Dialog supporting apparatus that selects similar dialog histories for utterance prediction

• US 7,752,051 B2
• Filed: 02/09/2006
• Issued: 07/06/2010
• Est. Priority Date: 10/08/2004
• Status: Expired due to Fees

• Current Assignee: Intertrust Technologies Corporation (Fidelio Acquisition Co LLC)
• Original Assignee: Panasonic Corporation (Panasonic Holdings Corporation)
• Inventors: Okimoto, Yoshiyuki, Mizutani, Kenji
• Primary Examiner(s): Hudspeth; David R
• Assistant Examiner(s): SHAH, PARAS D
• Application Number:
  US11/349,976
• Publication Number:
  US 20060129409A1
• Time in Patent Office: 1,608 Days
• Field of Search: 704/275.277, 704/9, 704/8, 704/270, 704/275, 704/277, 704231-257, 704 2- 7, 434/156, 707/3
• US Class Current: 704/275
• CPC Class Codes: G06F 40/58 Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

A dialog supporting apparatus is provided which can support an on-going dialog so that the dialog is smoothly completed irrespective of the other party in the dialog. The dialog supporting apparatus includes utterance receiving units which receive the utterances of the dialog participants and output utterance information for identifying an utterance. Moreover, the dialog supporting apparatus includes utterance outputting units which output the utterance information from the other participant in the dialog as an utterance, and an utterance processing unit which transforms the utterance identified by the utterance information into an other utterance form. Furthermore, the dialog supporting apparatus includes a dialog history selection unit which selects an optimum dialog history in a dialog history database and an utterance prediction unit which predicts utterances in the dialog based on the selected dialog history.

Claims

1. A dialog supporting apparatus which supports an on-going dialog made by a first dialog participant and a second dialog participant, said apparatus comprising:

a recording medium for storing a plurality of dialog histories each comprised of dialogs which are utterance strings of a plurality of people including the first dialog participant, the second dialog participant, and one or more dialog participants other than the first dialog participant and the second dialog participant, the dialog histories being stored in association with first information identifying the people, including the first dialog participant, the second dialog participant, and the one or more dialog participants other than the first dialog participant and the second dialog participant, of the dialog histories and second information indicating a topic of a dialog;

a dialog history selection unit operable to;

obtain the first information identifying the first dialog participant and the second dialog participant and the second information indicating the topic of the dialog;

extract, from said recording medium, one or more of the plurality of dialog histories based on the first information identifying the first dialog participant and the second dialog participant and the second information indicating the topic of the dialog, wherein each of the extracted one or more of the plurality of dialog histories is associated with a corresponding first dialog participant or second dialog participant indicated by the obtained first information and the topic of the dialog indicated by the obtained second information, and each of the extracted one or more of the plurality of dialog histories is either (i) a dialog history of a dialog made between the first dialog participant and a third dialog participant other than the second dialog participant or (ii) a dialog history of a dialog made between the second dialog participant and a fourth dialog participant other than the first dialog participant; and

select, from said recording medium, a dialog history that is different from any of the extracted one or more of the plurality of dialog histories, wherein the selected dialog history has a highest degree of similarity with respect to the extracted one or more of the plurality of dialog histories; and

an utterance prediction unit operable to predict a next utterance of either the first dialog participant or the second dialog participant in the dialog, based on the dialog history selected by said dialog history selection unit, and to predict, as a next utterance of either the first dialog participant or the second dialog participant in the dialog, an utterance which is included in the dialog history selected by said dialog history selection unit and appears next to an utterance corresponding to an utterance in the dialog.

2. The dialog supporting apparatus according to claim 1,

wherein said dialog history selection unit is further operable to

calculate a degree of similarity between each of the dialog histories other than the extracted one or more of the plurality of dialog histories, by using as an utterance block, an assembly which corresponds to successive utterances and is commonly included in each of the dialog histories and the extracted one or more of the plurality of dialog histories, to determine degrees of similarity which are inversely proportional to a number of the utterance blocks or to determine degrees of similarity which are proportional to a mean of numbers of utterances included in utterance blocks, wherein the dialog history having the highest degree of similarity is selected, from the determined degrees of similarity, and from among the dialog histories of each of the dialog participants.

