input characters from a database for storing the name, an attribute word associated with the name, and a degree of relevance indicating a degree of relevance between the name and the attribute word;

outputting the retrieved name as a candidate name;

extracting an attribute word associated with the candidate name from the database; and

outputting the extracted attribute word and the candidate name to an output unit of the computer,

wherein the extracting an attribute word associated with the candidate name from the database comprises;

obtaining a degree of relevance of the candidate name associated with the attribute word from the database with respect to a combination of the extracted attribute words;

calculating a degree of independency indicating a degree of difference between the extracted attribute words;

calculating a degree of coverage indicating an extent to which the combination of the extracted attribute words covers the candidate names;

calculating a degree of equality indicating uniformity of a number of corresponding candidate names for each attribute word;

calculating a score of the combination of the attribute words based on at least one of the calculated degree of independency, the calculated degree of coverage and the calculated degree of equality; and

outputting the combinations of the attribute words to the output unit in a descending order of the score.

20. The information retrieval method according to claim 19, wherein the attribute word stored in the database comprises a morpheme constituting a part of the candidate name.