Method of phrase verification with probabilistic confidence tagging

• US 7,010,484 B2
• Filed: 12/12/2001
• Issued: 03/07/2006
• Est. Priority Date: 08/14/2001
• Status: Expired due to Fees

• Current Assignee: Industrial Technology Research Institute
• Original Assignee: Industrial Technology Research Institute
• Inventors: Lin, Yi-Chung
• Primary Examiner(s): McFadden, Susan
• Assistant Examiner(s): Sked, Matthew J
• Application Number:
  US10/012,483
• Publication Number:
  US 20030083876A1
• Time in Patent Office: 1,546 Days
• Field of Search: 704/252, 704/251, 704/255
• US Class Current: 704/240
• CPC Class Codes: G10L 15/08 Speech classification or search

Abstract

A method of phrase verification to verify a phrase not only according to its confidence measures but also according to neighboring concepts and their confidence tags. First, an utterance is received, and the received utterance is parsed to find a concept sequence. Subsequently, a plurality of tag sequences corresponding to the concept sequence is produced. Then, a first score of each of the tag sequences is calculated. Finally, the tag sequence of the highest first score is selected as the most probable tag sequence, and the tags contained therein are selected as the most probable confidence tags, respectively corresponding to the concepts in the concept sequence.

Claims

1. A method of phrase verification with probabilistic confidence tagging, comprising the steps of:

receiving an utterance;

parsing the received utterance to find a concept sequence, in which the concept sequence comprises a plurality of concepts, each corresponding to at least one word in the utterance;

producing a plurality of tag sequences corresponding to the concept sequence, in which each of the tag sequences comprises a plurality of tags, each indicating whether the corresponding concept is required to be verified, and corresponding possible verification results;

calculating a first score of each of the tag sequences based on a probability corresponding to the tags therein; and

selecting the tag sequence of the highest first score as a most probable tag sequence, and selecting the tags in the most probable tag sequence as most probable confidence tags.

2. The method as claimed in claim 1, wherein the tag has a binary value, respectively representing acceptance and rejection.

3. The method as claimed in claim 1, wherein the tag has multiple values, respectively representing different levels of confidence.

4. The method as claimed in claim 1, wherein the first score is a contextual confidence score.

5. A method of phrase verification with probabilistic confidence tagging, comprising the steps of:

receiving an utterance;

parsing the received utterance to find a concept sequence, in which the concept sequence comprises a plurality of concepts, each corresponding to at least one word in the utterance;

assessing at least one confidence measure of each of the concepts in the concept sequence;

producing a plurality of tag sequences corresponding to the concept sequence, in which each of the tag sequences comprises a plurality of tags, each indicating whether the corresponding concept is required to verify, and corresponding possible verification results;

calculating a first score of each of the tag sequences based on the probability corresponding to the tags therein;

calculating a second score of each of the tag sequences;

proceeding a weighted calculation on the first score and second score, thus a total score is acquired; and

selecting the tag sequence of the highest total score as a most probable tag sequence, and selecting the tags in the most probable tag sequence as most probable confidence tags.

6. The method as claimed in claim 5, wherein the tag has a binary value, respectively representing acceptance and rejection.

7. The method as claimed in claim 5, wherein the tag has multiple values, respectively representing different levels of confidence.

8. The method as claimed in claim 5, wherein the first score is a contextual confidence score.

9. The method as claimed in claim 5, wherein the second score is computed from a set of confidence measure scores.