JP2003263192A - Information processing system, its method and information processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain an information processing system capable of improving processing performance in the information processing combined by all information processors. <P>SOLUTION: The information processing system 100 is provided with an input device 101 for receiving data input, first and second information processors 103<SB>1</SB>, 103<SB>2</SB>for generating outputs and an output device 105 for outputting the processed results of a voice, character strings, etc. When the first information processor 103<SB>1</SB>executes processing, the processed result is stored in an input history storage part 106 by an input history management part 108, and when the second information processor 103<SB>2</SB>executes processing, domain information stored in a domain knowledge storage part 107 is retrieved/extracted by using the processed result. Consequently the processing of the second information processor 103<SB>2</SB>is made suitable and efficient. <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 an information processing system, an information processing method, and an information processing program, and more particularly to specific information processing such as voice recognition and handwritten character recognition and other information processing related to this information processing. The present invention relates to an information processing system, an information processing method, and an information processing program capable of improving information processing when performing.

[0002]

2. Description of the Related Art In a voice recognition system that recognizes an input voice and outputs the recognition result as a character string, the voice is not recognized by using only the input data of the voice to be recognized but becomes the recognition target. It is usual to perform voice recognition processing while referring to voice data acquired in advance for a person. Similarly, a machine translation system that converts the input source language character string into the target language character string and outputs it, a voice synthesis system that converts the input character string into voice and outputs it, or refer to the input search request. Even in an information retrieval system that retrieves and outputs appropriate information from a database, it is usual to analyze input data based on reference information that is stored in advance and output the analysis result and newly generated information. .

In such an information processing system, although information processing is performed using reference information, errors or difficulties may occur during analysis or new information generation due to the variety, complexity or ambiguity of input data. Often occurs.
A character recognition device that recognizes handwritten characters has the same problem. Therefore, various methods for improving the processing when these information processing systems are actually used have been conventionally proposed.

FIG. 24 shows an example of a conventionally proposed information processing system. JP 2001-1011A
This system disclosed in Japanese Patent Publication No. 87 includes a voice recognition unit 12 that inputs a voice 11 and recognizes the voice 11, a machine translation unit 13 that performs machine translation, and a voice synthesis unit 15 that outputs a recognized voice 14. In addition, a display unit 16 and a dialogue management unit 17 are provided.

The voice recognition unit 12 performs voice recognition of the voice 11 that is interactively input, and supplies a voice recognition result 21 to the machine translation unit 13. In the machine translation unit 13, the voice recognition unit 1
The speech recognition result 21 from 2 is machine translated. When a dialogue between a Japanese user and an English user is being performed using this system, the dialogue management unit 17 stores the dialogue history, and the dialogue history and the Japan stored in advance in the dialogue history. Based on the knowledge of words and English, for example, it is determined whether or not the translation result of the machine translation unit 13 is unnatural from the flow of dialogues performed by two users. If the translation result is unnatural, the dialogue management unit 17
The fact that the translation result is unnatural is output to the machine translation unit 13. In this case, the machine translation unit 13 will re-translate based on the flow and knowledge of the dialogue between the two people.

As described above, in the prior art shown in FIG. 24, the performance of the information processing system is based on the input history which is the conversation history of two users who speak Japanese or English using this information processing system. We are trying to improve

[0007]

In the prior art shown in FIG. 24, the three main parts of the speech recognition unit 12, machine translation unit 13, and speech synthesis unit 15 are called Japanese to English or English to Japanese. Such a voice translation process 1
In charge of information processing. Then, in order to improve the accuracy of this one information processing, the dialogue management unit 17 uses the flow and knowledge of the dialogue between the two persons.

However, the processing that a person requests an information processing system such as a computer is not always such a single processing, but there are many cases where a human being requests information processing combined in parallel at the same time. For example, two or more types of information processing such as a process of simply converting a voice input in Japanese into English and a process of searching for an example sentence based on the voice input in Japanese and outputting this in voice in English. In many cases, actual processing is required in a combined form.

However, conventionally, in order to perform such complex processing, a dedicated information processing device for each processing or a processing program (software) specialized for each processing is started up and these processings are performed. It has been performed that an operator switches and uses each processing program or individually starts and uses each processing program each time. As a first example, consider an information processing system using integrated software for creating a New Year's card. In order to create a New Year's card, you can enter the name of the recipient or create a text using a picture or photo. Therefore, the software for creating a normal New Year's card comes with some software such as address book, digital camera editing software, painting software, calendar software, and map creation software. However, these pieces of software are basically independent, and the data processed by one piece of software is simply made available to other bundled pieces of software.

Next, as another example of the conventional information processing system using this kind of integrated software, the case of using office software will be described. This type of office software is often composed of word processing software, spreadsheet software, and a database. In the database software of office software, a plurality of attributes are input as data for each entry. As a specific example, it is assumed that there is a database in which a customer name, contact information of the customer, past transaction records, and the like are put together in one sheet, and the sheet has a plurality of sheets corresponding to the number of customers. At this time, by using the linkage function between the database and the spreadsheet software, each row represents each customer on the spreadsheet information of the database, the first column is the customer name, and the second and subsequent columns are the contact information of the customer. It can be created by a list showing attributes such as or transaction history in the past. Also, if you create a graph of transaction results for each customer from the table on this spreadsheet software, insert this graph into the document being created by the word processing software by using the linkage function of the spreadsheet software and word processing software. You can In such office software, a cooperation function between a plurality of software is incorporated, but the input data input to each information processing device can be used in another information processing device. As such, the compatibility of data formats is simply maintained.

As described above, in the conventional integrated software of this type, the cooperation between the software is insufficient, and the reality is that the cooperation depends on the operator. For example, taking the former integrated software for creating New Year's cards as an example, digital camera editing software has the function of only editing images taken with a digital camera, and one image data is received by software for creating text. It only has the function of passing. There is no cooperation such as automatically selecting a sentence that matches the content of the image taken by the digital camera, or selecting the most suitable sentence or typeface for the text from the content of the edited address book.

[0012] Therefore, if the conventional information processing system is made to perform a complex process, the system operator or the operator of the apparatus is in charge of coordinating a plurality of these processes, and that person has a lot of problems. It was a burden. Therefore, the smooth operation of the information processing system is hindered, the speed of the information processing is delayed, and the operator's knowledge of each process and the skill level of cooperation make a great difference in the information processing result. It was

Therefore, an object of the present invention is to provide an information processing system capable of effectively performing several different related information processes.

[0014]

According to a first aspect of the present invention, (a) first information processing means for inputting predetermined information to perform first information processing; and (b) this first information processing means. Second information processing means for performing second information processing different from the processing of the processing means; and (c) domain information storage in which domain-specific knowledge used for information processing of the second information processing means is stored as domain information. Means, (d) a field selecting means for selecting a field of the domain information stored in the domain information storing means based on the processing result of the first information processing means, and (e) selected by the field selecting means. The information processing system is provided with a specific domain information supply means for supplying the domain information of the specific field for the information processing of the second information processing means.

That is, according to the first aspect of the invention, the field of the domain information stored in the domain information storage means is selected based on the processing result of the first information processing means and the second information processing means is processed. The domain information used is limited so that the processing can be performed effectively.

According to the second aspect of the invention, (a) the first information processing means for inputting predetermined information to perform the first information processing, and (b) the processing of the first information processing means. Second information processing means for performing different second information processing; and (c) domain information storage means for storing domain-specific knowledge used for information processing of the second information processing means as domain information,
(D) first history storage means for storing a first history as a processing history of the first information processing means, and (e) this first history storage means.
Field selecting means for selecting the field of the domain information stored in the domain information storing means based on the first history stored in the history storing means, and (f) the domain of the specific field selected by the field selecting means. The information processing system is provided with a specific domain information supplying means for supplying information for the information processing of the second information processing means.

That is, according to the second aspect of the invention, the first history as the history of the processing of the first information processing means is stored, and the second information processing means performs the information processing using the domain information. The field of domain information at that time is selected based on the first history. Thereby, the processing can be effectively performed by limiting the domain information used for the processing of the second information processing means.

According to the third aspect of the invention, (a) a plurality of information processing means for inputting and processing information, and (b)
History storage means for storing information handled by any of the plurality of information processing means as a history; and (c) domain information storage means for storing domain information used for information processing of the plurality of information processing means, D) History extraction means for searching the history stored in the history storage means and extracting a specific history; and (e) Among the domain information stored in the domain information storage means using the history extracted by the history extraction means. Domain information searching / extracting means for searching for and extracting necessary domain information from (4), and (f) when one of the plurality of information processing means described above carries out information processing, the identification extracted by the domain information searching / extracting means The information processing system is provided with a specific domain information supply means for supplying the domain information of (3) as a reference for processing.

That is, according to the third aspect of the invention, when a plurality of information processing means perform their own information processing,
A history when any one of the information processing means performs information processing is stored, and when other information processing means performs information processing, necessary domain information is searched and extracted from this history, and this is used as a reference for processing. I am trying to supply it as. This allows each information processing unit to select and provide the domain information required when processing the information, so that the processing can be effectively performed.

According to a fourth aspect of the present invention, in the information processing system according to the first or second aspect, the field selecting means includes a similarity calculating means for calculating the similarity between the first history and each field. It is characterized by selecting the highest field among the similarities calculated by the similarity calculating means.

That is, in the invention according to claim 4, the field of domain information is selected using the similarity.

In the invention described in claim 5, (a) first information processing means for inputting voice information to perform speech translation processing, and (b) processing different from the processing of the first information processing means. Second information processing means for performing example sentence translation processing as
(C) A domain information storage unit that stores, as domain information, an example sentence for each scene used for translating the example sentence of the second information processing unit, and (D) a first history as a processing history of the first information processing unit. A first history storage means for storing
(E) Scene selection means for selecting a scene to which the example sentence is applied based on the first history stored in the first history storage means, and (f) Domain information of the scene selected by the scene selection means. And a specific domain information supplying means for supplying as a candidate for example sentence translation in the information processing system.

