JPH11203325A - Database generating method, record medium in which its program is stored and record medium recording group of data generated by the generating method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make contents of database to be stored sufficient and to obtain accurately labelled data by labelling the data inputted later based on data which are labelled and already registered and flexibly labelling the data further as learning the data on an interactive basis. SOLUTION: When inputted information from an information input means 1 is analog, it is converted into digital data in an inputted information processing means 2. When the inputted information is the one which is treated by a decomposition element decomposing means 3, the information is transmitted as it is to the decomposition element decomposing means 3. The inputted information is decomposed into the decomposition element formed by assembling it into a minimum unit to be decomposed by referring to the database 8 by the decomposition element decomposing means 3. And an inquiry whether the decomposition element obtained by the decomposition element decomposing means 3 is proper or not is made by a decomposition element interactive means 4 and when it is proper, it is defined as the decomposition element as it is. In addition, an inquiry whether a label obtained by a labelling means 5 is proper or not is made by a labelling interactive means 6 and when it is proper, it is defined as the label as it is.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for creating a database for inputting information, labeling the information, and storing the information as data in a database, a storage medium storing the program, and a data group created by the creating method. Related to a recorded medium.

[0002]

2. Description of the Related Art Conventionally, labeling of information input to a computer is performed by decomposing input information as it is into a minimum unit, a decomposed element, attaching a label or the like to the decomposed element, and storing data in a database. I have.

[0003]

In the conventional labeling method, only the labeling is fixedly performed on the decomposed element, and the labeled data is fixed. For this reason, existing stored data should be labeled with different data based on the labeled data, or the registered data should be rewritten based on the later data to enhance the content. It was not possible to respond flexibly to input information.

According to the present invention, learning is performed in an interactive manner by labeling data inputted later and rewriting original data in accordance with the data based on the data which has been labeled and registered. Method for creating a database capable of expanding a database in a format in which data can be organically associated with each other by flexibly labeling data, a recording medium storing a database creation program, and data created by the above steps It is intended to provide a recording medium on which a group is recorded

[0005]

According to the present invention, there is provided a database creation method for inputting information and storing the information as data in a database, and holding a program for inputting information and creating a database for storing the information as data in the database. In a recording medium and a recording medium in which information is input, data is processed, and a data group to be processed is recorded. A: information input means B: input information processing means C: data obtained by the information processing means is decomposed into decomposed elements D: labeling means for performing labeling processing of the decomposition element E: labeling dialogue means for performing labeling dialogue processing F: registration means for registering the labeled data in the information holding means

The information input means includes, if necessary, a discriminating means for discriminating whether the form of the information is digital or analog. Just convert it.

[0007] The information input means includes a discriminating means for discriminating the type of information (media type) as necessary.
The type of the information is at least one information selected from characters, voices, sounds, images (still images and moving images), vectors and cases, and an input device corresponding to each type of information may be specified.

[0008] The information processing means converts the information that is not data that can be processed by the decomposer into data that can be processed by the decomposer and sends it to the decomposer and leaves the data that is already processable as it is. May be sent to the decomposer.

The decomposing element decomposing means decomposes into a minimum unit which can be decomposed or a unit which can be assigned a sign by a set of the minimum units. The decomposing element decomposing means has reference means for referring to the decomposing elements registered in the database, and the decomposing element may be decomposed with reference to the registered decomposing elements.

This decomposition element decomposition means is provided with decomposition element dialogue means with the user as necessary. Then, the decomposing element dialogue means compares the obtained decomposing element with the format of the obtained decomposing element to determine whether or not the decomposing element has been properly decomposed. If not, indicate to the user in the output means to that effect. B: Ask the user whether the decomposition factor is correct. C: If it is correct, use the decomposition factor as it is. D: If it is not an appropriate decomposition factor, output it to the output unit. It may be configured to include the steps of prompting the user to input a correct decomposition element and making the input object a correct decomposition element.

