JP2003186896A - Conversation process device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide related information based on the content of a conversation being presently carried out without having a speaker being aware of it. <P>SOLUTION: The information of the conversation by the speaker is acquired by conversation inputting means 2 and 9 installed in the place where the conversation is taking place, important words are extracted by important word extracting parts 3 and 10 from the acquired information of the conversation, related information relating to the conversation is acquired from related information data base 6 and 13 by related information acquiring parts 5 and 12 using the extracted important words and outputted to related information providing means 7 and 14 installed in the place of conversation. When the device is being operated, the speaker can obtain information related to his own speech or to the speech of the conversing partner instantly even in a remote place without being aware of it. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for retrieving information related to a dialogue and providing the information to the interlocutor.

[0002]

2. Description of the Related Art The following is a conventional technique for extracting important words indicating contents or topics from dialogue information and providing information related to the important words to the interlocutor based on the important words. is there. In "Information providing device and method" described in Japanese Patent Laid-Open No. 11-203295, information generated by dialogue is input from a keyboard input device, a pen input device, a voice recognition device, an image recognition device, or the like, and the topic identification unit , The topic is identified from the input information, and based on the identified topic, the content of the provision information determined by the provision information determination unit is acquired from the provision information database, and the provision information output unit acquires the information recipient (interactive person). ), It is possible to output in a format suitable for.

In the "request prediction type information providing service device" described in Japanese Patent Application Laid-Open No. 8-147310, the recognition means recognizes the conversation content of a person, and the database retrieves information related to the recognized conversation content. Search using
It is possible to present the retrieved information to the person having the conversation by the presenting means. In the “advertising device for electronic dialogue” described in Japanese Patent Laid-Open No. 11-3348,
(Chat) When the word in the sentence input by the participant matches the word registered in the electronic interactive advertising device, it is possible to provide the advertising message that matches the word.

In the "dialogue processing device" described in Japanese Unexamined Patent Publication No. 10-69482, a statement processing module is provided in order to always activate a participating dialogue and thereby enhance the creativity of the dialogue. Morphological analysis of the participants' comments to extract the keywords, and the dialog structuring module arranges the relationships between the keywords and comments in a two-dimensional space to structure the contents of the dialog and recognize the topic development situation. The module grasps the development status of the topic from the content of the conversation and the transition information of the utterance, and the topic providing module activation module extracts the information which becomes the topic seed in response to the detection of the topic conversion. It is possible to provide. JP 20
In the "speech recognition device" described in Japanese Unexamined Patent Publication No. 00-122692, the speech recognition result of one speaker is compared with the speech recognition result of the other speaker, and two speech recognition results are predetermined. It is possible to provide information only when the conditions are met.

Also, Nishimoto et al.'S group thinking support environment AIDE
(Kazushi Nishimoto, Yasuyuki Kakubayashi, Rieko Kadobayashi et al .: "Group Thinking Support by Multi-agent", IEICE Transactions DI, Vol.J81-DI, No.5, pp.478-487, May 1998) Now, we will visualize the semantic structure of the current dialogue information, provide a shared information space for all participants, and, as a virtual dialogue participant, display the text related to the shared information space at that time from an external text base. It is possible to automatically extract and throw it into the information sharing information space. In addition, the conversation support agent Zinger (Kurabayashi Noriyuki, Yamazaki Tatsuya,
Taichi Yuasa: "Zinger: Conversation support agent based on interest similarity", IPSJ Human Interface Research Group, 2001-HI-94, pp.53-60), in which participants' interests were observed by observing conversation. It is possible to estimate and recommend appropriate topics.

Further, there are the following techniques as techniques for predicting, estimating or inferring from a sentence or a character string using the co-occurrence of words. In the “similarity search method and apparatus” described in Japanese Patent Laid-Open No. 9-153068, a search formula including a plurality of attributes, the strength of the co-occurrence relation of the attributes is calculated, and the importance of the attribute is increased. Proposed method technology.

In the "document composition device" described in Japanese Patent Laid-Open No. 10-207889, a method for reducing excessive indication of correct answer candidates for an error part in a text composition device is referred to by referring to a database relating to occurrence probabilities of words or word strings. The technology of is published. In the "language processing device" described in Japanese Patent Application Laid-Open No. 10-97533, a language processing device that understands a language expression is used as a sentence meaning determination means, based on concept co-occurrence frequency data, which is calculated in advance. The technology that is characterized by storing the reliability of the relationship is open to the public.

[0008]

Although a dialogue is made in relation to past dialogue contents, any of the above-mentioned conventional techniques can be used to provide related information that sufficiently reflects the past dialogue contents. There was a problem that I could not do it. In addition, for a plurality of remote interlocutors, topics or important words that are common to each area or both are extracted from the local dialogues, and information related to the topics or important words can be synchronized with each other remotely or synchronously. There was a problem that it could not be provided.

The present invention has been made in view of the above-mentioned conventional circumstances, and an object thereof is to provide relevant information to an interlocutor in accordance with the contents of the ongoing dialogue. Further objects of the present invention will be apparent from the following description.

[0010]

The present invention is a dialogue processing apparatus for providing a speaker with information related to a topic, a dialogue input means for inputting dialogue information spoken by the speaker, and the inputted dialogue. An important word extracting means for extracting an important word from the information,
A related information acquisition means for acquiring related information related to the dialogue from the related information database using the extracted important word; and an information providing means for providing the acquired related information to a speaker. There is. Therefore, it is possible to provide the speaker with relevant information that reflects the contents of past dialogue.

Further, in the dialogue processing apparatus according to the present invention, the dialogue processing apparatus is connected to another dialogue apparatus having the same function means via a network, and at least one of the dialogue processing apparatuses is provided with the important word. An important word transfer means is provided for transferring the important word extracted by the extracting means to another dialogue processing device for use by the related information obtaining means. Therefore, information related to the topic or the key word can be provided to a plurality of remote interlocutors synchronously or asynchronously.

