Conversation system and conversation software
First Claim
Patent Images
1. A conversation system comprising:
- a speech input device and a speech output device;
an utterance understanding unit adapted to understand an utterance to be inputted into the speech input device;
a conversation state storage unit adapted to store an utterance understanding result as understood by the utterance understanding unit;
a domain selection unit adapted to select one of a first domain and a second domain based on the utterance understanding result as understood by the utterance understanding unit;
a first utterance generation unit adapted to generate an utterance to be outputted based on the utterance understanding result stored in the conversation state storage unit, the utterance to be outputted from the speech output device being based on one of the first and second domains selected by the domain selection unit; and
a database for storing a plurality of texts;
the conversion state storage unit comprising a plurality of slots,each of the plurality of slots being adapted to store each of a plurality of language component candidates, and each of the language component candidates being stored is retrieved from each set of the language component candidates based on a likelihood with regard to each language component included in the utterance inputted into the speech input device,wherein the first utterance generating unit is adapted to retrieve a specific text from the database and to generate the utterance to be outputted using the text that has been retrieved when all of the slots are in a stored state, while the first utterance generating unit generates the utterance to be outputted using the language component candidate stored in a portion of the slots, when only the portion of the slots is in the stored state,wherein the domain selection unit is adapted;
to specify the first domain satisfying a first condition concerning a current utterance understanding result by the utterance understanding unit, and to specify the second domain satisfying a second condition concerning a selection history by the domain selection unit from among the first and second domains,to evaluate a part or all of first, second, and third indices for each of the first and second domains, the first index representing a reliability in consideration of an utterance understanding history by the utterance understanding unit, the second index representing a reliability in consideration of the selection history by the domain selection unit, and the third index representing a reliability in consideration of an utterance generation history by the first utterance generation unit, andto select one of the first domain, the second domain, and a third domain different from the first and second domains, as a current domain, based on the evaluation results and according to a selection rule,the system further comprising;
a second utterance generation unit adapted to generate a different utterance to be output from the speech output device,wherein when the domain selection unit has determined that the third domain satisfying a third condition that is different from either the first condition or the second condition has a second attribute, the utterance to be outputted from the speech output device is generated by the second utterance generation unit instead of the being generated by the first utterance generation unit.
1 Assignment
0 Petitions
Accused Products
Abstract
A first domain satisfying a first condition concerning a current utterance understanding result and a second domain satisfying a second condition concerning a selection history are specified. For each of the first and second domains, indices representing reliability in consideration of the utterance understanding history, selection history, and utterance generation history are evaluated. Based on the evaluation results, one of the first, second, and third domains is selected as a current domain according to a selection rule.
-
Citations
10 Claims
-
1. A conversation system comprising:
-
a speech input device and a speech output device; an utterance understanding unit adapted to understand an utterance to be inputted into the speech input device; a conversation state storage unit adapted to store an utterance understanding result as understood by the utterance understanding unit; a domain selection unit adapted to select one of a first domain and a second domain based on the utterance understanding result as understood by the utterance understanding unit; a first utterance generation unit adapted to generate an utterance to be outputted based on the utterance understanding result stored in the conversation state storage unit, the utterance to be outputted from the speech output device being based on one of the first and second domains selected by the domain selection unit; and a database for storing a plurality of texts; the conversion state storage unit comprising a plurality of slots, each of the plurality of slots being adapted to store each of a plurality of language component candidates, and each of the language component candidates being stored is retrieved from each set of the language component candidates based on a likelihood with regard to each language component included in the utterance inputted into the speech input device, wherein the first utterance generating unit is adapted to retrieve a specific text from the database and to generate the utterance to be outputted using the text that has been retrieved when all of the slots are in a stored state, while the first utterance generating unit generates the utterance to be outputted using the language component candidate stored in a portion of the slots, when only the portion of the slots is in the stored state, wherein the domain selection unit is adapted; to specify the first domain satisfying a first condition concerning a current utterance understanding result by the utterance understanding unit, and to specify the second domain satisfying a second condition concerning a selection history by the domain selection unit from among the first and second domains, to evaluate a part or all of first, second, and third indices for each of the first and second domains, the first index representing a reliability in consideration of an utterance understanding history by the utterance understanding unit, the second index representing a reliability in consideration of the selection history by the domain selection unit, and the third index representing a reliability in consideration of an utterance generation history by the first utterance generation unit, and to select one of the first domain, the second domain, and a third domain different from the first and second domains, as a current domain, based on the evaluation results and according to a selection rule, the system further comprising; a second utterance generation unit adapted to generate a different utterance to be output from the speech output device, wherein when the domain selection unit has determined that the third domain satisfying a third condition that is different from either the first condition or the second condition has a second attribute, the utterance to be outputted from the speech output device is generated by the second utterance generation unit instead of the being generated by the first utterance generation unit. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9)
-
-
10. A conversation program embodied on a non-transitory computer readable medium and configured to cause a computer to function as a conversation system, the conversation system comprising:
-
a speech input device and a speech output device; an utterance understanding unit adapted to understand an utterance to be inputted into the speech input device; a conversation state storage unit adapted to store an utterance understanding result as understood by the utterance understanding unit, a domain selection unit adapted to select one of a first domain or a second domain based on the utterance understanding result as understood by the utterance understanding unit; a first utterance generation unit adapted to generate an utterance to be outputted based on the result stored in the conversation state storage unit, the utterance to be outputted from a speech output device being based on one of the first and second domains selected by the domain selection unit; and a database for storing a plurality of texts; the conversion state storage unit comprising a plurality of slots, each of the plurality of slots being adapted to store each of a plurality of language component candidates, and each of the language component candidates being stored is retrieved from each set of the language component candidates based on a likelihood with regard to each language component included in the utterance inputted into the speech input device, wherein the first utterance generating unit is adapted to retrieve a specific text from the database and to generate the utterance to be outputted using the text that has been retrieved when all of the slots are in a stored state, while the first utterance generating unit generates the utterance to be outputted using the language component candidate stored in a portion of the slots, when only the portion of the slots is in the stored state, the program causing the domain selection unit of the conversation system to perform steps, including; specifying a first domain satisfying a first condition concerning a current utterance understanding result by the utterance understanding unit and a second domain satisfying a second condition concerning a selection history by the domain selection unit from among the first and second domains, evaluating a part or all of first, second, and third indices for each of the first and second domains, the first index representing a reliability in consideration of an utterance understanding history by the utterance understanding unit, the second index representing a reliability in consideration of the selection history by the domain selection unit, and the third index representing a reliability in consideration of an utterance generation history by the first utterance generation unit, and selecting one of the first domain, the second domain, and a third domain different from the first and second domains, as a current domain, based on the evaluation results and according to a selection rule, wherein the computer of the conversation system further comprises; a second utterance generation unit adapted to generate a different utterance to be output from the speech output device, and the program further causing the domain selection unit of the conversation system to perform steps, including; determining that the third domain satisfies a third condition that is different from either the first condition or the second condition, and causing the second utterance generation unit, rather than the third processing unit, to generate the utterance to be outputted from the speech output device.
-
Specification