The labeling means may obtain a label for the obtained decomposed element by referring to an already registered database and add the label. When the label to be given is not found, the fact is output to the output means, and the user is prompted to input a correct decomposing element, and the inputted thing is given a new new label and the labeling dialogue means is applied to the obtained decomposing element. For the label: A: A means for determining whether the label is appropriately labeled is provided. If the label is not properly labeled, the output means outputs the fact. B: Ask the user whether the labeling is correct. Label: D: If the labeling is not appropriate, the fact is output to the output means to prompt the user to input the correct label, and the input thing is set to the correct new labeling. E: The label of the reference database and the input by the user. If the label is different, the newly entered label is set to X, and the label already registered Is set to Y, data having a configuration in which X is added to a label common to both is displayed on the output means, and the user is asked for correctness. If correct, the data to which the label X has been added is already registered in the database If the data to which the label Y is added is not correct, the part already registered in the database is kept as it is. A labeling dialogue means comprising the above steps is provided.

This labeling means refers to a database registered for the decomposed element, and if there is no appropriate label, outputs the fact to the output means to the user, prompts the user to input a label, and checks the input result. It can be a label.

When the input information is a language composed of speech or text, the minimum decomposed element may be a character. Since the input information is a document, the label is an attribute that is a phrasing symbol including particles such as particles, auxiliary verbs, conjunctions, adverbs, parts of speech such as punctuation marks, and symbols such as spaces. A syntax composed of a plurality of minimum decomposers (characters) is regarded as one parser, and the types of attributes included in its own parser and those included in other parser are compared. It finds another parser that has all the attribute types, displays the attributes that are not in itself in the other parser, and displays them on the output means, and asks the user whether the attribute is correct. In the wording symbol,
If the question contains words such as who, when, where, why, and how, and if any of the decomposed elements of the input data is missing, the user will display the missing words on the output means. Then, input of the attribute may be prompted.

In addition, a similar word dictionary is provided, and when the data of the decomposed element is incomplete, according to the judgment result of the user's correctness,
What is necessary is just to make incomplete data of a decomposition element into correct data.

[0015] In the interactive means, an editing unit for editing the disassembly or data necessary for labeling into a format understandable to the user and displaying on the output means, The selected or input result may be used as a formal decomposition factor or label.

The data holding means for holding data in the database is an object database means. The data is object data. When there are a plurality of common labels between objects, only the common portion X is taken out to generate a new object. Then, an ID may be assigned to the object from which the common part X is extracted, and that part may be represented by the ID.

The means for registering in the database is classified according to the field of data and registered for each field. If necessary, data processing means registered in the database is provided. Then, the data processing means refers to the database, changes the contents of the registered data referred to in the labeling and labeling processing of the data by interactive processing with the user, obtains the data field in the labeling process, and determines the data field. Referring to the data in the field, the data is labeled and the contents of the registered data are updated at the same time.

First, the user's own database holding the knowledge data of the user is referred to. If labeling cannot be performed by the user's own database, the database of another user may be referred to. If labeling cannot be performed by referring to another person's database, a list of intellectuals having knowledge that enables labeling is obtained and displayed.

When the handling information is a language, the information processing means recognizes the input word type, and converts the input word type into a different word type.
Then, the difference between the pattern of the input information of the user and the pattern of the database of the other person is identified, the pattern of the input information of the user is replaced with the pattern of the database of the other person, and the database of the other person is searched.

When registering data for each field, in order to determine which field the input information is similar to,
Calculate the similarity with each field, use the field with the closest similarity, and if the similarity does not reach a certain threshold, determine the field based on the directory name or file name used immediately before What should I do?

The outline of the basic flow of the present invention will be described once again.
This will be described with reference to the drawings. The flow is shown in FIG. First, information is input by the information input means 1. If the input information is analog, the input information processing means 2 converts the information into digital data. If the input information is information that can be handled by the decomposing element decomposing means 3, the information as data is sent to the decomposing element decomposing means as it is.

