JP2791051B2 - Information processing system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to, for example, an information processing system for obtaining information, knowledge, and technology for solving problems in daily life.

[Prior art] At present, the knowledge involved in solving everyday problems, such as childcare, cooking, washing, etc., is in the form of publications, telephone consultations, computer searches and, best, listening to experienced he and herself. People get.

This information has various problems. For example, publications can be too informative and provide unnecessary information,
It becomes difficult to choose or causes psychological problems.

Telephone consultations lack privacy and do not provide personalized answer information.

[Problem to be solved by the present invention] In view of the above description, the present invention provides an information processing system that can obtain information more easily in order to obtain a conclusion on a problem.

SUMMARY OF THE INVENTION In view of the description, the present invention provides an information processing system that can obtain information most easily to obtain a conclusion on a problem and is appropriate for an operator of the system, that is, a user.

In view of the narrative, the present invention is a natural language that is commonly used by system operators (users), e.g., as requested, ordered, requested, so that information can be more easily obtained to reach a conclusion on a problem. By entering information such as
Provide an information processing system from which information can be obtained.

In view of the above description, the present invention provides an information processing system that can obtain information more easily in order to obtain a conclusion about a situation that can be understood from the description of the embodiments below.

[Means for Solving the Problem] In order to solve the above problem, the present invention is an information processing system for inputting a question described in a natural language and outputting a response to the question. Comprehension means for comprehending the purpose of the question asked, purpose / plan knowledge base means constructed so as to be searchable according to the purpose, and each element having a knowledge base consisting of at least a plan and an action corresponding to the purpose, Based on your understanding,
Having a means for accessing the purpose / plan knowledge base and formulating a plan and action corresponding to the purpose, and an output means for outputting the plan and action formulated by the formulating means as a response to an input question An information processing system is provided.

Embodiment FIG. 1 is a block diagram showing one embodiment according to the present invention. In the figure, IN is an input unit, which is composed of a keyboard or another input device, for example, a voice information input device, a digitizer, and for example, converts characters and symbol information expressed in a natural language as shown in FIG. It has character keys and function keys for input.

OUT is an output device, which includes a display, a printer, a voice synthesizer, and the like.

COM is an information processing unit, which has a processing unit CPU, a storage unit MEM, and an input / output interface PI. The processing procedure stored in the storage unit MEM, such as the procedure shown in FIG. Information arriving from the device via the interface PI, which is input in the form of a natural language used by the user of the system, for example, Japanese, or information stored in the storage unit MEM (this is the natural language The information is processed in a form or a language expression in the information processing unit, and the processed information is output in a natural language form via the interface PI.

Functional blocks realized by the information processing unit COM will be further described.

FIG. 2 is a detailed functional block diagram realized by the information processing unit shown in FIG. 1. In FIG. 2, ANA is a natural language analyzer, and the knowledge bases KB1, KB2 are arranged according to the intention of the user of the system of the embodiment. , KB3, KB4, the function of performing the processing shown in FIG.
It has a grammatical and semantic analysis function for phrases, a context analysis function, and an analysis function for the purpose plan of the user of the device (the person seeking advice). The purpose, plan, action, and action from the input natural language The information to be converted into information of a person or the like (the expression of such information is in the system, so an information expression convenient for the system is taken) is derived.

Prc is a switch, an inference engine for advice I
It instructs whether to execute E or the management unit AD of the data service.

IE is an inference engine that iterates through the expert knowledge base EKB to search and execute plans.

The inference engine IE requests the user to input new information from the input unit IN until an appropriate inference is obtained. The response to the user is performed by converting the information through a natural language synthesizer SYS so that the information of the internal expression has the same expression level as the input, and displaying the information in the form of a natural language or outputting the speech.

The AD is a management unit that receives a request from the natural language analyzer ANA or the natural language synthesizer SYS to search for specific information in the personal database IDB or write new personal information in the personal database IDB.

KB is a knowledge base, a knowledge base KB1 for a general knowledge concept shown in FIGS. A1 and A2, a knowledge base KB2 for a domain (field, domain) based concept shown in FIG. 4B, and a text creation concept shown in FIG. 4C. The knowledge bases KB3 have a dictionary structure as shown in the figure, and their meanings are derived from search words, that is, general words, and related information is derived. It is accessed as needed by the management unit AD, the natural language analyzer ANA, and the natural language synthesizer SYS.