Further, in the dialogue processing apparatus according to the present invention, the first dialogue processing apparatus and the second dialogue processing apparatus each including the important word transfer means are connected via a network, and the first dialogue processing is performed. The device inputs the dialogue information input by the user from its own dialogue input means, extracts the important word from the dialogue information by its own important word extraction means, and sets the extracted importance degree to its own important word transfer means. And related information acquisition means, and the important word transfer means of the first dialogue processing device outputs the extracted important word to the second via the network.
To the related information acquisition means of the dialogue processing device,
Related information acquisition means of the dialogue processing device from the related information database using the important word input from the important word extraction means of the first dialogue processing device and / or the important word transfer means of the second dialogue processing device as a search word. The related information relating to the dialogue is acquired and provided to the user by the relevant information providing means of the first dialogue processing apparatus. Therefore, for a plurality of remote interlocutors, a topic or an important word that is common to each area or both is extracted from the local dialogue, and information related to the topic or the important word is synchronously or asynchronously transmitted to each of the local and remote areas. Can be provided.

Further, in the dialogue processing apparatus according to the present invention, the important word extraction means has a morpheme analysis means, performs morpheme analysis on the dialogue information inputted from the dialogue input means, and inputs from the morpheme analysis result. Select important words related to dialogue information, calculate importance between important words from appearance frequency of important words, and select important words as candidates for search words to search information related to dialogue according to the importance. Extract. Further, in the dialogue processing device according to the present invention, the important word extracting means calculates the importance degree between the important words according to a predetermined time or according to the conversion of the contents of the dialogue. Therefore, it is possible to provide the related information in consideration of the importance calculated from the appearance frequency of the important word.

Further, in the dialogue processing apparatus according to the present invention, the important word extracting means extracts as a search word candidate for searching related information related to the dialogue based on the calculated importance of the important word. The important word and its frequency are stored in the important word storage means for subsequent importance calculation, and the important word and its frequency are used at the time of the next importance calculation. Therefore, it is possible to speed up the process of identifying the important word and provide the speaker with relevant information that reflects the important word based on the content of the past dialogue.

Further, in the dialog processing apparatus according to the present invention, the related information providing means provides the plurality of related information input from the related information acquisition means to the user sequentially or simultaneously on the screen and simultaneously provides them. Further, in the dialogue processing device according to the present invention, the related information providing means changes the providing time or the size of the providing area of the related information provided to the user according to the importance of the extracted important word. To do. Therefore, it is possible to provide a plurality of related information to the user in an easily understandable manner, and also to easily grasp the importance of these related information.

Further, in the dialogue processing apparatus according to the present invention, the important word transfer means has an important word translation means for translating the important word from the first language into a second language different from the first language. Therefore, even if there is a difference in the dialogue language between the dialogue processing devices, it is possible to use the important words with each other.

Further, the dialogue processing apparatus according to the present invention comprises user information input means for inputting user information relating to the characteristics of the user, and the related information acquisition means inputs from the important word and the user information input means. The relevant user information is used to search the relevant information database for relevant information related to the dialogue. Further, the dialogue processing apparatus according to the present invention comprises environment information input means for inputting environment information relating to the environment in which the dialogue is conducted, and the related information acquisition means uses the important words and the environment information inputted from the environment information input means. Then, the related information related to the dialogue is retrieved from the related information database. Therefore, it is possible to provide the speaker with relevant information that reflects the characteristics of the interlocutor and the environment in which the interlocutor is engaged.

Further, the present invention is a program for realizing a function of providing information related to a topic to a speaker by a computer, a function of extracting an important word from dialogue information spoken by the speaker, and the extraction. The computer realizes the function of acquiring the related information related to the dialogue from the related information database using the acquired important word and the function of providing the acquired related information to the speaker.

Further, the present invention is a method for providing information related to a topic to a speaker, which acquires conversation information spoken by a speaker by a microphone installed at a place where the conversation is conducted, and obtains the information. The important word is extracted from the dialogue information, the relevant information related to the dialogue is acquired from the relevant information database by using the extracted important word, and output from the relevant information providing means installed at the place where the dialogue is held. The operation to perform is performed by a series of computer processes. Therefore,
If the computer is running to perform the method,
The interlocutor can immediately (in synchronization) obtain information related to his / her own remarks and the remarks of the other party on the spot without any awareness, and can further promote the conversation.

In the dialog processing method, in extracting the important word, the acquired dialog information is morphologically analyzed,
Search for selecting important words related to dialogue information input from morphological analysis results, calculating importance between important words from appearance frequency of important words, and searching information related to dialogue according to the importance In the process of extracting important words as word candidates, and in the process of calculating the degree of importance between important words, the frequency of important words is calculated according to the predetermined time or the turning point of the content of the dialogue. An important word frequency calculation step and a vector about the frequency are formed from the calculated frequencies of the important words, and an included angle calculation step of calculating an included angle between the vectors, and a smaller included angle corresponds to the associated important word. A minimum included angle extraction step for extracting the minimum included angle with a high degree of importance, and a corresponding important word selection step for selecting the corresponding important word from the vector used to calculate this minimum included angle. Extracting important words that are different from the contrasting important words by comparing with the corresponding important word selecting step Using the different important word extracting step and the important words contrasting with the different important words, searching for information related to the dialogue It is preferable to include a search word candidate extraction step as a search word candidate for.

[0021]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, a dialogue processing apparatus of the present invention will be specifically described with reference to the drawings. FIG. 1 is a block diagram showing a dialogue processing apparatus according to the first embodiment of the present invention. The local dialogue processing apparatus 1 is configured to include a dialogue input unit 2, an important word extraction unit 3, an important word transfer unit 4, a related information acquisition unit 5, a related information database 6 and a related information providing unit 7. Further, the remote dialogue processing device 8 connected to the local dialogue processing device 1 via the network 15 similarly has the dialogue input unit 9, the important word extraction unit 10, the important word transfer unit 11, the related information acquisition unit 12, and the related information. Database 13
And a related information providing unit 14.