3. The dialog supporting apparatus according to claim 2,

wherein said dialog history selection unit is further operable to identify numbers of utterances included in respective assemblies, and calculate, by using the identified numbers of utterances, the degree of similarity which is proportional to a mean of the numbers of utterances included in the respective assemblies.

4. The dialog supporting apparatus according to claim 2,

wherein said dialog history selection unit is further operable to adjust the numbers of utterances in each of two selected dialog histories to a same number using a dynamic programming, based on the dialog histories and the degree of similarity calculated.

5. The dialog supporting apparatus according to claim 2,

wherein said dialog history selection unit is further operable to judge whether or not an utterance is commonly included in each of two selected dialog histories by performing at least one of the following matchings:

a matching based on utterance information for identifying the utterance;

a matching based on a surface expression; and

a matching based on a content word.

6. The dialog supporting apparatus according to claim 1,

wherein said utterance prediction unit is further operable to predict that the next utterance of one of the dialog participants in the dialog is an utterance in the dialog history which is stored next to an utterance in the dialog history corresponding to an utterance in the dialog, and wherein the dialog history is selected by said dialog history selection unit.

7. The dialog supporting apparatus according to claim 1, further comprising:

an utterance receiving unit operable to receive an utterance of one of the dialog participants;

an utterance processing unit operable to transform the utterance received by said utterance receiving unit into another utterance form; and

an utterance outputting unit operable to output the utterance of the other utterance form transformed by said utterance processing unit.

8. The dialog supporting apparatus according to claim 7,

wherein said utterance receiving unit is further operable to:

perform speech recognition of a speech to be inputted from one of the following vocabularies;

prediction utterances predicted by said utterance prediction unit;

sentences which are substantially the same as the prediction utterances;

words included in the prediction utterances; and

words associated with the prediction utterances; and

receive a result of the speech recognition which is the utterance.

9. The dialog supporting apparatus according to claim 7, said apparatus further comprising a display unit, wherein:

said utterance prediction unit is further operable to display the prediction utterance on said display unit; and

said utterance receiving unit is further operable to receive the prediction utterance which is the utterance, when the prediction utterance is selected by one of the dialog participants.

10. The dialog supporting apparatus according to claim 7, said apparatus further comprising a display unit, wherein said utterance prediction unit is further operable to predict a development of utterances made by the dialog participants in the dialog, based on the dialog history selected by said dialog history selection unit, and is further operable to display the predicted development of the utterances on said display unit.

11. The dialog supporting apparatus according to claim 1, said apparatus further comprising a history registration unit operable to register a history of the dialog made by the dialog participants in said recording medium.

12. A dialog supporting system comprising:

a dialog supporting apparatus and a server apparatus which support an on-going dialog made by a first dialog participant and a second dialog participant,

wherein said server apparatus comprises;

a recording medium for storing a plurality of dialog histories each comprised of dialogs which are utterance strings of a plurality of people including the first dialog participant, the second dialog participant, and one or more dialog participants other than the first dialog participant and the second dialog participant, the dialog histories being stored in association with first information identifying the people, including the first dialog participant, the second dialog participant, and the one or more dialog participants other than the first dialog participant and the second dialog participant, of the dialog histories and second information indicating a topic of a dialog; and

a dialog history selection unit operable to;

obtain the first information identifying the first dialog participant and the second dialog participant and the second information indicating the topic of the dialog;

extract, from said recording medium, one or more of the plurality of dialog histories based on the first information identifying the first dialog participant and the second dialog participant and the second information indicating the topic of the dialog, wherein each of the extracted one or more of the plurality of dialog histories is associated with a corresponding first dialog participant or second dialog participant indicated by the obtained first information and the topic of the dialog indicated by the obtained second information, and each of the extracted one or more of the plurality of dialog histories is either (i) a dialog history of a dialog made between the first dialog participant and a third dialog participant other than the second dialog participant or (ii) a dialog history of a dialog made between the second dialog participant and a fourth dialog participant other than the first dialog participant; and

select, from said recording medium, a dialog history that is different from the extracted one or more of the plurality of dialog histories, wherein the selected dialog history has a highest degree of similarity with respect to the extracted one or more of the plurality of dialog histories, and

wherein said dialog supporting apparatus comprises an utterance prediction unit operable to;

request one of the dialog histories from said server apparatus by sending, to said server apparatus, the first information identifying the first dialog participant and the second dialog participant; and

predict a next utterance of either the first dialog participant or the second dialog participant in the dialog, based on the dialog history sent by said server apparatus, and predict, as a next utterance of either the first dialog participant or the second dialog participant in the dialog, an utterance which is included in the dialog history selected by said dialog history selection unit and appears next to an utterance corresponding to an utterance in the dialog.