Since the type of input information (media type) may not be known, one information input means includes language, voice, sound, character, image (still image, moving image) vector information, and the like. What is necessary is just to provide a determination means for determining the type of information (media type).

In the decomposing means of (3), a decomposable minimum unit which can be decomposed with reference to the database 8 or a decomposed element which is collected into a decomposable unit by collecting and reconstructing the decomposed elements of this minimum unit is obtained. Decompose. If necessary, the user asks the user whether the decomposed element obtained by the decomposed element decomposer is appropriate or if it cannot be decomposed by the decomposed element decomposer, and if appropriate, If it is unsuitable or the number of decomposed elements is insufficient, the user is prompted to input the result, and the result is used as the correct decomposed element.

In the labeling means 5, a label is given to the decomposed element obtained by referring to the database 8. Then, in the labeling dialogue means 6, it is asked whether the label obtained by the labeling means is appropriate, or if labeling is not possible, asks the user whether the labeling is appropriate. If the label is appropriate or the label is insufficient, it prompts the user to input it and asks the user to input it. The result is the correct label.
Then, the result selected as described above is registered in the database 8 by the registering means 7.

[0025]

[Embodiment 1] A further description will be given below based on an actual example. In FIG. 2, reference numeral 10 denotes data processing means for giving meaning to input data. Numeral 11 denotes data input means for inputting text data such as text, multimedia data such as images and sounds. Reference numeral 12 denotes an input data holding unit for temporarily holding input knowledge data.

In the data processing means 10, reference numeral 13 denotes a knowledge data creation unit which decomposes input data and makes a meaning by referring to the knowledge data holding unit 20.
Reference numeral 14 denotes a decomposer for decomposing input data into decomposers of the smallest unit of meaning. Numeral 15 denotes a case recognition unit for assigning an attribute representing the meaning to the decomposed element of the input data. Is an attribute name by a general name such as, or an attribute value representing the specific content thereof).

Reference numeral 16 denotes a knowledge data reference unit for referring to the knowledge data held in the knowledge data holding unit. Reference numeral 17 denotes a registration unit for registering the data in the data holding means designated by the attribute. Reference numeral 20 denotes a data holding unit (hereinafter, referred to as a knowledge data holding unit), which holds knowledge data (case recognition data and data of a dictionary unit).

Reference numeral 21 denotes a case recognition data holding unit for holding meaningful input data. For example,
In the case of the input data "I go to school.", The decomposition factors are "I", "Ha", "School", "Nii", "Go", "." An ID is assigned to the sentence "I go to school." In addition, "I" goes to Nominative, and "No" to "Nominative" and "School" as attribute names and attribute values representing the case of "ha". On the other hand, an attribute name and an attribute value representing "Ni" are held as an object case, an action case is displayed for "Go", and an attribute name and an attribute value representing punctuation are stored as a case of ".".

Reference numeral 32 denotes a dictionary unit for holding attribute names and attribute values, and attribute values (nominative, objective, name, age, etc.) corresponding to the attribute names set in the input data (nominative nominative). , A decomposer representing the object case, and a decomposer representing the name, "XXX", and a decomposer representing age, such as "40 years old." , The contents are enriched. Reference numeral 41 denotes an output means, which is a display for displaying an image or a sound output device such as a speaker or a sound amplifier for outputting sound. The operation of the configuration of FIG. 1 will be described.

Data such as text data of a sentence and multimedia data are input to the data input means 11. The input data is temporarily stored in the input data storage unit 12. In the knowledge data creating unit 13, a decomposition element decomposition unit 14 decomposes input data into decomposition elements. For example, in the case of the input data “I go to school.” (Assumed to be syntax A), the decomposition factors are “I”, “H”, “School”, “N”, “Go”, “.”. . Decomposition of input data into decomposers is
For example, the data decomposed by the decomposer is displayed on the display unit, and the operator confirms the data. Inappropriate disassembly is performed while the operator and the display unit interact with each other, for example, by making corrections on the display screen by the operator. Alternatively, a voice output device asks the operator by voice whether the decomposition element is correct, and prompts the operator to input correct decomposition data.