The IDB is a user's personal database KB4 having a configuration as shown in FIG. 4D and accessed by the management unit AD. By providing such a personal database IDB, this system can output information more suitable for the individual, and the system can respond sufficiently even if the information from the input unit IN is made simpler information. You can do it.

EKB is a domain (field, domain) specific knowledge base,
Having expert knowledge to solve a specific area, for example, a problem of daily life, for example, childcare, cooking, choice, etc.
FIG. E shows the data structure.

SYS is a pre-language synthesizer, which performs processing as shown in FIG. 3B, ie, a sentence construction function / speaker,
It has a contextual search function, grammatical interpretation function, and word selection function in consideration of situations (speech mode) such as listeners, questions, instructions, and requests. Such a natural language synthesizer SY
S receives information from the inference engine IE and the management unit AD, and executes it using the knowledge bases KB1, BK2, and KB3. The natural language synthesizer SYS sorts and arranges information according to the knowledge base provided in the knowledge base KB3. Thereafter, the natural language synthesizer SYS converts the output information into the language of the user of the system (for example, Japanese in this case), and converts the output information into, for example, a Japanese conversational tone shown in FIG.
Output from T.

The operation of the embodiment having the above-mentioned configuration will be described below in the fifth,
This is performed with reference to FIGS.

First, as an explanation of the operation of the embodiment, an example of inputting information that "night crying does not stop" will be described.

This information is first input from the input unit IN, and is introduced into the parser Parser of the Japanese analyzer ANA. It is interpreted word by word, grammatical processing is performed for each phrase in consideration of the knowledge base EKB on childcare, and context analysis is performed. From the input words, "crying at night" is "crying at night" This is searched by the knowledge base EKB to obtain an inference output. The fact that "night crying does not stop" remains in the system as a common phenomenon for children.

The result, then, is the understander UNDERSTAND
Passed to ER. UNDERSTANDER started,
The subject missing as input information of this sentence is searched. In other words, who is the subject of the action? And
Infer that it is a child. That is, the child of the user is crying excessively at night. Thus, it finds a situation where it is in trouble for the user or parent.

From its domain-based concept KB1, then the system can remedy this situation.

The plan constructor plan constructor then creates an action plan for this purpose with reference to the parenting expert knowledge base EKB. The switch Prc then controls the direction of the remaining information according to the plan given by the plus constructor.

In this case, because we refer to the expert system,
The information concerns the expert system inference engine IE.

The inference engine IE exists in the function to solve the problem.

We need to find the cause of human problems. Therefore, the information processing section searches the knowledge base EKB.

Then, search the knowledge base to obtain the appropriate information for the purpose.

As in the example shown in FIG. 7, in this case, the system knows from the knowledge base or database that the reason for this problem was "hunger".

Therefore, before searching, it is confirmed that the same reason has been deleted.

This information is sent to the text generator preparetotor of the natural language synthesizer SYS for transfer to the user. In this case, the text creator preparetotor has already determined output items, for example, items such as purpose, plan, action, and actor. So, just send it to the generator GENRATOR. The generator GENRATOR converts this message from the expression in the system to the language expression of the operator who understands the natural language, and as shown in FIG. 7, "During this time --- what is it this time?"

Then, it is sent to the output device OUT. In this case, a display and a voice synthesizer are suitable.

In this regard, the user must obtain a response to the user's input information and respond to the system.

In this case, as shown in FIG. 7, the user responds, "Well, are you hungry?" For this, new information for the system is input from the input IN.
The procedure similar to the above-described processing is started again. Except at this time, the system is a process that has a standard and already fulfills the purpose of relieving night crying.

So, Anda Standa identifies the new purpose as a sub-purpose of the previous purpose.

The rest of the process is performed as before. The initial purpose of relieving night crying is satisfied.

Instead, the initial input was "When did Aki come to walk? ”Is controlled by the system. You can request to read the user's database IDB.