That is, the invention of claim 5 cites the first example of a specific information processing system. In this example, the first information processing means performs speech translation processing, and the second information processing means performs example sentence translation processing. When a voice is input and the first information processing unit performs a voice translation process, a first history as a history of the process is stored in the first history storage unit. When performing the example sentence translation process by the second information processing means, the scene to be the target of the example sentence is selected based on the first history, so that the domain information of the scene is cited as a candidate for the example sentence translation. Therefore, for example, when the first information processing unit is performing the translation process for shopping, the example sentence translation of the scene related to shopping is also performed, and the most appropriate process is performed by the second information processing unit. It will be possible.

According to the invention of claim 6, (a) first information processing means for inputting a position and extracting geographical information as information corresponding to the input position; and (b) this first information. Second information processing means for recognizing visual information input as processing different from the processing of the processing means, and (c) recognition of auditory information as processing different from the processing of the first and second information processing means. A third information processing unit, (d) a domain information storage unit that stores information for each scene used for recognition of visual information by the second information processing unit as domain information, and (e) a first information processing unit. First history storage means for storing a first history as a processing history, and (f) the first history stored by the first history storage means is supplied as domain information when recognizing visual information. First domain information supplier to do If, (g) second storing second and history storage unit, (viii) the second history storage means for storing second history as a history of the processing of the second data processing means
The information processing system is provided with a second domain information supply means for supplying the history of (1) as domain information when recognizing auditory information.

That is, in the invention described in claim 6, the second example of the concrete information processing system is given. In this example, the first information processing means performs processing for extracting geographical information, the second information processing means performs processing for recognizing visual information such as characters and images, and the third information processing means performs processing for recognizing voice information. Performs processing to recognize auditory information. A first history as a processing history of the first information processing means and a second history as a processing history of the second information processing means are stored in predetermined storage areas. Therefore, the recognition processing of the visual information by the second information processing means can be efficiently performed using the first history, and the auditory information of the third information processing means can be processed by using the second history. The recognition process can be performed efficiently. Depending on the system, the first history can be used for the processing of the third information processing means, and the second history can be used for the processing of the first information processing means.

In the invention according to claim 7, (a) first information processing means for analyzing the contents of the chat, and (b) appropriate advertisement as processing different from the processing of the first information processing means. Second information processing means for performing advertisement display processing to be displayed; (c) domain information storage means for storing advertisements to be displayed by the second information processing means as domain information for each field; and (d) first A first history storing means for storing a first history as a processing history of the information processing means of
(E) Type selecting means for selecting the type of advertisement based on the first history stored in the first history storing means, and (f)
The information processing system is provided with specific domain information supply means for supplying the type of domain information selected by the type selection means as an advertisement candidate.

That is, the invention of claim 7 cites the third example of a specific information processing system. In this example, the first information processing unit performs the process of analyzing the content of the chat, and the second information processing unit performs the advertisement display process for the person having the chat. Since the first history as the processing history of the first information processing means is stored in the first history storage means, by using this to select the domain of advertisement of the domain information, it is possible to relate to chat topics and the like. It is possible to efficiently display the created advertisement.

In the invention described in claim 8, (a) the first information processing step of inputting predetermined information to perform the first information processing, and (b) the first information processing step. A second information processing step of performing a second information processing different from the processing, and (c) knowledge of each field used for the information processing in the second information processing step is obtained in the first information processing step. A search step to search using history information,
(D) The information processing method is provided with a specific domain information supplying step of supplying the knowledge of the field searched in this searching step to the processing in the second information processing step.

That is, in the invention described in claim 8, knowledge of each field used for information processing in the second information processing step is searched by using the history information obtained in the first information processing step, and obtained by the search. By supplying the knowledge of the specified field to the processing in the second information processing step, the second information processing can be effectively performed using the domain information in which the application field is limited.

In the invention according to claim 9, (a) a history storing step of storing a history of processing when the information processing is performed by any of the information processing means, and (b) other than the information processing performed When other information processing is performed by the information processing means, a search step for searching a history related to the current information processing from the processing history stored in the history storage step, and (c) one used for the current information processing Specific domain information extraction step of extracting specific domain information in relation to the history searched in this search step from the domain information prepared in advance, and (d) identification extracted in this specific domain information extraction step The information processing method includes an information processing step of performing information processing this time using the domain information.

That is, in the invention according to claim 9, when information processing is performed by any of the information processing means, a history of the processing is stored, and other information processing means performs other information processing. At this time, the history related to the information processing of this time is searched from the processing history stored in the history storage step, specific domain information is extracted in relation to the searched history, and the information processing of this time is performed using this. I try to be able to do it efficiently.

In the information processing program according to the tenth aspect of the present invention, the computer is provided with (a) first information processing for processing input of predetermined information, and (b) a second information processing different from the first information processing. And (c) domain information extraction processing for extracting specific domain information from domain information as knowledge of each field used for the second information processing based on the result of the first information processing, (D) The specific domain information supply process of supplying the domain information extracted by the domain information extraction process for the second information processing is executed.

That is, according to the invention of claim 10, the first and second information processing programs are stored in the computer.
Information processing, and domain information extraction processing for extracting specific domain information from domain information as knowledge of each field used for the second information processing based on the result of the first information processing, and this domain information A specific domain information supply process for supplying the domain information extracted by the extraction process for the second information processing is executed so that the second information processing can be efficiently performed using appropriate domain information. ..

According to the information processing program of the invention described in claim 11, in the computer, (a) a first information processing for processing input of predetermined information and (b) a second information processing different from the first information processing. Information processing, (c) history storage processing for storing the history of the first information processing, and (d) knowledge by field used for the second information processing based on the history stored by this history storage processing. Domain information extraction processing to extract specific domain information from the domain information as
It is characterized in that the specific domain information supply processing for supplying the domain information extracted by the domain information extraction processing for the second information processing is executed.

That is, according to the invention of claim 11, the first and second information processing programs are stored in the computer.
Information processing, and domain information extraction processing for extracting specific domain information from domain information as knowledge of each field used for the second information processing based on the history of the first information processing, and this domain information A specific domain information supply process for supplying the domain information extracted by the extraction process for the second information processing is executed so that the second information processing can be efficiently performed using appropriate domain information. .

According to the information processing program of the invention as defined in claim 12, (a) a plurality of information processing for independently processing information and (b) when any one of these information processing is performed, Based on the history storage processing that stores the processing history, and (c) the history stored by this history storage processing, a specific domain is selected from the domain information as the knowledge of each field to be used in the next information processing. It is characterized by executing domain information extraction processing for extracting information and (d) specific domain information supply processing for supplying the domain information extracted by this domain information extraction processing for information processing.

That is, according to the twelfth aspect of the invention, a computer as an information processing program is caused to perform a plurality of information processes for independently processing information, and a history of the processes is stored when these processes are performed. Perform history storage processing. Then, based on the history stored by the history storage processing, specific domain information is extracted from the domain information as the field-specific knowledge to be used in the next information processing, and the extracted domain information is used as the information. By supplying for processing, it is possible to improve the efficiency of information processing by using appropriate domain information.

According to the invention of claim 13, a plurality of computers for executing the information processing program according to claims 10 to 12 are provided, and each computer executes independent information processing. I am trying.

That is, the invention according to claim 13 shows the case where the computer which executes the information processing program according to claims 10 to 12 is independent for each information processing. One computer may perform a plurality of information processes in a time division manner, or a part of the plurality of computers may perform a part of the plurality of information processes in an overlapping manner.

According to the invention of claim 14, in the information processing program according to claim 11 or 12, the domain information extracting process retrieves a desired history from the history stored in the history storing process. It includes a history search / extraction process to extract and extract, and a domain information search / extract process to search and extract specific domain information from the domain information based on the history extracted by this history search / extract process. It is characterized by that.

That is, the invention of claim 14 shows a specific example of the domain information extraction processing. In this example,
A history search / extraction process for searching and extracting a desired history from the history stored in the history storage process, and this history search / extraction process.
Domain information search / extraction processing is performed in which specific domain information is searched and extracted from the domain information based on the history extracted by the extraction processing.

According to the invention of claim 15, the information processing program according to claim 14 is used to search for domain information.
The extraction process is characterized by searching and extracting specific domain information based on the similarity between the history and the domain information.

That is, according to the fifteenth aspect of the invention, the history and the specific domain information are associated with each other using the similarity.

[0044]

DETAILED DESCRIPTION OF THE INVENTION

[0045]

EXAMPLES The present invention will be described in detail below with reference to examples.

<First Embodiment>

FIG. 1 shows the overall configuration of an information processing system according to the first embodiment of the present invention. The information processing system 100 includes an input device 101 that receives inputs such as voice and type, and a first device that processes input data 102 input from the input device 101 to generate an output.
And the second information processing devices 103 ₁ and 103 ₂ and the output device 105 that receives the output data 104 from the first and second information processing devices 103 ₁ and 103 ₂ and outputs the output data 104 as a processing result such as a voice or a character string. Is equipped with.

This information processing system 100 is used by both the input history storage unit 106 for storing the history of the input data 102 received by the input device 101 and the first and second information processing devices 103 ₁ and 103 _2. It also includes a domain storage unit 107 that stores the knowledge that can be obtained. The input history management unit 108 stores the input data 102 in the input history storage unit 106, searches the input data 102, and extracts necessary data. The domain knowledge management unit 109 is adapted to search and extract data stored in the domain knowledge storage unit 107.

The outline of the operation of the information processing system 100 of this embodiment will be described below. The input device 101 is composed of, for example, one or a combination of a voice input device, a type input device, a mouse input device as a pointing device, or a touch panel input device using a pressure sensor. Then, depending on the form of the input, the input data 102 such as the voice by the utterance input, the character string by the type input, the position information by the mouse input, the position information touched by the fingertip, or the like is input to the first or second information processing device 103 ₁ , 103. Will be sent to ₂ .

These first or second information processing device 10
3 _1, 103 _2, in addition to these input data 102, receives the auxiliary information such as input history 111 and domain knowledge 112 respectively from the input history management unit 108 and the domain knowledge management section 109. Then, based on these auxiliary information, the input data 102 is analyzed and the appropriate output data 104 is generated. The input data 102 of the first or second information processing device 103 ₁ or 103 ₂ and the analysis result of these are sent to the input history management unit 108 and the domain knowledge management unit 109. Also, the output data 104 generated by the first or second information processing device 103 ₁ , 103 _2.
Is an input history management unit 108 and a domain knowledge management unit 109.
And output device 105.