13. A dialog supporting method for supporting an on-going dialog made by a first dialog participant and a second dialog participant, said method comprising:

storing, in a recording medium, a plurality of dialog histories each comprised of dialogs which are utterance strings of a plurality of people including the first dialog participant, the second dialog participant, and one or more dialog participants other than the first dialog participant and the second dialog participant, wherein the dialog histories are stored in association with first information identifying the people, including the first dialog participant, the second dialog participant, and the one or more dialog participants other than the first dialog participant and the second dialog participant, of the dialog histories and second information identifying a topic of a dialog;

obtaining the first information identifying the first dialog participant and the second dialog participant and the second information indicating the topic of the dialog;

extracting, from said recording medium, one or more of the plurality of dialog histories based on the first information identifying the first dialog participant and the second dialog participant and the second information indicating the topic of the dialog, wherein each of the extracted one or more of the plurality of dialog histories is associated with a corresponding first dialog participant or second dialog participant indicated by the obtained first information and the topic of the dialog indicated by the obtained second information, and each of the extracted one or more of the plurality of dialog histories is either (i) a dialog history of a dialog made between the first dialog participant and a third dialog participant other than the second dialog participant or (ii) a dialog history of a dialog made between the second dialog participant and a fourth dialog participant other than the first dialog participant;

selecting, from said recording medium, a dialog history that is different from any of the extracted one or more of the plurality of dialog histories, wherein the selected dialog history has a highest degree of similarity with respect to the extracted one or more of the plurality of dialog histories; and

predicting a next utterance of either the first dialog participant or the second dialog participant in the dialog, based on the dialog history selected in said selecting of the dialog history, and predicting, as a next utterance of either the first dialog participant or the second dialog participant in the dialog, an utterance which is included in the dialog history selected in said selecting the dialog history and appears next to an utterance corresponding to an utterance in the dialog.

14. A non-transitory computer-readable recording medium having stored therein a program for supporting an on-going dialog made by a first dialog participant and a second dialog participant, said program causing a computer to execute at least the following:

storing, in a dialog history database, a plurality of dialog histories each comprised of dialogs which are utterance strings of a plurality of people including the first dialog participant, the second dialog participant, and one or more dialog participants other than the first dialog participant and the second dialog participant, wherein the dialog histories are stored in association with first information identifying the people, including the first dialog participant, the second dialog participant, and the one or more dialog participants other than the first dialog participant and the second dialog participant, of the dialog histories and second information identifying a topic of a dialog;

obtaining the first information identifying the first dialog participant and the second dialog participant and the second information indicating the topic of the dialog;

extracting, from said dialog history database, one or more of the plurality of dialog histories based on the first information identifying the first dialog participant and the second dialog participant and the second information indicating the topic of the dialog, wherein each of the extracted one or more of the plurality of dialog histories is associated with a corresponding first dialog participant or second dialog participant indicated by the first information and the topic of the dialog indicated by the obtained second information, and each of the extracted one or more of the plurality of dialog histories is either (i) a dialog history of a dialog made between the first dialog participant and a third dialog participant other than the second dialog participant or (ii) a dialog history of a dialog made between the second dialog participant and a fourth dialog participant other than the first dialog participant;

selecting, from said dialog history database, a dialog history that is different from any of the extracted one or more of the plurality of dialog histories, wherein the selected dialog history has a highest degree of similarity with respect to the extracted one or more of the plurality of dialog histories; and

predicting a next utterance of either the first dialog participant or the second dialog participant in the dialog, based on the dialog history selected in said selecting of the dialog history, and predicting, as a next utterance of either the first dialog participant or the second dialog participant in the dialog, an utterance which is included in the dialog history selected in said selecting the dialog history and appears next to an utterance corresponding to an utterance in the dialog.