LANGUAGE UNDERSTANDING DEVICE

• US 20080294437A1
• Filed: 05/20/2008
• Published: 11/27/2008
• Est. Priority Date: 05/22/2007
• Status: Active Grant

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Okuno, Hiroshi, Nakano, Mikio, Funakoshi, Kotaro, Fukubayashi, Yuichiro, Komatani, Kazunori
• Granted Patent: US 8,244,522 B2
• Time in Patent Office: Days
• US Class Current: 704/251
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning

Abstract

A language understanding device includes: a language understanding model storing unit configured to store word transition data including pre-transition states, input words, predefined outputs corresponding to the input words, word weight information, and post-transition states, and concept weighting data including concepts obtained from language understanding results for at least one word, and concept weight information corresponding to the concepts; a finite state transducer processing unit configured to output understanding result candidates including the predefined outputs, to accumulate word weights so as to obtain a cumulative word weight, and to sequentially perform state transition operations; a concept weighting processing unit configured to accumulate concept weights so as to obtain a cumulative concept weight; and an understanding result determination unit configured to determine an understanding result from the understanding result candidates by referring to the cumulative word weight and the cumulative concept weight.

Claims

1. A language understanding device comprising:

a language understanding model storing unit configured to store, as a language understanding model, word transition data including pre-transition states, input words, predefined outputs corresponding to the input words, word weight information, and post-transition states, and concept weighting data including concepts obtained from language understanding results for at least one word, and concept weight information corresponding to the concepts;

a finite state transducer processing unit configured to output understanding result candidates including the predefined outputs based on words included in a word sequence being input and present states, and in accordance with the word transition data read out from the language understanding model storing unit, to accumulate word weights so as to obtain a cumulative word weight, and to sequentially perform state transition operations by which transitions to the post-transition states are carried out;

a concept weighting processing unit configured to accumulate concept weights, which correspond to the concepts included in the understanding result candidates output from the finite state transducer processing unit, in accordance with the concept weighting data read out from the language understanding model storing unit so as to obtain a cumulative concept weight; and

an understanding result determination unit configured to determine an understanding result from the understanding result candidates by referring to the cumulative word weight and the cumulative concept weight.

2. The language understanding device according to claim 1, wherein

the language understanding model storing unit is further configured to store, as the language understanding model, filler transition data including pre-transition states, fillers matching an arbitrary word, filler weight information, and post-transition states,

the finite state transducer processing unit is further configured to accumulate filler weights in accordance with the filler transition data read out from the language understanding model storing unit so as to obtain a cumulative filler weight, and to perform the state transition operations by which transition to the post-transition states are carried out, and

the understanding result determination unit is further configured to determine the understanding result by also referring to the cumulative filler weight.

3. The language understanding device according to claim 1, wherein

the word sequence being input is obtained by a speech recognition process,

the language understanding model storing unit is further configured to store the word weight information and the concept weight information which are determined depending on speech recognition accuracy during the speech recognition process, and

the understanding result determination unit is further configured to determine the understanding result by referring to the cumulative word weight obtained based on the word weight information which depends on the speech recognition accuracy and the cumulative concept weight obtained based on the concept weight information which depends on the speech recognition accuracy.

4. The language understanding device according to claim 1, wherein

the word sequence being input includes N word-sequences, where N is a natural number equal to or greater than 2,

the finite state transducer processing unit is further configured to output the understanding result candidates for each of the N word-sequences by performing the state transition operations for each of the N word-sequences, and

the understanding result determination unit is further configured to determine the understanding result from all of the understanding result candidates corresponding to the N word-sequences.