The first or second information processing device 103 ₁ ,
103 The input data _{102 2} receives the voice as described above, there are data such as strings or position information, as the output data 104 to be sent to the output device 105,
There are data such as voices, character strings, and image information. The auxiliary information received from the input history management unit 108 and the domain knowledge management unit 109 includes the input history and its analysis result, the output data 104 generated by the first or second information processing device 103 ₁ or 103 ₂ , and the input history. There are related domain information and data indicating the degree of association. Also, the first or second
The information processing apparatus 103 of _1, 103 ₂ are input history management unit 10
8 and the data transmitted to the domain knowledge management unit 109 include the input data 102, the analysis result thereof, the output data 104 generated by the first or second information processing device 103 ₁ , 103 ₂ , and the input history. These first or second information processing apparatus 103 _1, 103 _2, when a user performs an input from the input device 101, executes a predetermined process by calling the respective units described above in a predetermined order.

The output device 105 constituting the information processing system 100 is constituted by one of a voice output device for outputting a voice, a display device for displaying image information such as characters, or any combination thereof. To be done. The output device 105 outputs the output data 104 sent from the first or second information processing device 103 ₁ or 103 _2.
Based on, the data such as voice, character string, and image information is output.

On the other hand, the input history management unit 108 manages the input history used by the first or second information processing device 103 ₁ , 103 ₂ during information processing. As an example, the input history is stored and searched. And means for extraction. When the input history management unit 108 receives a request for searching or extracting an input history from the first or second information processing device 103 ₁ or 103 ₂ , the input history management unit 108 extracts an input history satisfying the request from the input history storage unit 106. Then, it returns this to the one of the first or second information processing devices 103 ₁ and 103 ₂ that issued the request. When the input history management unit 108 also receives the input data 102 or the analysis result thereof or the generated output data 104 from the first or second information processing apparatus 103 ₁ or 103 ₂ , the input history storage unit 1
These data are stored in 06.

In the input history storage unit 106, the input history management unit 108 is used by the first or second information processing device 103 ₁ , 10 2.
It is designed to store auxiliary information exchanged with 3 ₂ . Examples of such data include the input data 102 input in the past, the analysis result thereof, and the generated output data 104. Input history storage unit 10
6 will provide these auxiliary information in response to a request from the input history management unit 108.

The domain knowledge management unit 109 is means for managing domain knowledge used by the first or second information processing device 103 ₁ or 103 ₂ for information processing. The domain knowledge management unit 109 is composed of search and extraction means and the like as described above. The domain knowledge management unit 109
When the domain knowledge storage unit 107 receives a domain knowledge search or extraction request from the first or second information processing apparatus 103 ₁ or 103 ₂ , the domain knowledge storage unit 107 retrieves domain knowledge satisfying the request. Then, this is returned to the requesting device of the first and second information processing devices 103 ₁ and 103 ₂ .

In the domain knowledge storage unit 107, the domain knowledge management unit 109 has the first or second information processing device 103.
_1, it stores the supplementary information to be exchanged between the 103 _2.
Such auxiliary information may include domain information related to the input data 102 and the degree of association thereof.
The domain knowledge storage unit 107 will provide auxiliary information according to the request of the domain knowledge management unit.

FIG. 2 shows an outline of the control of the information processing system of the first embodiment as described above. Now, the first information processing device 10 shown in FIG.
It is assumed that 3 ₁ is first selected and input processing is performed from the input device 101. When the input device 101 receives the user's input, the input device 102 receives the input data 102 from the first information processing device 103.
Send to ₁ . The first information processing apparatus 103 ₁ analyzes the received input data 102 and newly generates output data 104. The first information processing apparatus 103 ₁ sends the input data 102 to the input history management unit 108, an output data 10 the analysis was the result and the newly generated input data 102
4 is sent to the input history management unit 108 and the output device 105. The input history management unit 108 receives the input data, the analysis result thereof, and the newly generated output data 104.
Is stored in the input history storage unit 106 (step S2 in FIG. 2).
01). Upon receiving the output data 104, the output device 105 outputs this to the user.

Next, it is assumed that the user selects the second information processing device 103 ₂ as the information processing device. Input device 102 receives an input from the user, and sends the input data 102 to the second information processing apparatus 103 ₂ (step S202). When the second information processing apparatus 103 ₂ receives the input data 102, past input data, the search and extraction request of the input history to the input history management unit 108 to obtain output data 104 and the analysis result and generating Is sent (step S203).

When the input history management unit 108 receives a request for an input history and a request for extraction from the second information processing apparatus 103 ₂ , the input history management unit 108 inputs the input data satisfying the request, the analysis result thereof, and the generated output data 104. The history storage unit 106 is searched to extract the relevant one. The input history data thus obtained is sent to the _second information processing apparatus 103 ₂ (step S204).

The second information processing apparatus 103 ₂ has the input history 1
When the input history data indicating 11 is received, the input history data and the domain knowledge search / extraction request are sent to the domain knowledge management unit 109 to obtain the domain information of the input data 102 (step S205).

When the domain knowledge management unit 109 receives the input history data and the domain knowledge search / extraction request from the second information processing apparatus 103 ₂ , the domain knowledge storage unit 107 stores the domain information based on the input history data.
Search and extract from. Then, the obtained domain information is sent to the _second information processing device 103 ₂ (step S20).
6).

[0062] The second information processing apparatus 103 ₂ receives the domain information, and generates the analysis and output data of the input data 102 and the domain information to the group (step S207). The second information processing apparatus 103 ₂ also sends the input data 102 to the input history management unit 108,
The analysis result of the input data and the newly generated output data 104 are sent to the input history management unit 108 and the output device 105. The input history management unit 108 receives the input data 1
02, the analysis result and newly generated output data 1
04 is stored in the input history storage unit 106 (step S2
08). Upon receiving the output data 104, the output device 105 outputs it to the user (step S209).

Therefore, in the first embodiment of the present invention,
When ambiguity occurs in the analysis of the input data 102 or the generation of the output data 104 in the process of the information processing by the first or second information processing devices 103 ₁ and 103 ₂ ,
By referring to the input history and domain knowledge, auxiliary information for this input data 102 can be obtained.
Therefore, compared with the conventional information processing system, there is an advantage that the analysis of the input data 102 and the generation processing of the output data 104 are easy and accurate.

Next, the information processing system 100 described above
A case in which is configured as an automatic interpretation system will be specifically described. In this case, the input device 101 shown in FIG. 1 is composed of a microphone, a keyboard, a mouse and a touch panel. In addition, the first and second information processing devices 103
_1, 103 ₂ is composed of a personal computer. The input history storage unit 106, the domain storage unit 107, and the data storage device storing the control program (not shown) are
It is composed of a magnetic disk storage device. Output device 105
Is composed of a speaker and a display.

The input history management unit 108 and the domain knowledge management unit 109 can also be realized by independent personal computers or personal computers that also have other functions. The first information processing device 103 ₁ constitutes a speech translation processing device, the second information processing device 103 ₂ constitutes a parallel translation example sentence translation processing device, and these are combined to realize an automatic interpretation system. In this automatic interpretation system, a Japanese speaker (hereinafter referred to as a Japanese speaker) and an English speaker (hereinafter referred to as an English speaker) are input devices 101.
A dialogue is performed via the output device 105 and the voice translation by the first information processing device 103 ₁ and the second
Two functions of parallel translation example sentence translation by the information processing apparatus 103 ₂ are used.

FIG. 3 is a diagram for explaining a process when a Japanese speaker speaks Japanese when the Japanese speaker uses the voice translation function to interact with an English speaker. Japanese speaker is Figure 1
Heading to the microphone as the input device 101 shown in, speak "Is the room empty?" The voice input from the microphone is the first information processing device 10
A CPU (not shown) in 3 ₁ converts the character string into a character string by a voice translation function by executing a predetermined control program, and the converted character string is translated from Japanese into English. As a result of this translation, "Do you have a
room? It is assumed that the English sentence "" is obtained. The converted English sentence is converted into voice data and output device 10
A voice is output from the speaker as 5, and a character string is output from the display as the output device 105 as shown in FIG.

At this time, the input history storage unit 1 shown in FIG.
In 06, the speaker is a "Japanese speaker", the recognition result of the utterance by the speaker is "is the room free?", And the output data 104 is the result of translating the speaker.
It is memorized that the English sentence "Do you have a room?" Was obtained and that the unique speaker ID (IDentification) for this speaker was "001". Further, morphological analysis is performed on the speaker sentence of the input data, and as a result, the input utterance is divided into words as registered in the dictionary and stored in a temporary storage area (not shown) together with the part-of-speech information. The information stored is "room (noun), ha (particle), vacant (verb), te (particle), there is (auxiliary verb), masu (auxiliary verb),
Ka (particle) ". Here, refer to the "unnecessary word / part-of-speech list" that is a list of words and parts of speech that are thought to be unnecessary for analysis, and only the words that are not listed as unnecessary words and part-of-speech in this list,
The data is extracted from the temporary storage area and stored in the input history storage unit 106. Here, in the unnecessary word list, "particle",
If there is an "auxiliary verb", the "room (noun)" and "vacant (verb)" that do not correspond to these are input history storage unit 1
It is stored in 06. The English translation of the word extracted at this time is also stored in the input history storage unit 106 in the same manner.
Here, the translations for "room" and "vacant" are "have (verb)" and "room (nou).
n) ”will be stored.

FIG. 4 is a diagram for explaining the processing when a Japanese speaker speaks English when he or she uses the voice translation function to interact with an English speaker. It is assumed that the English speaker answers “Yes.” As an affirmative English word from the microphone constituting the input device 101 in response to the Japanese speaker's utterance “Is the room empty?”. The utterance of this answer is translated into Japanese "Yes." As a result of the same processing as described above. Then, based on this voice data, this voice is output from the speaker constituting the output device 105, and a character string is displayed on the display.