Whether the information flow is to be inquired to the inference engine IE is determined by the information processing unit based on the contents of the created plan, and if it is NO, whether to write to the personal database IDB, if NO, read the personal database IDB go to. Therefore, the flow 2 in FIG. 6 is selected.

On the other hand, if "Aki-chan has been vaccinated now" is entered, the next judgment is made based on the information of the target plan already obtained, that is, whether to inquire the inference engine IE or not, and write it to the personal database IDB If the answer is YES, it is interpreted as a request to write this information to the user's database IDB.
Is selected and performed. The system then processes the user's desired command and has no way to give too many instructions. For that purpose, the search is based on natural language processing.

FIG. 8 is a knowledge base for explaining another embodiment of the present invention. If such a knowledge base is provided, information on problem solving such as cooking as shown in FIG. As such, the system of the present invention is suitable for solving not only childcare problems but also other problems.

[Effects of the Invention] According to the present invention, as a response to an input question, a plan and an action in consideration of the purpose of the question are output, so that an appropriate response that meets the intention of the operator's question can be obtained. effective.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention, FIG. 2 is a detailed diagram of a CPU shown in FIG. 1, FIG. 3 is a diagram showing a processing procedure shown in FIG. 1, and FIG. FIG. 4 is a detailed view of the analyzer and the parser shown in FIG. 2, FIG. 4A2 is a detailed view of the synthesizer and the generator shown in FIG. 2, FIG. 4B is a view showing the contents of the knowledge base KB1, and FIG. 4D is a diagram showing the contents of the knowledge base KB3, FIG. 4E is a diagram showing the contents of the knowledge base KB4, FIG. 5 is a detailed block diagram shown in FIG. 2, and FIG. Is a diagram illustrating a control procedure, FIG. 7 is a diagram illustrating the operation of the embodiment, FIG. 8 is a diagram illustrating another embodiment, FIG. 9 is a diagram illustrating another embodiment, KB. Base IDB: Personal database IE: Inference engine IN: Input section OUT: Output section COM: Information processing section

Continuation of the front page (56) References Tanaka, Fuchi, "Artificial Intelligence Handbook Volume III", Kyoritsu Shuppan Co., Ltd., pp. 667-690 (Showa 59-7-5) Takano et al. Topic Management Method ", Information Processing Society of Japan Research Report (AI-57-2), Vol. 88, No. 18, 1988 (Showa 63-3-10) Kato and Nakagawa, "Understanding Intention in Question Answering and Managing Topics", Information Processing Society of Japan Research Report (NL-58-6), Vol. 86, No. 79, 1986 (Showa 61-11-21) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 17/30 G06F 9/00

Claims (6)

(57) [Claims] An information processing system for inputting a question described in a natural language and outputting a response to the question, comprising: an understanding means for understanding the purpose of the input question; A goal / plan knowledge base means constructed and having a knowledge base consisting of at least a plan and an action corresponding to the goal, and accessing the goal / plan knowledge base based on the understanding by the understanding means; An information processing system comprising: a planning unit that formulates a plan and an action corresponding to the above; and an output unit that outputs the plan and the activity formulated by the planning unit as a response to an input question. 2. The information processing system according to claim 1, wherein said objective / planning knowledge base means is a knowledge database means constructed for each field. 3. The information processing system has personalized database means constructed so as to be searchable in units of actors, the understanding means understands actors, the formulating means accesses the personalized database means, The information processing system according to claim 1, wherein a plan and an action corresponding to (1) are formulated. 4. The information processing system according to claim 1, wherein said question is described in a natural language, and said understanding means includes means for grammatically analyzing the input described in the natural language. 5. An output means comprising: a sentence structure knowledge base means capable of retrieving a sentence structure knowledge necessary for configuring each element of a sentence composed of a natural language; and a sentence structure knowledge base means by the formulating means. Means for retrieving the sentence composition knowledge corresponding to a search based on the determined plan and action, and conversion means for converting the plan and action to natural language using the extracted sentence composition knowledge The information processing system according to claim 1, wherein: 6. The information processing system according to claim 1, wherein said objective / planning knowledge base means has a user-specific knowledge base, and said user-specific knowledge base is updated.