That is, the local dialogue processing apparatus 1 and the remote dialogue processing apparatus 8 installed apart from each other are connected to each other via the network 15. Each of the functional units of the local dialogue processing apparatus 1 and the remote dialogue processing apparatus 8 is configured by executing a predetermined program by a computer that constitutes the dialogue processing apparatus, and each functional unit performs each processing described below. Take action.

Here, the user A and the user B use the local dialogue processing apparatus 1 to have a dialogue, while the users C and D use the remote dialogue processing apparatus 8 to have a dialogue. The following description will be made assuming that there is a case.

First, the operation of the local dialogue processing apparatus 1 will be outlined. Dialog input unit 2 of the local dialog processor 1
Inputs dialogue information of users A and B and outputs the obtained dialogue information to the important word extraction unit 3. The important word extraction unit 3 inputs the dialogue information from the dialogue input unit 2, extracts the important word from the dialogue information, and uses the extracted important word as a candidate for the search word, the important word transfer unit 4 and the related information acquisition unit 5.
Output to. Important word transfer unit 4 of local dialogue processing apparatus 1
Inputs an important word from the important word extraction unit 3 and outputs the important word to the related information acquisition unit 12 of the remote dialogue processing device 8.

The related information acquisition unit 5 of the local dialogue processing device 1 uses the important words input from the important word extraction unit 3 or the important words input from the important word transfer unit 11 of the remote interaction processing device 8 via the network 15, A search word (search expression) for searching the information related to the dialogue is configured, and the information related to the dialogue is related to the dialogue from the related information database 6 by using the search word (hereinafter, may be abbreviated as dialogue related information). To get. Further, the related information acquisition unit 5 outputs the dialog related information, which is the search result from the related information database 6, to the related information providing unit 7.

The related information database 6 uses the search word input from the related information acquisition unit 5 to select information related to the search word, that is, information related to the dialogue, from the information stored inside, and the related information is acquired. Return to the acquisition unit 5. The related information providing unit 7 provides the user A and / or the user B with the dialog related information input from the related information acquisition unit 5.

Next, the operation of the remote dialogue processing device 8 will be outlined. Dialog input unit 9 of the remote dialog processing device 8
Inputs dialogue information of the users C and D, and outputs the obtained dialogue information to the important word extraction unit 10. The important word extraction unit 10 inputs the dialogue information from the dialogue input unit 9,
An important word is extracted from the dialogue information, and the extracted important word is output to the important word transfer unit 11 and the related information acquisition unit 12 as a search word candidate. The important word transfer unit 11 of the remote dialogue processing device 8 inputs the important word from the important word extraction unit 10,
The important word is output to the related information acquisition unit 5 of the local dialogue processing device 1.

The related information acquisition unit 12 of the remote dialogue processing device 8 uses the important words input from the important word extraction unit 10 or the important words input from the important word transfer unit 4 of the local interaction processing device 1 via the network 15 to A retrieval word (retrieval expression) for retrieving information related to the dialogue is configured, and the dialogue related information is acquired from the relevant information database 13 using the retrieval word. Further, the related information acquisition unit 12 outputs the dialog related information, which is the search result from the related information database 13, to the related information providing unit 14.

The related information database 13 uses the search word input from the related information acquisition unit 12 to select information related to the search word, that is, information related to the dialogue, from the information stored inside, and the related information is acquired. Return to the acquisition unit 12. The related information providing unit 14 provides the user C and / or the user D with the dialog related information input from the related information acquisition unit 12.

Next, the function and operation of each part of the local dialogue processing apparatus 1 and the remote dialogue processing apparatus 8 will be described in detail. In this example, since the functions and operations of the respective units of the local dialogue processing apparatus 1 and the remote dialogue processing apparatus 8 are the same, the local dialogue processing apparatus 1 will be described as a representative.

FIG. 2 is an example of a configuration diagram of the dialogue input unit 2. As shown in the figure, the dialogue input unit 2 includes a microphone device 21 and a dictation processing device 22. The microphone device 21 inputs voice-type dialogue information from the user, and the dictation processing device 22
Output to. The dictation processing device 22 converts voice-type dialogue information input from the microphone device 21 into text-type dialogue information. In addition, as software that currently realizes dictation processing, there is ViaVoice (trade name) manufactured by IBM Corporation.

FIG. 3 is an example of a configuration diagram of the important word extraction unit 3. As shown in the figure, the important word extraction unit 3 includes a dialogue comparison device 23, a dialogue recording device 24, a morphological analysis device 25, a word dictionary 26, a significant word extraction device 27, and a significant word storage device 28. The dialogue comparison device 23 inputs the dialogue information in text format output from the dictation processing device 22 of the dialogue input unit 2, and compares the input dialogue information with the dialogue information recorded in the dialogue recording device 24. , Dialogue information including a matching word is extracted, and the extracted dialogue information is output to the morphological analyzer 25. The dialogue information is unit information in which the contents of the dialogue such as phrases and paragraphs of the dialogue are divided into certain units. Further, the dialogue comparison device 23 accumulates and records the compared dialogue information in the dialogue recording device 24 as a dialogue log.

The morpheme analysis device 25 inputs the comparison result result from the dialogue comparison device 23, performs morpheme analysis on the dialogue information appearing as the comparison result while referring to the word dictionary 26, and extracts the morpheme analysis result from the important word. Output to the device 27. The word dictionary 26 is a collection of word data to be referred to when performing morphological analysis.

The important word extraction device 27 is a morphological analysis device 2
From the words input from 5, the number of important words within a certain period of time or between the turning points of the dialogue (topic) is counted, the count result is output to the important word storage device 28, and the important words satisfying a specific criterion are searched word candidates. As the important word transfer unit 4 and the related information acquisition unit 5
Output to. Here, the specific criterion is, for example, the top three most important words having the highest frequency or the highest frequency. The important word storage device 28 inputs and holds the important words from the important word extracting device 27.