At this time, the input history storage unit 1 shown in FIG.
In 06, the speaker is an “English speaker”, the recognition result of the utterance by this speaker is “Yes.”, And the result of translating the speaker is “Yes.” As the output data 104.
It is stored that the Japanese sentence is obtained and that the unique utterance ID (IDentification) for this speaker is “002”. Further, morphological analysis is performed on the speaker sentence of the input data, and as a result, the input speaker is divided into words as registered in the dictionary and stored in the temporary storage area together with the part-of-speech information. Here, refer to the "unnecessary word / part-of-speech list", which is a list of words and parts of speech that are considered unnecessary for analysis, and temporarily store only those words that are not listed as unnecessary words or part-of-speech in this list. Input history storage unit 10 extracted from area
Store in 6. Here, the non-word list has "artic"
"le" and "preposition"
“Yes (advert)” that does not correspond to these is stored. Further, the Japanese translation of the word extracted at this time is also stored in the input history storage unit 106 in the same manner. In this case, the translated word "Yes" for "Yes" is additionally stored.

Next, a case where a Japanese speaker interacts with an English speaker using the parallel translation example sentence translation function will be described. Here, it is assumed that the bilingual example sentences are classified into rough scenes.
For example, taking an overseas travel as an example, each scene such as “hotel” or “restaurant” corresponds to this. Examples of sentences classified into each scene are, for example, "Is there an air conditioner in the room?" Or "What time is check-out?" I want to order. "And" Please make the meat rare. "

FIG. 5 illustrates a process in which a Japanese speaker uses the bilingual example sentence translation function to interact with an English speaker, and FIG. 6 shows a flow of this process. Consider a case where a Japanese speaker voice-inputs the keyword "Ikura" from the input device 101 to search for parallel translation example sentences. Receiving this keyword from the input device 101 (FIG. 6 step S221) the second information processing apparatus 103 having the translated example sentence translation function _2, stores the input history, using the input history storage unit 106 for retrieval and extraction input Search history. Then, the input sentence input in the past and the analysis result thereof are extracted. If you search the input history, you will find "Utterance ID-0" which is the utterance of the previous "Japanese speaker".
01 ”“ Is the room vacant? ”And its analysis results“ Room (noun), vacant (verb) ”, parallel translation“ have ”
(Verb), room (noun) ”and“ Yes. ”Of“ Utterance ID-002 ”, which is the utterance of“ English speaker ”immediately before the search for the Japanese speaker this time, and the analysis result“ Yes (adverb) ”. ", The parallel translation" yes (adverb) "is obtained (see FIG. 4, step S22 in FIG. 6).
2).

The domain knowledge storage unit 107 stores information called “domain” that can be commonly used by the first information processing apparatus 103 ₁ and the second information processing apparatus 103 ₂ . Here, the above-mentioned "scene" is information that can be used by both the voice translation function and the parallel translation example sentence translation function. This "scene" is still the first
Information processing device 103 ₁ and second information processing device 10
It is characterized by equivalent obtained information and by 3 _second processing. Here, in both the voice translation function and the parallel translation example sentence translation function, the words that are its constituent elements are obtained as a result of analyzing each speaker and example sentence, so each "scene" is characterized by this set of "words". To be For example, if the scene is “hotel”, the set of words such as “check-in, reception, room, ...
Then, it is characterized by a set of words such as "seat, wait, menu, ...".

In the second information processing device 103 ₂ , the domain knowledge storage unit 109 is used by vectorizing the result of the input history as an input history vector (step S223) and searching and extracting the domain knowledge. The vector representing each scene stored in 107 is extracted (step S224), and the degree of similarity with the input history vector is calculated (step S225). At this time, the scene having the highest similarity vector is determined to be the scene in which the dialogue is being performed, and this scene information is extracted (step S22).
6). In the case of this example, "hotel" is extracted as a scene having a vector having the highest similarity to the vector representing the input history (hereinafter, simply referred to as the input history vector), and this information has the second translation example sentence translation function. Is sent to the information processing device 103 ₂ (step S227).

Here, as a method for calculating the similarity between the input sentence and each domain, there is a vector space model described in, for example, "Information Retrieval and Language Processing (Kennobu Tokunaga, The University of Tokyo Press)". . The outline of this method is as follows. Utterances exchanged in the "Restaurant" domain (such as "Please give me a menu") and Utterances exchanged in the "Shopping" domain ("Can I try it on?")
And so on). It is assumed that words (“menu” and “try-on”) appearing in both domains are given unique IDs, and the total number of words becomes “n”. At this time, each domain is represented by an n-dimensional vector, and I
When the value of the i-th element is a value calculated in consideration of the appearance frequency or the appearance frequency of the word in which D is “i”, a vector having a high value in the element (frequently occurring) that characterizes each domain is created. This is regarded as a vector representing each domain.

Next, similarly to the input sentence, an n-dimensional vector corresponding to the frequency of appearance of words in the sentence is created. We regard this as a vector representing the input sentence. At this time, when the inner product of the input sentence vector and each domain vector is calculated, the value increases as the number of elements having higher values is shared.
In other words, if the value of this inner product is large, the more frequent words are shared, which means that the input sentence is similar to the sentence of the domain. As a result, the domain having the vector with the largest inner product can be regarded as the domain of the input sentence. That is, if there is an input sentence "Please give me a menu in Japanese."

[0076] The second information processing device 103 having a bilingual example sentence translation function ₂ receives the "hotel" as the scene information,
It is determined that the search using the keyword “Ikura” described above mainly targets the example sentences in this scene. And
When actually searching based on a keyword and displaying the search result, a search request is made for the parallel translation example sentences classified for each scene corresponding to the scene in the domain knowledge storage unit 107, which is classified as “hotel”. Will be displayed on the top as a match.

FIG. 7 shows a display example of the display in the case of this example. In the case of this example, dialogue example sentences such as "How much is the room charge?" Or "How much is the single room with a shower?" Is displayed. The Japanese speaker who sees this result can easily obtain an example sentence suitable for the current conversation scene, and thus can smoothly communicate with the English speaker.

In the first embodiment described above, a word is assumed as the input history information stored in the input history storage section. Another type of information stored in the input history storage unit is the syntactic structure of the input sentence. By using the syntactic structure, the relationship between words can be described by numerical information, etc., and more detailed and accurate results can be obtained when searching domain knowledge based on the input history.

When the information processing system of this embodiment is realized on a personal computer, it is possible to store the input date and time as the input history. "Tomorrow" and "1 hour later" during the utterance by referring to the date and time of the input history
Such expressions can be converted into concrete numerical data. In addition, by using the geographical information system, it is possible to obtain information on the specific position of the speaker using the automatic interpretation system, whether it is a hotel or a restaurant. As another input history, an operation procedure at the time of example sentence search can be considered. Here, it is assumed that the example sentences of the example sentence search function are classified into folders corresponding to the domains in advance, but when the example restaurant search function is used first, the "Restaurant" folder is selected and the example sentence in it is used. To do. Then, by leaving the selected restaurant in the input history, the "restaurant" domain can be immediately obtained when acquiring domain knowledge, and when the next voice translation is performed, the domain information can be used in the translation word selection, etc. it can. Domains can also have a hierarchical structure. For example, by creating subdomains such as "reservation" and "check-in" under the "hotel" domain under "hotel", more detailed domain information can be used.

<Second Embodiment>

FIG. 8 shows the overall configuration of an information processing system according to the second embodiment of the present invention. The information processing system 300 includes an input device 301 that receives inputs such as voice and type, and a first device that processes input data 302 input from the input device 301 to generate an output.
-Third information processing devices 303 _{1 to} 303 ₃ and these first
And an output device 305 for outputting a processing result such as a voice and a character string From the third information processing apparatus 303 _{1 to 303 3} receive output data 304.

This information processing system 300 includes an input device.
The history of the input data 302 accepted by 301 is stored.
Input history storage unit 306 and first to third information processing devices 3
03 ₁~ 303₃Knowledge that can be referred to and used by
It also includes a domain storage unit 307 that stores Input shoes
The history management unit 308 stores the input data in the input history storage unit 306.
302 is stored, and the search is performed and necessary
It is designed to extract data. Domain knowledge management
The unit 309 is stored in the domain knowledge storage unit 307.
It is designed to search for and extract data that is used.

In the information processing system 300 as described above, the input device 301 firstly outputs data such as voice by utterance input, character string by type input, position information by mouse input as a pointing device or position information touched by a fingertip. third adapted to deliver the information processing apparatus 303 _{1 to 303 3} ~.

[0084] The first to third information processing apparatus 303 ₁ - 30
3 ₃ is the input data 30 input from the input device 301
In addition to 2, the auxiliary information such as the input history 311 and the domain knowledge 312 is received from the input history management unit 308 and the domain knowledge management unit 309, respectively. Then, based on these auxiliary information, the input data 302 is analyzed and the appropriate output data 304 is generated. Input data 302 and results of these analyzes of the first to third information processing apparatus 303 _{1 to 303 3} are fed to the input history management unit 308 and the domain knowledge manager 309. Further, the output data 30 generated in the first to third information processing apparatus 303 _{1 to 303 3}
4 is an input history management unit 308 and a domain knowledge management unit 30.
9 and the output device 305.

[0085] The first to third information processing apparatus 303 ₁ - 30
As described above, the input data 302 received by the 3 ₃ includes data such as voice, character string, or position information.
The output data 304 sent to the output device 305 includes data such as voice, character strings, and image information. The auxiliary information received from the input history management unit 308 and the domain knowledge management section 309, the input history and the analysis result, the first through third information processing apparatus 303 _{1 to 303} output data 304 generated by _3, and the input history There are related domain information and data indicating the degree of association. As the data first to third information processing apparatus 303 _{1 to 303 3} are sent to the input history management unit 308 and the domain knowledge manager 309, the input data 302 and the analysis result, the first through third information there are processing apparatus 303 _{1 to 303 3} output data 304 and the input history and the like generated by. These first to third information processing apparatus 303 _{1 to 303 3,} if the user performs an input from the input device 301, executes a predetermined process by calling the respective units described above in a predetermined order.

