Apparatus, method and computer program product for optimum translation based on semantic relation between words

• US 20070100601A1
• Filed: 06/06/2006
• Published: 05/03/2007
• Est. Priority Date: 10/27/2005
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Digital Solutions Corporation (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Kimura, Kazuhiro
• Granted Patent: US 8,060,359 B2
• Time in Patent Office: Days
• US Class Current: 704/4
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation
• CPC Class Codes: G06F 40/55 Rule-based translation

Abstract

A machine translation apparatus includes a identification information detection unit that detects information identifiable to a designated object; a receiving unit that receives a source language sentence; a word dividing unit that divides the source language sentence into a plurality of first word by morphological analysis; a deixis detection unit that detects, from the divided word, a deixis indicating the object directly; a correspondence setting unit that sets the identification information of the designated object and the deixis in correspondence with each other; a semantic class determining unit that determines the semantic class indicating a semantic attribute of the designated objectbased on the identification information of the designated object corresponding to the deixis; and a translation unit that translates the source language sentence where the deixis is attached the attribute having the semantic class of the designated object.

Claims

1. A machine translation apparatus comprising:

a identification information detection unit that detects identification information identifiable to a designated object;

a receiving unit that receives a source language sentence;

a word dividing unit that divides the source language sentence into a plurality of first word by morphological analysis;

a deixis detection unit that detects, from the first words, a deixis indicating the object directly;

a correspondence setting unit that sets the identifiable information of the designated object and the deixis in correspondence with each other;

a semantic class determining unit that determines a semantic class indicating a semantic attribute of the designated object based on the identifiable information of the designated object corresponding to the deixis; and

a translation unit that translates the source language sentence where the deixis is attached with the semantic class of the designated object.

2. The machine translation apparatus according to claim 1,

wherein the identification information detection unit detects the identification information of the designated object including the semantic class thereof, and

the semantic class determining unit acquires the semantic class included in the identification information of the designated object and determines the acquired semantic class as the semantic class of the designated object.

3. The machine translation apparatus according to claim 1, further comprising a semantic class information storage unit that stores semantic class information where an identifier to identify the designated object is associated with the semantic class of the designated object,

wherein the identification information detection unit detects the identification information of the designated object including the identifier, and

the semantic class determining unit acquires, from the semantic class information storage unit, the semantic class corresponding to the identifier included in the identification information of the designated object, and determines the acquired semantic class as the semantic class of the designated object.

4. The machine translation apparatus according to claim 1, further comprising a communication unit that is connected to a network, transmits the identifier to a semantic class management server, returns the semantic class corresponding to the identifier for identifying the designated object, and receives the semantic class returned from the semantic class management server,

wherein the identification information detection unit detects the identification information of the designated object including the identifier, and

the semantic class determining unit transmits the identifier to the semantic class management server through the communication unit, acquires through the communication unit the semantic class returned from the semantic class management server in response to the identifier transmitted, and determines the acquired semantic class as a semantic class of the designated object.

5. The machine translation apparatus according to claim 2, further comprising a semantic class hierarchy storage unit that stores a hierarchical relation between a plurality of the semantic classes,

wherein the semantic class determining unit accesses the semantic class hierarchy storage unit, determines whether the acquired semantic class is included in the semantic class of the word included in the deixis, and upon determination that the semantic class is so included, determines the acquired semantic class as the semantic class of the designated object.

6. The machine translation apparatus according to claim 5,

wherein when the semantic class determining unit determines that the acquired semantic class is not included in the semantic class of the word included in the deixis, the correspondence setting unit sets the identification information of the designated object detected again by the identification information detection unit in correspondence with the deixis detected by the deixis detection unit.

7. The machine translation apparatus according to claim 1, further comprising a distance attribute storage unit that stores the deixis and the corresponding distance attribute including information on the range of the distance to the object indicated by the deixis,

wherein the correspondence setting unit acquires the distance attribute corresponding to the deixis from the distance attribute storage unit, determines whether the distance to the designated object from the identification information detection unit is included in the distance range included in the acquired distance attribute, and when so determined, sets the identification information of the designated object detected by the identification information detection unit in correspondence with the deixis detected by the deixis detection unit.

8. The machine translation apparatus according to claim 7,

wherein upon determination that the distance from the identification information detection unit to the designated object is not included in the distance range included in the acquired distance attribute, the correspondence setting unit determines whether the distance from the identification information detection unit to the designated object detected again is included in the distance range included in the acquired distance attribute.

9. The machine translation apparatus according to claim 1, further comprising a rule storage unit that stores a rule in which the semantic class is set in correspondence with a second word in a target language,

wherein the translation unit acquires, from the rule storage unit, the second word corresponding to the semantic class attached to the deixis, and translates the deixis into the second word acquired.

10. The machine translation apparatus according to claim 1, further comprising a rule storage unit that stores a rule in which a third word in a source language is set in correspondence with a second word in a target language, the third word depending on the semantic class grammatically or semantically,

wherein the translation unit acquires, from the rule storage unit, the second word corresponding to the third word which depends on the semantic class attached to the deixis and which is included in the source language sentence, and translates the third word into the second word.

11. The machine translation apparatus according to claim 1,

wherein the identification information detection unit detects the identification information of the designated object from a no-contact IC tag attached to the designated object.

12. The machine translation apparatus according to claim 1,

wherein the identification information detection unit detects the identification information of the designated object from a bar code attached to the designated object.

13. The machine translation apparatus according to claim 1, wherein the identification information detection unit includes

an image pickup unit that picks-up an image of the designated object;

and

an image recognition unit that analyzes the image picked up and acquiring the identification information of the designated object including the semantic class of the designated object.

14. The machine translation apparatus according to claim 1,

wherein the identification information detection unit receives positional information including information on latitude and longitude where the designated object exists, and detects the received positional information as the identification information of the designated object.

15. The machine translation apparatus according to claim 14, further comprising a communication unit that is connected to a network, transmits the positional information to the positional information management server for returning the information on the position corresponding to the positional information, and receives the information on the position returned from the positional information management server,

wherein the semantic class determining unit transmits the positional information to the positional information management server through the communication unit, acquires, through the communication unit, the information on the position returned from the positional information server in response to the transmitted positional information, and based on the information on the position thus acquired, determines the semantic class of the designated object.

16. A machine translation method comprising:

receiving an source language sentence;

dividing the source language sentence into plurality of first word by morphological analysis;

detecting from the first words a deixis indicating the object directly;

detecting identification information of the designated object;

setting identification information of the designated object in correspondence with the deixis;

determining a semantic class indicating the semantic attribute of the designated object based on the identification information of the designated object set in correspondence with the deixis; and

translating the source language sentence where the deixis is attached the semantic class of the designated object.

17. A computer program product having a computer readable medium including programmed instructions for machine translation, wherein the instructions, when executed by a computer, cause the computer to perform:

receiving an source language sentence;

dividing the source language sentence into plurality of first word by morphological analysis;

detecting from the first words a deixis indicating the object directly;

detecting identification information of the designated object;

setting identification information of the designated object in correspondence with the deixis;

determining a semantic class indicating the semantic attribute of the designated object based on the identification information of the designated object set in correspondence with the deixis; and

translating the source language sentence where the deixis is attached the semantic class of the designated object.