FIG. 4 is an example of a configuration diagram of the important word transfer unit 4. As shown in the figure, the important word transfer unit 4 includes an important word storage device 29 and an important word transmission device 30.
The important word storage device 29 provides a buffer function for transmitting and receiving important words between the local interaction processing device 1 and the remote interaction processing device 8. In order to transmit the important word output from the important word extracting device 27 of the important word extracting unit 3 from the local dialogue processing device 1 to the remote dialogue processing device 8, the important word storage device 29, conversely, the remote dialogue processing device. In order to transmit from 8 to the local dialogue processing device 1, the important word is temporarily held.

The important word transmitting device 30 includes the important word storage device 2
The important word held in 9 is transmitted from the local dialogue processing apparatus 1 to the remote dialogue processing apparatus 8, or is transmitted from the remote dialogue processing apparatus 8 to the local dialogue processing apparatus 1.

FIG. 5 is an example of a configuration diagram of the related information acquisition unit 5. As shown in the figure, the related information acquisition unit 5 is composed of a search word configuration device 31, a search device 32, and a search result storage device 33. The search word constructing device 31 inputs an important word that is a candidate for the search word output from the important word extracting unit 3, or the search word constructing device 31 uses the important word transferring unit 11 of the remote dialogue processing device 8 to connect to the network. Input the important word output via 15.

The retrieval word construction device 31 constructs a retrieval word (retrieval expression) used to retrieve the dialogue related information from the related information database 6 based on the inputted important word. For example,
When the important word extracted from the important word extracting unit 3 is “wine”, the important word transferring unit 11 of the remote dialogue processing device 8
When the important word output from the network via the network 15 is "glass", the important word "glass" is added to the important word "wine", and "(wine) or (glass)" or "(wine)" is added. Configure the search term as "and (glass)" or the like. Further, in order not to search the pornographic information, the search word is configured as “((wine) or (glass)) not (pornography)” or the like. In addition, such a logical child and a condition can be set by the user.

The search word construction device 31 outputs the constructed search word to the search device 32, and the search device 32 inputs the search word outputted from the search word construction device 31 and uses the search word to make a relation. The database 6 is searched and the search result is recorded in the search result storage device 33. The related information database 6 inputs the search word output from the search device 32 of the related information acquisition unit 5, searches the information held in the database with the search word, and retrieves the search result as the related information. It returns to the search device 32 of the acquisition unit 5. Here, it is assumed that the related information database holds information such as the contents of the encyclopedia. Note that the content held in the related information database is not limited to text format information, but is audio format, still image format, moving image format, and / or composite format information thereof.

FIG. 6 is an example of a block diagram of the related information providing unit 7. As shown in the figure, the related information providing unit 7 includes a search result update detection device 34, a display 35, a provided result storage device 36, and a timer 37. Here, there are two methods of providing the dialog-related information to the user: a method of using the search result update detection device 34 and a method of using the timer 37.

First, a method for providing interactive information using the search result update detection device 34 will be described. In this case, the timer 37 is not used. The search result update detection device 34 monitors the search result storage device 33 of the related information acquisition unit 5, and when the search result of the search result storage device 33 is updated, inputs the search result from the search result recording device 33. , Recording a copy of the content in the provision result storage device 36,
Output to the display 35. Provision result storage device 36
Is a storage device for holding the dialogue related information provided to the user. The display 35 inputs the dialogue related information output from the search result update detection device 34 and provides it to the user.

Next, a method of using the timer 37 to provide dialogue information will be described. In this case, the search result update detection device 34 is not used. According to the default value set in the timer 37 (for example, every 39 seconds), the display 35 inputs the dialogue related information as the search result from the search result storage device 33 and provides it to the user.

Next, the processing procedure of the local dialogue processing apparatus 1 and the remote dialogue processing apparatus 8 will be described. Figure 7
6 is a flowchart showing a processing procedure of dialogue processing in the local dialogue processing apparatus 1 and the remote dialogue processing apparatus 8. As shown in the figure, the dialogue process includes an initialization step S1, a dialogue input step S2, and an important word extraction step S.
3, the important word transfer determination step S4, the important word transfer step S5, the related information acquisition step S6, the related information providing step S7, and the continuation determination step S8.

In the initialization step S1, each part or each device is initialized. For example, initialization is performed by adjusting the volume of the microphone of the dialogue input unit 2 (or 9) or setting the set value of the timer 37 of the related information providing unit 7 (or 14) to 30 seconds. In the dialogue input step S2, first, the microphone device 21 of the dialogue input unit 2 (or 9) inputs dialogue information in a voice format from the user,
Output to the dictation processing device 22. Next, the dictation processing device 22 converts the speech-type dialogue information input from the microphone device 21 into text-type dialogue information.

In the important word extraction step S3, the dialogue comparison device 23 of the important word extraction unit 3 (or 10) inputs the textual dialogue information output from the dictation processing device 22 of the dialogue input unit 2 (or 9). Then, the dialogue information recorded in the dialogue recording device 24 is compared, the corresponding dialogue information is extracted, and the extracted dialogue information is output to the morpheme analysis device 25. Further, the dialogue comparison device 23 records the extracted dialogue information in the dialogue recording device 24 as a log. Then, the morpheme analysis device 25 inputs the dialogue information in text format output from the dialogue comparison device 23, performs morpheme analysis on the input dialogue information with reference to the word dictionary 26, and outputs the morpheme analysis result to the important word. Output to the extraction device 27.

Then, the important word extracting device 27 counts the important words from the input words within a certain period of time or between the turning points of the dialogue (topic), outputs the counting result to the important word storage device 28, and calculates the importance degree. Outputs to the important word transfer unit 4 and the related information acquisition unit 5 an important word satisfying a specific criterion as a search word candidate. In addition, the important word storage device 28 uses the important word extraction device 2
The important word and the degree of importance (for example, the frequency of appearance) are input from 7 and held. Therefore, the extracted important word and its importance are stored in the important word storage device 28 every time, and are used for subsequent importance calculation.