The output device 305 constituting the information processing system 300 is one of a voice output device for outputting a voice, a display device for displaying image information such as characters, or a combination thereof. . The output device 305 is the first to third information processing devices 303.
And the output data 304 sent from _{1-303 3} based on output voice, text, data such as image information.

On the other hand, the input history management unit 308 uses the first to third
But the information processing apparatus 303 _{1 to 303 3} to manage the input history for use in information processing, as one example is provided with means for the storage, retrieval and extraction input history. The input history management unit 308 is the first to third information processing devices 303.
Upon receiving the search or request the extraction of the input histories from _{1 to 303 3,} take out the input history satisfy the request from the input history storage unit 306. Then, return it to the device that issued the request of the first to third information processing apparatus 303 _{1 to 303 3.} Input history management unit 308, also first to third input data 302 or the result of the analysis from the information processing apparatus 303 _{1 to 303 3} output data 30 or generated,
When 4 is received, these data are stored in the input history storage unit 306.

The domain knowledge management unit 309 is the first to third
The information processing apparatus 303 _{1 to 303 3} are means for managing the domain knowledge to be used when the information processing. The domain knowledge management unit 309 is composed of search and extraction means as described above. The domain knowledge management unit 309 is the first
When From the third information processing apparatus 303 _{1 to 303 3} receives a request for search or extraction of domain knowledge, taking out the domain knowledge to satisfy the request from domain knowledge storage unit 307. Then, this is the first to third information processing devices 303
It returned to the device that originated the request of _{1-303 3.}

FIG. 9 shows an outline of the control of the information processing system of the second embodiment. The first information processing device 30 shown in FIG.
It is assumed that 3 ₁ is selected and input processing is performed from the input device 301. Upon receiving the user's input, the input device 301 sends the input data 302 to the first information processing device 303 ₁ . The first information processing apparatus 303 ₁ analyzes the input data 302 received, to generate a new output data 304. The first information processing apparatus 303 ₁ sends the input data 302 to the input history management unit 308, the input data 30
The analysis result of No. 2 and the newly generated output data 304 are sent to the input history management unit 308 and the output device 305.
The input history management unit 308 stores the received input data, the analysis result thereof, and the newly generated output data 304 in the input history storage unit 306 (step S40 in FIG. 4).
1). Upon receiving the output data 304, the output device 305 outputs it to the user.

Next, it is assumed that the user has selected the second information processing device 303 ₂ as the information processing device. When the input device 302 receives input from a user, and sends the input data 302 to the second information processing apparatus 303 ₂ (step S402). When the second information processing apparatus 303 ₂ receives the input data 302, past input data, the search and extraction request of the input history to the input history management unit 308 to obtain output data 304 the analysis result and generating Is sent (step S403).

[0091] The input history management unit 308 receives a request and the request of extraction of the input history from the second information processing apparatus 303 _2, inputs the input data or output data 304 the analysis result and generating, satisfy the request A search is made from the history storage unit 306 and the relevant one is extracted. The input history data thus obtained is sent to the _second information processing device 3032 (step S404).

[0092] The second of the information processing apparatus 303 ₂ input history 3
When the input history data indicating 11 is received, the input history data and the domain knowledge search / extraction request are sent to the domain knowledge management unit 309 to obtain the domain information of the input data 302 (step S405).

[0093] domain knowledge manager 309 receives the input history data and domain knowledge search and retrieval request from the second information processing apparatus 303 _2, domain knowledge storage unit 307 domain information by the input history data of this group
Search and extract from. Then, the obtained domain information is sent to the _second information processing device 3032 (step S40).
6).

[0094] The second information processing apparatus 303 ₂ receives the domain information, and generates the analysis and output data of the input data 302 and the domain information to the group (step S407). The second information processing apparatus 303 ₂ also sends the input data 302 to the input history management unit 308,
The analysis result of the input data and the newly generated output data 304 are sent to the input history management unit 308 and the output device 305. The input history management unit 308 receives the input data 3
02, the analysis result and newly generated output data 3
04 is stored in the input history storage unit 306 (step S4).
08). Upon receiving the output data 304, the output device 305 outputs it to the user (step S409).

Therefore, for example, when ambiguity arises in the analysis of the input data 302 or the generation of the output data 304 in the process of the information processing by the first and second information processing devices 303 ₁ and 303 _2. Further, auxiliary information for this input data 302 can be obtained by referring to the input history and domain knowledge. Therefore, compared to the conventional information processing system, there is an advantage that the analysis of the input data 302 and the generation processing of the output data 304 are easy and accurate.

[0096] Now, it is assumed that the user after this has selected the third information processing apparatus 303 ₃ as an information processing apparatus.
When the input device 301 receives input from a user, and sends the input data 302 to the third information processing apparatus 303 ₃ (step S410).

[0097] The third information processing apparatus 303 ₃ receives the input data 302, past input data, the input history search, to the input history management unit 308 to obtain output data 304 the analysis result and generating An extraction request is sent (step S411).

When the input history management unit 308 receives the input history search / extraction request from the third information processing apparatus 303 ₃ ,
The input data satisfying the requirement, the analysis result thereof, and the generated output data are searched and extracted from the input history storage unit 306. The obtained input history data is sent to the _third information processing apparatus 3033 (step S412).

[0099] The third information processing apparatus 303 ₃ receives the input history data from the input history management unit 308, domain knowledge management section 30 to obtain the domain information of the input data
The input history data and the domain knowledge search / extraction request are sent to 9 (step S413).

[0100] Once the domain knowledge manager 309 receives the input history data and domain knowledge search and retrieval request from the third information processing apparatus 303 _3, based on the input history data search domain information from the domain knowledge storage unit 307 To extract. And sends the domain information obtained to the third information processing apparatus 303 ₃ (step S414).

[0102] The third information processing apparatus 303 ₃ receives the domain information from the domain knowledge management section 309, and generates the analysis and output data 304 of the input data 302 based on this (step S415). Then, the input data 302 is sent to the input history management unit 308, and the analysis result of the input data 302 and the newly generated output data 3
04 is sent to the input history management unit 308 and the output device 305. The input history management unit 308 receives the input data,
The analysis result and newly generated output data 304
Is stored in the input history storage unit 306 (step S41).
6). Upon receiving the output data 304, the output device 305 outputs it to the user (step S41).
7).

[0102] In a second embodiment of the present invention, in the process of any one of the first to third information processing apparatus 303 _{1 to 303 3} is performing data processing, analysis and output of the input data 302 If such ambiguities in the generation of the data 304 has occurred, the input data 102 by referring to the input history and domain knowledge of the remaining devices of the first to third information processing apparatus 303 _{1 to 303 3} Can obtain supplementary information for. Therefore, as compared with the conventional information processing system, the analysis of the input data 302 and the output data 30
There is an advantage that the generation process of 4 becomes easy and accurate.

Next, the information processing system 300 described above
The case where the above is configured as a video editing system will be specifically described. In this case, the input device 301 shown in FIG. 8 is composed of an image input terminal, a voice input terminal, a microphone, a keyboard, a mouse and a touch panel. Also configured in the first to third information processing apparatus 303 _{1 to 303 3} are personal computers. The input history storage unit 306, the domain storage unit 307, and the data storage device that stores the control program (not shown) are magnetic disk storage devices. The output device 305 includes an image output terminal, an audio output terminal, a speaker, and a display.

[0104] The input history management unit 308 and the domain knowledge management section 309, first to third information processing apparatus 303 ₁
It can be composed of a personal computer as in the case of 303 ₃ . In the second embodiment, the first information processing device 303 ₁ constitutes a gesture recognition processing device, the second information processing device 303 ₂ constitutes a telop recognition processing device, and the third information processing device 303 ₃ Configures a voice recognition processing device, and video editing is performed by them.

It is assumed that the user uses the video editing system to perform editing work such as voice recognition and indexing on the contents of the lecture program and news program broadcast on television. In the present embodiment, as an example, with respect to the contents of a lecture program related to "four major civilizations" of "world history", a work of converting the lecture contents into a character string using a voice recognition device and creating an index or the like will be described.

FIG. 10 is for explaining the initial state of this work. Let's assume that the speaker points to the area on the map to explain the "China's Yellow River Basin", which is one of the four major civilizations. Input data 30 as input image
2 is processed by the gesture recognition processing function for realizing the first information processing apparatus 303 _1, the position coordinate information corresponding to the location pointed is generated. In the example shown in FIG. 10, the position coordinate information has coordinates (x, y) of (136, 78). Further, it is assumed that the position information “Yellow River Basin of China” is obtained as a region corresponding to the recognition result of the position coordinate information by collating with the world map in the database prepared in advance. At this time, as input information, the corresponding input data is its ID, the position coordinate information of the recognition result by gesture recognition is numerical information, and the position information corresponding to the position coordinate information is a character string.
06 and is displayed on the display as the output device 305 when necessary. At this time, it is assumed that the position information, which is a character string, is subjected to morphological analysis, and the constituent words and part-of-speech information thereof are also stored in the input history storage unit 306.

FIG. 11 illustrates the processing when a telop appears in a program, and FIG. 12 shows the flow of this processing. Here, a telop means a character or image information added on the screen of a program. For example, when watching a television program, characters that inform about earthquakes, incidents, accidents, etc. may suddenly appear. These are telops. When a telop appears in a lecture program, the content of the lecture is often emphasized or summarized. Here, it is assumed that a telop called "Yellow River Civilization" has appeared in the lecture to introduce the civilization of the Yellow River basin.

When the telop appears, the input image is sent to the _second information processing apparatus 3032 as the input data 302, and the telop recognition unit (not shown) recognizes the telop (step S421 in FIG. 12). That is,
The second information processing apparatus 303 ₂ stores the input history by using the input history management unit 308, a feature for searching and extraction, to find the input history stored in the input history storage unit 306. Then, the input data input to the input history storage unit 306 in the past and the analysis result thereof are extracted (step S422).

