Information processing apparatus and method and program for controlling the same

An information processing method for an information processing apparatus performing dialogue processing, comprising the steps of:

• controlling acquisition of text data described in a natural language, utilizing an acquisition unit,
• wherein the controlling step includes the steps of:
• acquiring a speech signal from users' natural language spoken by the user, utilizing a speech signal acquisition unit;
• and
• analyzing the speech signal acquired in the step of acquiring speech signal and outputting the text data corresponding to the speech signal, utilizing a speech processing unit,
• wherein the speech signal is an electronic signal;
• generating an input word string based on the text data acquired by the acquisition unit, and calculating a similarity between the input word string and each of a plurality of examples each including an example statement and frame information described using a frame format and corresponding to the example statement, the plurality of examples being stored in a predetermined storage unit, utilizing a similarity calculation unit,
• wherein the similarity calculation between the input word string and the example statement comprises weighting on the similarity in according with reliability representing a certainty of a speech recognition result, the reliability being an occurrence probability which is calculated using a state transition probability and an output symbol probability, and
• wherein the weighting on the similarity is performed in accordance with user profile information of the user's personal information and normal behavior pattern stored in a user profile storage unit,
• wherein the frame format includes a plurality of slots representing semantic contents of the example statement;
• selecting an example corresponding to an example statement whose similarity to the text data, the acquisition of which is controlled by the acquisition controlling step, is the highest from among the plurality of examples stored in the predetermined storage unit in accordance with the similarity calculated by the calculating step, utilizing a example selection unit;
• processing the text data, the acquisition of which is controlled by the acquisition controlling step, to convert the text data into the frame format n accordance with the frame information corresponding to the example selected by the selecting step, utilizing a text data processing unit;
• and
• controlling the dialogue processing in accordance with the text data processed and converted into the frame format by the processing step, utilizing a dialogue processing unit,
• wherein the dialogue processing unit puts information corresponding to a target of the acquired text data to be reported to the user in one of the plurality of slots included in the frame format for generating an output statement, and
• wherein the similarity calculation unit uses the similarity for the example statement calculated in a last dialogue processing immediately preceding a current dialogue processing as a correction value or a value for the weighting.

A non-transitory computer-readable storage medium storing a computer program that when executed on a computer causes dialogue processing, the program comprising the steps of:

• controlling acquisition of text data described in a natural language,
• wherein the controlling step includes the steps of:
• acquiring a speech signal from users' natural language spoken by the user;
• and
• analyzing the speech signal acquired in the step of acquiring speech signal and outputting the text data corresponding to the speech signal,
• wherein the speech signal is an electronic signal;
• generating an input word string based on the text data;
• calculating a similarity between the input word string and each of a plurality of examples each including an example statement and frame information described using a frame format and corresponding to the example statement, the plurality of examples being stored in a predetermined storage unit, utilizing a similarity calculation unit,
• wherein the similarity calculation between the input word string and the example statement comprises weighting on the similarity in according with reliability representing a certainty of a speech recognition result, the reliability being an occurrence probability which is calculated using a state transition probability and an output symbol probability, and
• wherein the weighting on the similarity is performed in accordance with user profile information of the user's personal information and normal behavior pattern stored in a user profile storage unit,
• wherein the frame format includes a plurality of slots representing semantic contents of the example statement;
• selecting an example corresponding to an example statement whose similarity to the text data, the acquisition of which is controlled by the acquisition controlling step, is the highest from among the plurality of examples stored in the predetermined storage unit in accordance with the similarity calculated by the calculating step;
• processing the text data, the acquisition of which is controlled by the acquisition controlling step, to convert the text data into the frame format in accordance with the frame information corresponding to the example selected by the selecting step;
• controlling the dialogue processing in accordance with the text data processed and converted into the frame format by the processing step;
• and
• putting information corresponding to a target of the acquired text data to be reported to the user in one of the plurality of slots included in the frame format for generating an output statement,
• wherein the similarity calculation unit uses the similarity for the example statement calculated in a last dialogue processing immediately preceding a current dialogue processing as a correction value or a value for the weighting.

An information processing apparatus performing dialogue processing, comprising:

• a processor;
• memory;
• an acquisition unit acquiring text data described in a natural language,
• wherein the acquisition unit includes:
• speech signal acquisition unit acquiring a speech signal from users' natural language spoken by the user;
• and
• speech processing unit analyzing the speech signal acquired by the speech signal acquisition unit and outputting the text data corresponding to the speech signal,
• wherein the speech signal is an electronic signal;
• an example storage unit storing a plurality of examples each including an example statement and frame information described using a frame format and corresponding to the example statement,
• wherein the frame format includes a plurality of slots representing semantic contents of the example statement;
• a similarity calculation unit generating an input word string based on the text data acquired by the acquisition unit, and calculating a similarity between the input word string and the example statement stored by the example storage unit,
• wherein the similarity calculation between the input word string and the example statement comprises weighting on the similarity in according with reliability representing a certainty of a speech recognition result, the reliability being an occurrence probability which is calculated using a state transition probability and an output symbol probability, and
• wherein the similarity calculation unit performs the weighting on the similarity in accordance with user profile information of the user's personal information and normal behavior pattern stored in a user profile storage unit;
• an example selection unit selecting an example corresponding to an example statement whose similarity to the text data acquired by the acquisition unit is the highest from among the plurality of examples stored by the example storage unit in accordance with the similarity calculated by the similarity calculation unit;
• a text data processing unit processing the text data acquired by the acquisition unit to convert the text data into the frame format in accordance with the frame information corresponding to the example selected by the example selection unit;
• and
• a dialogue processing unit performing the dialogue processing in accordance with the text data processed and converted into the frame format by the text data processing unit,
• wherein the dialogue processing unit puts information corresponding to a target of the acquired teat data to be included in the frame format for generating an output statement, and
• wherein the similarity calculation unit uses the similarity for the example statement calculated in a last dialogue processing immediately preceding a current dialogue processing as a correction value or a value for the weighting.