In the important word transfer judgment step S4,
Whether or not the important word is transferred from the important word transfer unit 4 of the local interaction processing device 1 to the related information acquisition unit 12 of the remote interaction processing device 8 and / or the important word is important in the remote interaction processing device 8 It is determined whether or not the important word is to be transferred from the word transfer unit 11 to the related information acquisition unit 5 of the local dialogue processing apparatus 1. When transferring, the important word transfer step S5.
If the procedure is performed and the transfer is not performed, the procedure of the related information acquisition step S6 is performed. It should be noted that this determination is made based on the user setting regarding whether or not to share the dialogue information between the dialogue processing devices 1 and 8.

In the important word transfer step S5, first, the important word storage device 29 of the important word transfer unit 4 (or 11) is used.
Is an important word extraction device 27 of the important word extraction unit 3 (or 10).
In order to transmit the important word output from the local dialogue processing device 1 to the remote dialogue processing device 8 and conversely to the remote dialogue processing device 8 to the local dialogue processing device 1, Hold. Next, the important word transmission device 30 transmits the important word held in the important word storage device 29 from the local dialogue processing device 1 to the remote dialogue processing device 8. Alternatively, it is transmitted from the remote dialogue processing device 8 to the local dialogue processing device 1.

In the related information acquisition step S6, the search word construction device 31 of the related information acquisition unit 5 (or 12)
The important word which is the candidate of the search word output from the important word extraction unit 3 (or 10) is input. When it is determined in the important word transfer determination step S3 that the important word is to be transferred, the search word construction device 31 outputs the important word or the local dialogue output from the important word transfer unit 11 of the remote dialogue processing device 8 via the network 15. Important word transfer unit 4 of processing device 1
The important words output from the network via the network 15 are input.

The search word constructing device 31 constructs a search word to be used for searching the dialogue related information from the related information database 6 (or 13) based on the inputted important word. The search word construction device 31 outputs the constructed search word to the search device 32. The search device 32 inputs the search word output from the search word construction device 31, searches the related database 6 (or 13) using the search word, and records the search result in the search result storage device 33. To do. Here, the related information database 6 (or 13) inputs the search word output from the search device 32 of the related information acquisition unit 5 and searches the information held in the database using the search word. , And returns the search result to the search device 32 of the related information acquisition unit 5.

In the related information providing step S7, first, the search result update detecting device 34 of the related information providing unit 7 (or 14) is connected to the search result storage device 33 of the related information acquiring unit 5.
When the search result of the search result storage device 33 is updated, the search result is input from the search result recording device 33, a copy of the content is recorded in the provided result storage device 36, and output to the display 35. To do. Next, the display 35 inputs the dialogue related information output from the search result update detection device 34 and provides it to the user.

In the continuation judging step S8, it is judged whether or not the dialogue processing is continued. If the dialogue processing is continued, the processing from S2 to S7 is continued, and if the dialogue processing is stopped, , The interactive processing ends. It should be noted that this judgment is made based on the number of times of information provision and the information provision time set by the user.

Next, a method of calculating the importance degree between the important words from the appearance frequency of the important words and extracting the important words that are candidates for the search word for searching the dialogue related information according to the importance degree explain. This method includes a method of calculating the degree of importance between the important words according to a predetermined time and a method of calculating the degree of importance between the important words according to the conversion of the content of the dialogue.

First, a method of calculating the degree of importance between important words according to a predetermined time will be described. FIG. 8 shows that the similarity between important words is calculated from the appearance frequency of the important words according to a predetermined time, and the dialogue related information is calculated according to the importance of the important words based on the calculated similarity between the important words. The procedure of the method of extracting the important word which is a candidate of the search word for searching is shown.
This procedure includes initialization step S10, important word frequency calculation step S11, included angle calculation step S12, minimum included angle extraction step S13, corresponding important word selection step S14, search word candidate extraction step S15, continuation determination step S16.
And update step S17. Note that these procedures are executed by the important word extraction device 27 of the important word extraction unit 3 (or 10).

In the initialization step S10, a time interval for counting the frequency of important words is set. Here, this fixed time is called a slot. That is, setting of slots is performed as the initialization step S10. In the important word frequency calculation step S11, the appearance frequency of the important word is counted at regular intervals (for example, every 120 seconds) using a timer or the like, and the count result is held in the important word storage device 28. That is, the frequency of appearance of important words is counted in slot units. Here, the important word storage device 28 of the important word extracting unit 3 (or 10) holds the important words of the past slots (slots from slot 1 to slot 4), their frequencies and slot numbers as shown in Table 1. I shall. For example, it is assumed that the important words are calculated at the frequencies shown in Table 1 below in slots 1 to 4 that precede the current slot in time.

[0056]

[Table 1]

In the included angle calculation step S12, the importance of the important word is calculated. Therefore, first, the frequency (W1
To W4).

[0058]

[Equation 1]

Next, the sum vector W0 from W1 to W4 is given by Equation 2
Calculate as

[0060]

[Equation 2] Here, W0 = (6, 9, 5, 6).

From vector W0 and vector W1 respectively
Find the included angle up to W4 (the included angle between vector W0 and vector Wi is Ti. Here i = 1, 4). Here, for example, the included angle T1 formed by the vector W0 and the vector W1 is defined as the following Expression 3.

[0062]

[Equation 3] Here, cos indicates a cosine.・ Indicates dot product. | W0 | indicates the magnitude of the vector W0. | W1 | indicates the magnitude of the vector W1.

Each included angle is approximately as follows. T1 = 0.876 radians, T2 = 0.556 radians, T3 = 0.355 radians, T4 = 0.406 radians,

It can be said that the smaller the included angle, the higher the importance of the important word corresponding to the vector used in the calculation of the included angle, and the reciprocal of the value is regarded as expressing the importance.
That is, the included angle becomes smaller in the order of T3 <T4 <T2 <T1, and the important word "bottle" represented by the vector W3 is the vector W4.
Compared with the important word "France" expressed by, it can be regarded as important (greater importance).