When the input history is searched, as a result of the gesture recognition for the immediately preceding input image data, information that the place pointed by the speaker is “China Yellow River Basin” is “China (noun), Yellow River (noun),” It is obtained by the word set "basin (noun)". Second with the function of telop recognition
The information processing apparatus 303 ₂ converts the information of the input history to the input history vector (step S423), the domain knowledge manager 309 using domain knowledge management section 309 having a function of searching and extracting a domain knowledge The stored vector representing each domain is extracted (step S424), and the degree of similarity with the input history vector is calculated (step S425). Then, at this time, the domain having the highest similarity vector is determined as the domain to which the main topic of the lecture content is related, and this domain information is extracted (step S426). In the example shown in FIG. 11, the domain corresponds to the country (region) in which the lecture content is spoken, and there are, for example, “China” and “Japan”. Each "country (region)" is characterized by a set of "words". For example, if the country (region) is "China", it is a set of words such as "Yellow River, Han, Guangmu Emperor, ...", and if it is "Japan", the word is "Kinji, Yamadaikuni, Yayoi period, ..." Characterized by a set. Here, when extracting the domain information, the “China” vector having the highest similarity between the vectors of “0.74” is extracted. Therefore, so that the "Chinese" as the extracted domain is sent from the domain knowledge manager 309 to ₂ second information processing apparatus 303 having the functions of the telop recognition (step S
427).

[0110] The second information processing apparatus 303 ₂ receives the "Chinese" as the domain information, the telop is determined that a deep relationship with this domain. When actually recognizing this telop, it is assumed that "Yayoi civilization" classified into the "Japan" domain and "Yellow river civilization" classified into the "China" domain are obtained as recognition candidates.
Since "China" is obtained as the domain information based on the input history, "Yellow River civilization" is adopted as the recognition result with a higher likelihood among the recognition candidates that are classified and stored for each domain. .

FIG. 13 shows a state of subsequent processing of the second information processing apparatus which has received the domain information as the recognition result of the telop. The recognition result of this telop is stored in the input history storage unit 306 as a character string. That is, “Yellow river civilization” is stored in the input history storage unit 306 corresponding to “telop”. Such a recognition result will be output on a display as the output device 305 when necessary.

Next, a case will be described in which the third information processing apparatus 303 ₃ as a speech recognition processing apparatus converts the speech of the speaker in this information processing system 300 into a character string. It is assumed that the speaker uttered “I found Saitou” while explaining the Yellow River civilization.

FIG. 14 is for explaining the processing for converting the speech of such a speaker into a character string. When the third information processing apparatus 303 ₃ receives input data 302 representing the speech store input history, looking for input history using the input history management unit 308 has a function of searching and extraction. Then, the input data input in the past and the analysis result thereof are extracted. When the input history management unit 308 searches the input history, “Chinese Yellow River basin” and the word set “China (noun), Yellow River (noun), basin (noun)” as the result of the previous gesture recognition result are Further, "Yellow River Civilization" and its analysis results "Yellow River (Noun) and Civilization (Noun)" are obtained as the immediately preceding telop recognition results.

[0114] The third information processing apparatus 303 ₃ and vectorized these results, each domain that are stored in the domain knowledge storage unit 307 by using the domain knowledge manager 309 having the function of searching and extracting a domain knowledge Compare with the vector. Then, at this time, it is determined that the domain having the vector with the highest similarity is the vector related to the main topic of the lecture content, and this domain information is extracted. In the example shown in FIG. 14, the “China” vector having the highest similarity of “0.74” is extracted. "China" as this extracted domain
There will be sent from the domain knowledge manager 309 to ₂ second information processing apparatus 303 having the function of the telop recognition.

[0115] The third information processing apparatus 303 ₃ receives a "Chinese" as the domain information, a previous utterance "Saito is found." Is determined to be those deep relevant to this domain. Then, when performing this voice recognition, instead of "Saito" that is classified as the scene of "Japan" among the recognition candidates that are classified and stored for each domain as the recognition result, the scene of "China" is selected. The classified “Saito” will be adopted as the recognition result with higher likelihood. The voice recognition result is stored in the input history storage unit 306 as a character string.

FIG. 15 shows how the result of voice recognition is displayed on the display. Output device 30
On the display as 5, "Saito was found."
Will be displayed.

[0117] By as described is first to third information processing apparatus 303 _{1 to 303 3} use the input history of each other than the user more accurate gesture recognition result, telop recognition result and the speech recognition result You can get
Video editing work can be performed with less modification work than the prior art.

In the second embodiment described above, a word is assumed as the input history information stored in the input history storage unit, but the present invention is not limited to this. For example, the position information on the image can be used as one of the information stored in the input history storage unit. In this case, by storing each domain of domain knowledge such as "China" and "Japan" in the domain knowledge storage unit as the location information on the image of the world map, the location information of the input history is stored. You can get domain information directly.

In the second embodiment, in FIG. 12, the telop recognition unit accepts the input of the telop image and extracts the domain information of the “China” domain having the highest degree of similarity. It is also possible to perform such processing by

FIG. 16 shows a process in which the voice recognition unit receives a voice input and extracts domain information as a modification of the second embodiment. In this modified example, the voice recognition unit accepts a voice input (step S421A), and performs the processes of step S422 and subsequent steps based on the recognized voice information. Since the processes after step S422 are essentially the same as those in FIG. 12, description of these processes will be omitted.

<Third Embodiment>

FIG. 17 shows the overall structure of the information processing system in the third embodiment of the present invention.
The information processing system 500 includes an input device 501 that receives a user's input, and first and second information processing devices 503 ₁ and 503 ₂ that process input data 502 input from the input device 501 and generate an output. An output device 505 is provided for receiving output data 504 from the first and second information processing devices 503 ₁ and 503 ₂ and outputting the output data 504 as a processing result such as a voice or a character string.

This information processing system 500 is used by the input history storage unit 506 storing the history of the input data 502 received by the input device 501 and the first and second information processing devices 503 ₁ and 503 ₂ both referring to. It also includes a domain storage unit 507 that stores knowledge that can be obtained. The input history management unit 508 stores the input data 502 in the input history storage unit 506, searches the input data 502, and extracts necessary data. The domain knowledge management unit 509 is adapted to search and extract data stored in the domain knowledge storage unit 507.

In the information processing system 500 as described above,
Information processing device 503 of No. 1₁Is input from the input device 501.
Data 502 and analyze it and output appropriate data 5
04 is generated. Then, the input data 502 and the solution
Input analysis results History management unit 508 and domain knowledge management
To the section 509 and output data 504 to the input history management section 5
08, domain knowledge management unit 509 and output device 505
Send to. Input data 5 received from the input device 501
02 is a data such as voice, character string or position information.
Output data 504 to be sent to the output device 505.
Data such as voices, character strings or image information.
It In addition, the input history management unit 508 and domain knowledge manager
As the data to be sent to the processing unit 509, the input data 502 and
Analysis results, generated output data or input history, etc.
There is. First information processing device 503 ₁Is the user input
In accordance with the above, the information processing system shown in FIG.
Calling each device or each part in the stem 500
It is designed to perform the processing of.

[0125] input history management unit 508, the second information processing apparatus 503 ₂ is adapted to manage the input history to be used when the information processing. Second information processing device 503
_{The reference} numeral ₂ is composed of various means such as storage of input history, retrieval and extraction. Input history management unit 508 receives the input history search and retrieval request from the second information processing apparatus 503 _2, retrieves the input history 511 to satisfy the request from the input history storage unit 506, the second information having issued this request It returns it to the processing unit 503 _2. In addition, the input history management unit 508 uses the first or second information processing device 50.
When receiving the input data 502, the analysis result thereof, or the generated output data 504 from 3 ₁ , 503 ₂ , these data are stored in the input history storage unit 506.

On the other hand, the domain knowledge management unit 509 uses the second
The information processing apparatus 503 ₂ is composed of manages the domain knowledge to be used when the information processing, for example, search and extraction means, such as domain knowledge. Domain knowledge management unit 509 receives the domain knowledge search and retrieval request from the second information processing apparatus 503 _2, retrieves the domain knowledge 512 to satisfy the request from domain knowledge storage unit 507, a second information the requesting and it sends it to the processing unit 503 _2.

FIG. 18 shows an outline of the control of the information processing system of the third embodiment.

In the information processing system 500 shown in FIG.
The user operates the first information processing device 503. ₁Is selected
It The input device 502 receives the input data 502 from the user.
When received, the first information processing device 503₁Sent to
To do. First information processing device 503₁Is the input data received
Data 502 is analyzed and new output data 504 is generated.
To do. First information processing device 503₁Is the input data 502
To the input history management unit 508,
Input analysis result and newly generated output data 504
It is sent to the history management unit 508 and the output device 504. Input history
The management unit 508 receives the input data, the analysis result
And the newly generated output data, the input history storage unit 506
(Step S601 in FIG. 18). Output device 50
5 receives the output data 504 and outputs it to the user.
Force

It should be noted that S602 following this shown in FIG.
Processing ~S609 is never changed in the processing essentially of steps S202~S209 in the first and second information processing apparatus 103 _1, 103 ₂ in the first embodiment. Therefore, these explanations are omitted.

Therefore, in the third embodiment of the present invention,
If the first or second information processing apparatus 503 _1, 503 ₂ is in the process of doing information processing, such as ambiguities in the generation of the analysis or the output data 504 of the input data 502 is generated,
By referring to the input history and domain knowledge, auxiliary information for this input data 102 can be obtained.
Therefore, compared with the conventional information processing system, there is an advantage that the analysis of the input data 102 and the generation processing of the output data 104 are easy and accurate.

Next, the information processing system 500 described above
The case where is configured as the optimal advertisement distribution chat system will be specifically described. In this case, the input device 501 shown in FIG. 17 is composed of a microphone, a keyboard, a mouse and a touch panel. The first and second information processing apparatus 503 _1, 503 ₂ is composed of a personal computer. Input history storage unit 506, domain storage unit 5
The data storage device for storing 07 and the control program (not shown) is composed of a magnetic disk storage device. The output device 505 includes a speaker and a display.

Input history management unit 508 and domain knowledge
The management unit 509 uses the first and second information processing devices 503.
₁, 503₂Configure with a personal computer as in
be able to. First information processing device 503₁Chat
A second information processing device 503 which constitutes a processing device.₂Is the best
Configures an advertisement display processing device, and these are optimally combined
Ad delivery chat system realized. This optimal advertisement distribution
In a remote chat system, a user
First information processing apparatus 503₁Chat using
Second information as an optimum advertisement display processing device when performing
Processor 503 ₂To dynamically show the best ads
Has become.

FIG. 19 is for explaining the processing of the first information processing apparatus. Chat user inputs device 5
Suppose that the user inputs "I will go to Shibuya next Sunday." From the keyboard as 01. The input data 502 including the input character string is processed by the _first information processing device 5031 and output on the display as the output device 505. At this time, the input sentence and the speaker are “speaker A”, and the unique utterance ID (ID
Entry) is “001” is stored in the input history storage unit 506 by the input history management unit 508 having a function of storing, searching or extracting the input history.

It is assumed that another chat user replies to this utterance, "Let's go eat something." The input data 502 including the character string of this answer is also output on the display. As shown in FIG. 19, the input history storage unit 506 indicates that the character string input at this time and the speaker are “speaker B” and that the unique utterance ID (IDentification) for this utterance is “002”. Stored in.

[0135] In this case, the second of the information processing apparatus 503 ₂ as the best advertisement display processing device analyzes the content of the chat, consider the case to display the ads. The advertisements displayed here are classified by domain. By domain, for example, classifications such as “gourmet”, “theme park”, and “shopping” are given. Second information processing device 5
03 _2, input history in the input history management unit 508 has a function of searching and extracting to perform the search of the input history, to extract the inputted input sentence and its analysis result in the past.

FIG. 20 illustrates the state of processing by the second information processing apparatus, and FIG. 21 shows the flow of this processing. As described above, the speaker B responds to the utterance of the speaker A (step S621 in FIG. 21), and when the input history is searched, the user inputs "the next Sunday to play in Shibuya.""Go." And its analysis result "Sunday (Noun / Date), Shibuya (Noun / Place name), Play (Verb)" and "Let's go eat something." As a result, "what (pronoun), eat (verb)" is obtained (step S62).
2). Second information processor 503 ₂ as the optimum advertisement display processing device, these results is vectorized to create input history vector (step S623). Then, the domain knowledge storage unit 507 is provided to the domain knowledge management unit 509 having a function of searching and extracting domain knowledge.
A vector representing each domain stored in is extracted (step S624), and the degree of similarity with the input history vector is calculated (step S625). Here each "domain" is characterized by a set of words. For example, if the domain is "gourmet", "Thai food, paella, delicious,
"..." is a set of words, and "theme park" is characterized by a set of words such as "amusement park, roller coaster, ...". At this time, the domain having the highest similarity vector is determined as the domain representing the topic of chat, and this domain information is extracted. In the case of this example, "gourmet" is extracted as a scene having a vector having the highest similarity to the input history vector (step S
626), this information is sent to the _second information processing device 5032 as the optimum advertisement display processing device (step S627).

FIG. 22 shows an example of the display contents on the display in the state where the chat is being performed.
The second information processing apparatus 503 ₂ receives the "gourmet" as the domain information, history chats is determined that targeted topics in this domain. Therefore, on the display as the output device 505 in which the chat is being performed, the advertisement classified into the “gourmet” scene and belonging to the keyword “Shibuya” is placed beside the place where the chat is displayed. Will be displayed in the part.

The user who sees this display result can obtain an advertisement that matches the content of the dialogue, and can easily obtain necessary information such as the place of the advertisement and the menu of the advertisement for which this display is performed. .

In the third embodiment described above, the word is assumed as the input history information stored in the input history storage unit, but the present invention is not limited to this. For example, date and time information exists as one of the information stored in the input history storage unit. By referring to the date and time of the input history, expressions such as "tomorrow" and "1 hour later" in the input sentence can be treated as specific numerical data. Therefore, it is possible to divide the domain information into more detailed domains such as "theme park (January)" as well as "theme park".

Other information includes sex, age, name, etc. input by the user when participating in a chat. In chat, since strangers often interact with each other, there is a method of inputting gender, age, occupation, etc. in order to express the feature. By using these pieces of information, it becomes possible to deliver the optimum advertisement that matches the preferences of the participants. In addition, there are some chats in which a certain theme is set for dialogue, and the theme or the like that such a chat has can be used. In this example, the domain divisions such as “gourmet” and “theme park” are used, but there may be other transversal domains in each of these domains. For example, if a place such as "Shibuya" or "Shinjuku" is set as a domain, there may be a case where conversations are made about various topics such as gourmet food and shopping. In this case, it is possible to determine the domain as "Shinjuku" and distribute information such as movie theaters as an advertisement.

As the method of creating domain knowledge described here, data labeled for a set of many elements can be used. For example, in an international classification of patents or a directory search on a search site, a large number of elements (here, patents and web pages, respectively) are grouped as a set of items that belong to the same category, and each set is labeled. At this time, the label may be a domain, and the elements in each label may characterize the domain. In the case of patents and web pages, the content can be analyzed, and words and figures that frequently appear in only a certain domain can be extracted as those characterizing the domain, and domain knowledge can be constructed from these. Further, even for a document set which is not labeled, it is possible to divide it into some subsets by using the clustering method, and each of these subsets can be used as domain knowledge.

<Fourth Embodiment>

FIG. 23 shows the overall structure of the information processing system in the fourth embodiment of the present invention.
The information processing system 700 includes an input device 701 that receives inputs such as voice and type, and first and second information processing devices 703 ₁ that process input data 702 input from the input device 701 to generate outputs. 703 ₂ and an output device 705 for outputting the processing result such as voice and character string. The information processing system 700 also includes an input history storage unit 706 and a domain knowledge storage unit 707. The input history management unit 708 includes the input history storage unit 7
The input data is stored in 06, the search is performed, and necessary data is extracted. The domain knowledge management unit 709 is adapted to search and extract data stored in the domain knowledge storage unit 707.

However, the input device 701 and the output device 705
Each of these parts other than exists as a control program for realizing those parts stored in the storage part 712 as a RAM (random access memory) area connected to the data processing device 711. Data processing device 71
Reference numeral 1 denotes each of these devices, the input device 701, and the output device 7.
In addition to 05, it is connected to a recording medium 713, and has a CPU and an input / output interface circuit not shown. The recording medium 713 stores a control program body for utilizing the input history and domain knowledge, which is expanded in the storage unit 712 and the data processing device 711 executes each control program to input the input history management unit 70.
Each part such as 8 is functionally realized. Therefore, the recording medium 713 is a magnetic disk, an optical disk,
It can be constituted by a semiconductor memory, a CD (Compact Disk) -ROM, and other recording media.

Information processing system 70 of the fourth embodiment
0 realizes exactly the same circuit function as the information processing system 100 of the first embodiment. Therefore, the description of the specific control operation of the information processing system 700 is omitted.

In each of the embodiments described above, a plurality of information processing devices share the same input device or output device, but the present invention is not limited to this. That is, each information processing device may be provided with a dedicated input device or output device, or some of the information processing devices may share some input device or output device. May be.

In each of the above-described embodiments, the first information processing apparatus has been described as processing information without using the input history or domain knowledge, but the input history or domain knowledge already exists. In such a case, it is natural that the first information processing apparatus may use these.

Further, in each of the embodiments described above, the information processing devices other than the first information processing device such as the second information processing device always use the input history or the domain knowledge when performing the processing. However, it is not limited to this. That is, if the reliability of domain estimation does not deserve reference so that it does not reach the threshold value, or if the user does not want to use domain knowledge from the input history due to the characteristics of information processing, those It is free to provide options so that processing can be performed without using.

Further, in each of the embodiments described above, one function has been described as being independently processed by each information processing apparatus, but the present invention is not limited to this. That is,
For example, a plurality of information processing devices may simultaneously be in charge of the same processing as a device having a predetermined function such as a voice output device, or one device may realize a plurality of functions in a time division manner. .

Further, although the input device and the output device are described as separate devices in the embodiments, one device such as a touch panel has both a function as an input device and a function as an output device. Good.

[0151]

As described above, according to the first aspect of the present invention, the field of domain information as auxiliary information stored in the domain information storage means is based on the processing result of the first information processing means. Since the domain information to be selected and used in the processing of the second information processing unit is limited and the processing is performed, the domain information to be processed is limited, and the effective and appropriate processing can be performed.

According to the second aspect of the present invention, the first history as the processing history of the first information processing means is stored, and the second information processing means uses the domain information to process information. Since the field of the domain information when performing the process is selected based on the first history, the domain information as the auxiliary information used for the process of the second information processing unit can be limited and the process can be effectively performed. Can be done accurately and accurately.

Further, according to the invention described in claim 3, when a plurality of information processing means perform their own information processing,
The history when one of the information processing means performs the information processing is stored, and when the other information processing means performs the information processing, the necessary domain information is obtained from this history while observing the change of the history with time. Since the information is searched and extracted and supplied as a reference for processing, each information processing means can select and provide the necessary domain information when processing information, and the processing of all related information is possible. Can be performed effectively and appropriately.

According to the invention described in claim 8, knowledge for each field used for information processing in the second information processing step is searched by using the history information obtained in the first information processing step, and is searched. Since the knowledge of the field obtained by the above is supplied to the processing in the second information processing step, the second information processing can be performed effectively and appropriately by using the domain information in which the application field is limited.

Further, according to the invention described in claim 9, when information processing is performed by any of the information processing means, a history of the processing is stored, and other information processing means processes other information processing. The history related to the information processing this time is searched from the history of the processing stored in the history storage step when it is performed, and the specific domain information is searched in relation to the searched history. Since this is extracted, the information processing this time can be performed efficiently and accurately using this.

According to the invention described in claim 10, the computer as an information processing program, the first and second information processing, and the second information processing based on the result of the first information processing.
Domain information extraction processing for extracting specific domain information from domain information as field-specific knowledge used for information processing of the domain, and domain information extracted by this domain information extraction processing is supplied for the second information processing. Since the specific domain information supply processing to be performed is executed, the second information processing can be efficiently performed using appropriate domain information.