In the minimum included angle extraction step S13, the included angle T1
Extract the vector related to the minimum value from to T4. That is, a vector expressing a set of important words having the highest importance is extracted. Here, T3 is the minimum value and the corresponding vector is the vector W3. In the corresponding important word selection step S14, the minimum included angle extraction step S13
Select the important word corresponding to the vector extracted in. The key word corresponding to the vector W3 is "bottle". Then, the important word "bottle" is selected.

In the retrieval word candidate extraction step S15, the important word extracted in the corresponding important word selection step S14 is extracted as a retrieval word candidate. That is, the important word “bottle” is used as a search word candidate. In the continuation step S16, it is determined whether or not to continue the procedure from the important word frequency calculation step S11 to the search word candidate extraction step S15.
If it is to continue, the subsequent updating step S17
The procedure is performed, and the processing is performed by using the important word and the importance thereof, which are extracted in the previous time and the importance thereof is calculated and accumulated in the storage device 28. If it does not continue, the process ends. In the updating step S17, the important word and its frequency are stored and the slot number is updated. For example,
The important word and its frequency are stored in the important word storage device 28 of the important word extracting unit 3 (or 10). If the slot number of the current slot is 5, the current slot number is incremented to 6.

Next, a method of calculating the degree of importance between important words according to the conversion of the content of the dialogue will be described. In the above description, the importance of a certain time unit (slot unit) was calculated, but the turning point of the dialogue is extracted and the candidate of the search word is extracted by calculating the importance of the dialogue unit in the same manner. be able to. As a technique for extracting the turning point of the dialogue, a conventional technique such as "dialogue processing device" described in Japanese Patent Laid-Open No. 8-137874 can be applied.

Further, instead of calculating the frequency (vector) of the important word for each fixed time (slot) unit or each turning point of the dialogue, the frequency of the important word is counted for each predetermined document unit and the value is calculated. It is also possible to refer to and extract the important word by the same method as described above. Further, in the above description, a method has been described in which only the important words corresponding to the vector having the smallest included angle are candidates for the search word having a high degree of importance, but from the important word corresponding to the vector having a small included angle in order. It is also possible to select a plurality of important words from (important words of high importance) and use them as search word candidates.

Here, the method of outputting the dialog related information to the display 35 in the related information providing step S7 will be described in detail. As a method of outputting this dialogue related information, FIG. 9 shows a method for sequentially displaying the dialogue related information on the display 35 and a plurality of dialogue related information recorded in the provision result storage device 36 as described above. Thus, there is a method of simultaneously displaying in different display areas.

When a plurality of dialogue-related information items are simultaneously displayed in the different display areas, a larger display area can be assigned to a new dialogue-related information item when the importance is calculated for each slot. For example, interaction-related information for slot 5 (the current slot is referred to as slot 5) is allocated 50% of the display area, and interaction-related information for slots 1 through 4 is 12.5% of the display area. You can assign each one. Further, when the dialog related information of past slots is simultaneously displayed on different display areas, the display area can be assigned according to the calculated included angle in the process of calculating the importance.

Also, when displaying sequentially in time, it is possible to display at a time corresponding to the reciprocal of the included angle. For example, when extracting candidate search words, for example, when an important word is selected in the corresponding important word selection step S14,
Since the reciprocal of the included angle T3 is about 2.8, it is possible to display the dialogue-related information that is the search result for only 2.8 seconds. In the above description, the related information database 6 (or 13) is the local dialogue processing device 1
Alternatively, the configuration is described as being configured inside the remote interaction processing device 8, but it is also possible to configure to use one or more external databases via the network 15. For example, a so-called search engine provided on the World Wide Web can be used as the related information database 6 (or 13).

Next, a second embodiment according to the present invention will be described. FIG. 10 is a configuration example of the important word transfer unit 4 (or 11) of the second embodiment. The important word transfer unit 4 of the second embodiment is the important word transfer unit 4 (or 1) described in the first embodiment.
The important word translation device 38 is added to 1), and is composed of an important word storage device 29, an important word translation device 38, and an important word transmission device 30.

The important word translation device 38 converts the important word into the first word.
Provides a function of translating from the first language into a second language different from the first language. For example, when the user of the local dialogue processing apparatus 1 is a person whose native language is Japanese and the user of the remote dialogue processing apparatus 8 is a person whose native language is English, the important word transfer unit in the local dialogue processing apparatus 1 is used. The important word translation device 38 in 4 provides a function of converting from Japanese to English, and the important word transfer unit 11 in the remote dialogue processing device 8 is provided.
On the contrary, the important word translation device 38 provides a function of converting from English to Japanese, so that it is possible to provide the dialogue related information corresponding to both languages to the users in various places.

Next, a third embodiment according to the present invention will be described. In this example, as shown in FIG. 11, in the configuration of the local dialogue processing apparatus 1 and the remote dialogue processing apparatus 8 of the first embodiment, the user information input unit 39 and the environment information input unit with respect to the local dialogue processing apparatus 1 are used. 40 is added, and a user information input unit 41 and an environment information input unit 42 are also added to the remote dialogue processing device 8.

An application example of using the user information and environment information will be described below. FIG. 12 is a block diagram of the user information input unit 39 (the same applies to the user information input unit 41). The user information input unit 39 includes a user information input device 43 and a user information storage device 44. FIG. 13 is a block diagram of the environment information input unit 40 (the same applies to the environment information input unit). The environment information input unit 39 includes an environment information input device 45 and an environment information storage device 46.