[0157] According to the invention described in claim 11, the first and second information processing programs are stored in the computer.
Information processing, and domain information extraction processing for extracting specific domain information from domain information as knowledge of each field used for the second information processing based on the history of the first information processing, and this domain information Since the specific domain information supply processing for supplying the domain information extracted by the extraction processing for the second information processing is executed, it is possible to perform the second information processing efficiently and accurately by narrowing it down to appropriate domain information. it can.

Further, according to the information processing program of the invention described in claim 12, when the computer is caused to perform a plurality of information processing for independently processing the information as the information processing program, and the processing is performed. Since the history storage processing that stores the processing history is performed, the specific domain information is extracted from the domain information as the knowledge for each field to be used for the next information processing while watching the history change over time. Can be made. Therefore,
By supplying the extracted domain information for the information processing, the efficiency of the information processing can be improved by using the appropriate domain information.

[Brief description of drawings]

FIG. 1 is a block diagram showing an overall configuration of an information processing system according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing an outline of control of the information processing system according to the first embodiment.

FIG. 3 is an explanatory diagram showing a process when a Japanese speaker speaks Japanese when the Japanese speaker uses the voice translation function to interact with an English speaker in the first embodiment.

FIG. 4 is an explanatory diagram showing a process when English is uttered when a Japanese speaker interacts with an English speaker using a voice translation function in the first embodiment.

FIG. 5 is an explanatory diagram illustrating a process when a Japanese speaker has a dialogue with an English speaker using the parallel translation example sentence translation function in the first embodiment.

FIG. 6 is a flowchart showing a process when a Japanese speaker has a dialogue with an English speaker by using a parallel translation example sentence translation function in the first embodiment.

FIG. 7 is a plan view showing an example of a display result of a display in the first embodiment.

FIG. 8 is a block diagram showing an overall configuration of an information processing system in a second exemplary embodiment of the present invention.

FIG. 9 is a flowchart showing an outline of control of the information processing system according to the second embodiment.

FIG. 10 is an explanatory diagram showing a process when a speaker points to a corresponding area on a map in the second embodiment.

FIG. 11 is an explanatory diagram illustrating a process when a telop appears in a program according to the second embodiment.

FIG. 12 is a flowchart showing a process in which a telop recognition unit receives an input of a telop image and extracts domain information in the second embodiment.

FIG. 13 is an explanatory diagram showing a state of subsequent processing of the second information processing device which has received the domain information as a telop recognition result in the second embodiment.

FIG. 14 is an explanatory diagram showing a process of converting a speaker's utterance into a character string in the second embodiment.

FIG. 15 is a plan view showing how the result of voice recognition is displayed on the display in the second embodiment.

FIG. 16 is a flowchart showing a process in which a voice recognition unit receives a voice input and extracts domain information as a modification of the second embodiment.

FIG. 17 is a block diagram showing an overall configuration of an information processing system in a third exemplary embodiment of the present invention.

FIG. 18 is a flowchart showing an outline of control of the information processing system of the third example.

FIG. 19 is an explanatory diagram showing the processing of the first information processing apparatus in the third embodiment.

FIG. 20 is an explanatory diagram illustrating a process performed by a second information processing device according to the third embodiment.

FIG. 21 is a flowchart showing a process of extracting domain information in the third embodiment.

FIG. 22 is a plan view showing an example of display contents on a display in a state where a chat is being performed.

FIG. 23 is a block diagram showing an overall configuration of an information processing system in a fourth exemplary embodiment of the present invention.

FIG. 24 is a block diagram showing an example of a conventionally proposed information processing system.

[Explanation of symbols]

100, 300, 500, 700 Information processing system 101, 301, 501, 701 Input device 103 ₁ , 303 ₁ , 503 ₁ , 703 ₁ First information processing device 103 ₂ , 303 ₂ , 503 ₂ , 703 ₂ Second Information processing device 108, 308, 508, 708 Input history management unit 109, 309, 509, 709 Domain knowledge management unit 303 ₃ Third information processing device 711 Data processing device 712 Storage unit 713 Recording medium

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Kenji Sato             5-7 Shiba 5-1, Minato-ku, Tokyo NEC Corporation             Inside the company F-term (reference) 5B091 AA03 BA03 CB12 CB32 CC04                       CC15 CC16 EA01                 5D015 KK02 KK04 LL07

Claims (15)

[Claims] 1. A first information processing means for inputting predetermined information to perform first information processing, and second information for performing second information processing different from the processing of the first information processing means. A processing means, a domain information storage means storing domain-specific knowledge used for information processing of the second information processing means as domain information, and a domain information storage means based on a processing result of the first information processing means. Field selection means for selecting a field of the stored domain information, and specific domain information supply means for supplying the domain information of the specific field selected by the field selection means for information processing of the second information processing means. An information processing system comprising: 2. First information processing means for inputting predetermined information to perform first information processing, and second information for performing second information processing different from the processing of the first information processing means. A processing unit, a domain information storage unit that stores domain-specific knowledge used for information processing of the second information processing unit as domain information, and a first history as a processing history of the first information processing unit. First history storage means, field selection means for selecting a field of the domain information stored in the domain information storage means based on the first history stored in the first history storage means, and field selection An information processing system, comprising: specific domain information supplying means for supplying domain information of a specific field selected by the means for information processing of the second information processing means. 3. A plurality of information processing means for inputting and processing information respectively, a history storage means for storing information handled by any one of the plurality of information processing means as a history, and a plurality of information processing means of the plurality of information processing means. Domain information storage means for storing domain information used for information processing, history extraction means for searching the history stored in the history storage means and extracting a specific history, and using the history extracted by this history extraction means Domain information retrieval / extraction means for retrieving and extracting necessary domain information from the domain information stored in the domain information storage means, and domain information when any one of the plurality of information processing means performs information processing. Specific domain information supply means for supplying the specific domain information extracted by the search / extraction means as a reference for processing. An information processing system to be. 4. The field selection means includes a first history and a similarity calculation means for calculating a similarity between the respective fields, and the highest field among the similarities calculated by the similarity calculation means. 3. The information processing system according to claim 1, wherein the information processing system is selected. 5. A first information processing unit for inputting voice information to perform a voice translation process, and a second information processing unit for performing an example sentence translation process different from the process of the first information processing unit. Means, a domain information storage means that stores, as domain information, an example sentence for each scene used for translating the example sentence of the second information processing means, and a first history as a processing history of the first information processing means. A first history storage means, a scene selection means for selecting a scene to which an example sentence is applied based on the first history stored in the first history storage means, and a scene selected by the scene selection means. An information processing system comprising: specific domain information supplying means for supplying domain information as a candidate for example sentence translation. 6. A first information processing means for inputting a position and extracting geographical information as information corresponding to the input position, and a processing different from the processing of the first information processing means. Second information processing means for recognizing visual information, third information processing means for recognizing auditory information as processing different from the processing of the first and second information processing means, and the second information processing means. Domain information storage means that stores information for each scene used for recognition of the visual information as domain information, and first history storage means that stores a first history as a history of processing of the first information processing means, The first history information stored in the first history storage means is supplied as the domain information when the visual information is recognized, and the first domain information supply means is used as the processing history of the second information processing means. 2 shoes The second history storage means for storing the history, and the second domain information supply means for supplying the second history stored in the second history storage means as the domain information when recognizing the auditory information. An information processing system characterized by: 7. A first information processing means for analyzing chat contents, and a second information processing for executing an advertisement display processing for displaying an appropriate advertisement as processing different from the processing of the first information processing means. Means, a domain information storage means for storing advertisements to be displayed by the second information processing means as domain information for each field, and a first history as a processing history of the first information processing means No. 1 history storage means, type selection means for selecting the type of advertisement based on the first history stored in this first history storage means, and advertisement of domain information of the type selected by this type selection means An information processing system, comprising: a specific domain information supply means that is supplied as a candidate for. 8. A first information processing step for inputting predetermined information to perform first information processing, and a second information processing for performing second information processing different from the processing performed in the first information processing step. 2, an information processing step, a search step for searching for knowledge by field used for information processing in the second information processing step, using the history information obtained in the first information processing step, and a search step And a domain information supplying step for supplying the knowledge of the field searched in step 2 to the processing in the second information processing step. 9. A history storage step of storing a history of processing when information processing is performed by any of the information processing means, and another information processing performed by information processing means other than the information processing means. The search step for searching a history related to the current information processing from the processing history stored in the history storage step, and the search step from the domain information prepared in advance for use in the current information processing. A specific domain information extracting step for extracting specific domain information in relation to the history searched in step 1, and an information processing step for performing information processing this time using the specific domain information extracted in the specific domain information extracting step. An information processing method comprising: 10. A computer based on a first information processing for processing input of predetermined information, a second information processing different from the first information processing, and a result of the first information processing. Domain information extraction processing for extracting specific domain information from domain information as domain-specific knowledge used for the second information processing, and domain information extracted by this domain information extraction processing for the second information processing. An information processing program, characterized by causing a specific domain information supply process to be supplied to a computer. 11. A computer is provided with a first information processing for processing input of predetermined information, a second information processing different from the first information processing, and a history for storing a history of the first information processing. Based on the storage processing and the history stored by this history storage processing, the second
Domain information extraction processing for extracting specific domain information from domain information as field-specific knowledge used for information processing of the domain, and the domain information extracted by this domain information extraction processing is supplied for the second information processing. An information processing program, characterized by causing a specific domain information supply process to be performed. 12. A computer, comprising a plurality of information processings for independently processing information, a history storage processing for storing a history of the processing when any one of the information processings is performed, and a history storage processing by the history storage processing. Based on the stored history, the domain information extraction process of extracting specific domain information from the domain information as knowledge of each field to be used in the next information processing, and this domain information extraction process An information processing program, characterized by causing a specific domain information supply process to supply domain information for the information processing thereof. 13. The information processing program according to claim 10, wherein a plurality of the computers are provided and each of them executes independent information processing. 14. The domain information extraction processing is a history search / extraction processing for searching and extracting a desired history from the history stored in the history storage processing, and the history search / extraction processing. 13. The information processing program according to claim 11 or 12, further comprising domain information search / extract processing for searching and extracting specific domain information from the domain information based on the history. 15. The information processing program according to claim 14, wherein in the domain information search / extraction processing, specific domain information is searched and extracted from the similarity between the history and the domain information.