The user information input device 43 inputs information characterizing each user such as the user's user name, affiliation, and preference from the user, and gives the user identifier to the input user information. , Is stored in the user information storage device 44 for each user. FIG. 14 shows an example of the data format of user information. The environment information input device 45 uses the above-mentioned interlocutor identifier and the place where the user interacts with the environment information input unit 45.
Information about the environment in which a conversation is being performed, such as the conversation time, is input, and an environment identifier is given to the input environment information and stored in the environment information storage device 46. Figure 15
Shows an example of the data format of environmental information.

The search word constructing device 31 of the related information acquisition unit 5 (or 12) inputs the user information and / or the environment information from the user information storage device 44 and / or the environment information storage device 46, and the dialogue related information A search term for searching can be constructed. For example, when "Chardonnay" is set as the user's preference information, the important word input from the important word extracting unit 3 (or 10) is "wine".
And "France", it is possible to construct "(wine) and (France) and (Chardonnay)" as a search term. By using the user information in this way, it is possible to search and provide the user with customized dialogue-related information.

Next, an application example using the environment information will be described. First, as shown in FIG. 16, the search word construction device 31 of the related information acquisition unit 5 (or 12) adds a dialog history item in addition to the user information data format of FIG. It is assumed that the search term constructed by the user's dialogue and the dialogue time as the environmental information are recorded.

FIG. 17 shows a flow chart of the environmental information utilization processing procedure. The environmental information use procedure includes an initialization step S20, a past environment information acquisition step S21, a user identification step S22, and a dialogue history identification step S23.
And a search word constructing step S24. In the initialization step S20, the related information acquisition unit 5 (or 12)
The search word configuration device 31 of acquires the current time. Further, the current environment information is input from the environment information storage device 48 of the environment information input unit 40 (or 42), and the dialogue time is specified from the input environment information.

In the past environmental information acquisition step S21,
Search word construction device 31 of related information acquisition unit 5 (or 12)
Inputs from the environment information storage device 48 of the environment information input unit 40 (or 42) the past environment information in which the interaction time and the interaction time specified in the initialization step S20 match the current time. This environmental information will be referred to as past environmental information. In the user identification step S22, the related information acquisition unit 5
(Or 12) The search word configuration device 31 identifies the interlocutor identifier from the past environment information, and uses the extended user information matching the interlocutor identifier to use the user information input unit 39 (or 41). Input from the person storage device 44.

In the dialog history specifying step S23, a search word of the dialog history is input from the expanded user information input in the user specifying step S22. In the search word construction step S24, a search word for searching the dialogue related information is constructed using the search word input in the dialogue history specifying step S23. By using the environment information as described above, it is possible to provide the user with the dialog-related information corresponding to the environment (the dialog time in the above description).
That is, it is possible to provide the user with information related to the conversation content that was spoken at the same time (time) in the past.

In the above description of each embodiment, the case where the local dialogue processing device 1 and the remote dialogue processing device 8 are respectively single has been explained, but as shown in FIGS. 18, 19 and 20, one The local interaction processing device 1 and / or one or more remote interaction processing devices 8 described above may be connected via the network 15. In this case, the transfer of important words between the local interaction processing device and the remote interaction processing device can be performed using the multicast technique.

Further, FIG. 21 shows a user interface for displaying the dialogue related information of the present invention when the local dialogue processing apparatus 1 and the remote dialogue processing apparatus 8 are individually connected via the network 15. For example: As shown in the figure, the user is the related information providing unit 7
The display 35 (or 14) can be used to view both information related to local and remote interactions.

Further, in FIG. 22, for displaying the dialogue related information of the present invention when the local dialogue processing apparatus 1 is single and the two remote dialogue processing apparatuses 8 are connected via the network 15. An example of the user interface of As shown in the figure, the user is the related information providing unit 7
Using display 35 (or 14), it is possible to view three pieces of information related to the interaction of the two local and remote. It should be noted that a method for providing information related to local and / or remote interactions by dividing the area for providing interaction-related information (user interface)
However, it is also possible to provide information related to local and / or remote interaction in a temporally segmented manner.

[0085]

As described above, according to the present invention,
Information related to the dialogue can be provided to the interlocutor to promote the dialogue. Further, according to the present invention, it is possible to provide the related information related to the local and remote interactions locally or / and remotely, synchronously or asynchronously, and the interaction between remote users can be synchronously or asynchronously. Be promoted.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of a dialogue processing device of the present invention.

FIG. 2 is a block diagram showing a configuration example of a dialogue input unit of the present invention.

FIG. 3 is a block diagram showing a configuration example of an important word extraction unit of the present invention.

FIG. 4 is a block diagram showing a configuration example of an important word transfer unit of the present invention.

FIG. 5 is a block diagram showing a configuration example of a related information acquisition unit of the present invention.

FIG. 6 is a block diagram showing a configuration example of a related information providing unit of the present invention.

FIG. 7 is a flowchart showing an example of a dialogue processing procedure of the present invention.

FIG. 8 is a flowchart showing an example of a retrieval word candidate extraction procedure based on importance according to the present invention.

FIG. 9 is a diagram showing an example of a user interface for displaying interaction-related information of the present invention.

FIG. 10 is a block diagram showing another configuration example of the important word transfer unit of the present invention.

FIG. 11 is a block diagram of a dialogue processing apparatus according to another embodiment of the present invention.

FIG. 12 is a block diagram showing a configuration example of a user information input unit of the present invention.

FIG. 13 is a block diagram showing a configuration example of an environment information input unit of the present invention.

FIG. 14 is a diagram showing an example of a data format of user information of the present invention.

FIG. 15 is a diagram showing an example of a data format of environment information of the present invention.

FIG. 16 is a diagram showing an example of extended user information of the present invention.

FIG. 17 is a flowchart showing a procedure for applying environment information of the present invention.

FIG. 18 is a diagram showing a connection example of a local dialogue processing device and a remote dialogue processing device of the present invention.

FIG. 19 is a diagram showing another connection example of the local dialogue processing device and the remote dialogue processing device of the present invention.

FIG. 20 is a diagram showing another connection example of the local dialogue processing device and the remote dialogue processing device of the present invention.

FIG. 21 is a diagram showing an example of a user interface for displaying dialogue related information of the present invention.

FIG. 22 is a diagram showing another example of a user interface for displaying dialogue-related information of the present invention.

[Explanation of symbols]

1: local dialogue processing device, 2, 9: dialogue input unit,
3, 10: important word extraction unit, 4, 11: important word transfer unit,
5, 12: Related information acquisition unit, 6, 13: Related information database, 7, 14: Related information providing unit, 8: Remote dialogue processing device, 15: Network,

Claims (14)

[Claims] 1. A dialogue processing device for providing information related to a topic to a speaker, a dialogue input means for inputting dialogue information spoken by a speaker, and an important word for extracting an important word from the inputted dialogue information. Extraction means, related information acquisition means for acquiring related information related to the dialogue from the related information database using the extracted important words, and information providing means for providing the acquired related information to the speaker. And a dialogue processing device. 2. The dialogue processing apparatus according to claim 1, wherein the dialogue processing apparatus is connected to another dialogue apparatus having a similar function means via a network, and at least one of the dialogue processing apparatuses is provided with the above-mentioned A dialogue processing device, comprising: important word transfer means for transferring the important word extracted by the important word extraction means to another interaction processing device for use by the related information acquisition means. 3. The dialogue processing apparatus according to claim 2, wherein the first dialogue processing apparatus and the second dialogue processing apparatus each including the important word transfer means are connected via a network, The dialogue processing device inputs the dialogue information input by the user from its own dialogue input means, extracts the important word from the dialogue information by its own important word extraction means, and sets the extracted importance to the own important word. The important word transfer means of the first dialogue processing device outputs the extracted important word to the relevant information acquisition means of the second dialogue processing device via the network. The relevant information acquisition means of the first dialogue processing device uses the important word input from the important word extraction means of the first dialogue processing device and / or the important word transfer means of the second dialogue processing device as a search word. Related Information Day Acquires related information related from the base to the dialogue, the dialogue processing apparatus characterized by providing the user with related information providing unit of the first dialogue processing apparatus. 4. The dialogue processing apparatus according to claim 1, wherein the important word extraction unit has a morpheme analysis unit, and stores the dialogue information input from the dialogue input unit. Performs morphological analysis, selects important words related to dialogue information input from morphological analysis results, calculates importance between important words from appearance frequency of important words, and retrieves information related to dialogue according to the importance. A dialogue processing device characterized by extracting an important word as a candidate of a search word for performing. 5. The dialogue processing device according to claim 4, wherein the important word extraction means is in accordance with a predetermined time, or
A dialogue processing device, characterized in that the degree of importance between important words is calculated according to the conversion of the content of the dialogue. 6. The dialogue processing device according to claim 5, wherein the important word extracting means is a candidate of a search word for searching related information related to the dialogue based on the calculated importance of the important word. The important word extracted as is stored in the important word storage means for calculating the importance degree thereafter, and the important word and its frequency are used in the next importance degree calculation. Interaction processing device. 7. The dialogue processing apparatus according to claim 1, wherein the related information providing unit provides the user with a plurality of related information input from the related information acquisition unit. An interactive processing device characterized by being provided sequentially or arranged side by side on a screen at the same time. 8. The dialogue processing apparatus according to claim 5, wherein the related information providing unit provides the user with information according to the importance of the extracted important word. A dialogue processing device, characterized in that the provision time of the relevant information or the size of the provision area is changed. 9. The dialogue processing device according to claim 2, wherein the important word transfer means changes the important word from a first language to a second language different from the first language. A dialogue processing device having an important word translation means for translating and converting into a language. 10. The dialogue processing apparatus according to claim 1, further comprising: user information input means for inputting user information regarding characteristics of the user, wherein the related information acquisition means is A dialogue processing device, characterized in that the relevant information related to the dialogue is searched from the relevant information database by using the important word and the user information input from the user information input means. 11. The method according to any one of claims 1 to 10.
In the dialogue processing device according to the item, an environment information input unit for inputting environment information about an environment for dialogue is provided, and the related information acquisition unit uses the environment information input from the important word and the environment information input unit, A dialogue processing device, which searches for relevant information related to a dialogue from a relevant information database. 12. A program for realizing by a computer the function of providing information related to a topic to a speaker, the function of extracting an important word from dialogue information spoken by a speaker, and the extracted important A program that realizes by a computer a function of acquiring related information related to a dialogue from a related information database using words and a function of providing the acquired related information to a speaker. 13. A method of providing information related to a topic to a speaker, wherein the conversation information spoken by a speaker is acquired by a microphone installed in a place where the conversation is performed, and the acquired conversation information is acquired. The important word is extracted from the related information, the related information related to the dialogue is acquired from the related information database using the extracted important word, and output from the related information providing means installed in the place where the dialogue is performed. An interactive processing method characterized by performing a series of computer processes. 14. The dialogue processing method according to claim 13, wherein in the extraction of the important words, the acquired dialogue information is subjected to morphological analysis, and the important words relating to the dialogue information input from the morphological analysis result are selected. The importance between the important words is calculated from the frequency of occurrence of the words, and the important words are extracted as candidates for the search word for searching the information related to the dialogue according to the importance. In the process of calculating the importance between
An important word frequency calculation step for calculating the frequency of important words according to a predetermined time or according to a turning point of the content of the dialogue, and a vector for the frequency is constructed from the calculated frequencies of the important words, and An included angle calculation step of calculating an included angle, a minimum included angle extraction step of extracting the minimum included angle as the importance of the corresponding important word is larger as the calculated included angle is smaller, and this minimum included angle is calculated. Corresponding important word selection step of selecting a corresponding important word from the vector used for, and a different important word extracting step of extracting an important word different from the contrasting important word by comparing with the corresponding important word selecting step, and And a step of extracting a search word candidate as a search word candidate for searching information related to a dialogue using the important word contrasted with different important words. Processing method.