The data decomposed into decomposed elements is output by an output unit 41.
Will be displayed. When nothing is registered in the dictionary unit 32, “I”, “ha”,
Since “school”, “ni”, “go”, and “.” Are displayed, the operator assigns “Nominative” and “School” to “I” as attribute values indicating the nominative case and “ha” case. On the other hand,
For "Ni", an objective case is set as an attribute value representing the case, "Go" is an action case, and an attribute name and an attribute value representing punctuation are set as a case of ".". Then, the registration unit 17 registers the attribute value “ha” for the attribute name “Nominative” and the attribute value “O” for the attribute name “Objective” in the dictionary 31. Also, for "I", an ID is added and registered, and the nominative is registered as the attribute value. The ID is also assigned to “school”, and “object” is registered in the dictionary unit 32 as the attribute value. Then, an ID is assigned to the whole input data “I am going to school.”, An attribute name and an attribute value are assigned to each decomposition element, and stored in the case recognition data holding unit 21.

Further, in the state where the attribute value “ha” is registered for the attribute name “Nominative” and the attribute value “Ni” is registered for the attribute name “Objective Case” in the dictionary unit 32, Is input, the decomposing element decomposing unit 14 decomposes the input data into decomposing elements, and the knowledge data referencing unit 16 searches the knowledge data holding unit 20 to determine that nominative The goal case for "school" and the behavioral case for "go" are obtained, and the case recognition unit 15 assigns the attribute names (case recognition data) obtained in this way to each of the decomposed elements, thereby inputting. Automatically give meaning to data. The meaning-added knowledge data is stored in the knowledge data holding unit 2.
0 is stored.

Further, in this state, when a sentence "Yamada-kun is going." The “kun” nominative case and the “going” behavioral case are displayed on the output means 41. Then, as compared with the already registered syntax A, there is no purpose case, so the case recognition unit 15 displays a question "Where?" By outputting to an audio output device, it points out the lack of input data and prompts the input of the missing data.

Then, when the operator inputs "to school" by voice input using a keyboard or a microphone,
The sentence “Yamada goes to school.” Is disassembled into decomposed elements, confirmed and identified (with an attribute name given), and stored in the knowledge data holding unit 20. Alternatively, when the knowledge data holding unit 20 is set with a measure indicating a means such as “on foot” or “on a bicycle” and a time scale indicating time such as “8:30”, the syntax B Further, "what?" Indicating shortage of measure, "what time?" Indicating shortage of time, etc. are displayed on the output means to prompt the operator to input data. Then, if "by bicycle" or "at 8:30" is entered,
The knowledge data with case recognition indicating that "Yamada will go to school at 8:30 by bicycle" is held in the knowledge data holding unit 20. The structure of the case recognition data at this time is "Yamada-kun"
Registered as "8:30" time, "school" purpose, "bicycle" means. In this way, a sentence created with a rich content based on the syntax B is assigned an ID and registered.

Further, according to the present invention, "name", "age", etc. can be set as attribute names of case data, so that "data (or ID) named Yamada-kun" As “name”, attribute value as “Yamada”,
Enter attribute value “40” for attribute name “age”
Register in the dictionary unit 31. Further, when a case such as "no" used as the nominative case and "wo" used as the nominative case appears, the attribute name "nominative" is used for the attribute name "nominative" and the " Is added. Also, attribute names are given to “no” and “wo”, respectively, and registered with IDs. The registered "no" and "wo" are added later when a new usage appears, and the content of the data is enriched.

In the present invention, a method of setting an attribute name in a decomposed element is, for example, when determining whether or not “ha” is a nominative case, statistics of the attribute name in which “ha” is used are taken, and the frequency is calculated. Select the higher one and set it. In addition, various methods can be used for determining the probability, such as obtaining a probability and selecting a high probability. As described above, the present invention refers to the meaning of the input data by referring to the knowledge data such as dictionaries and sentences which are already stored with the meaning. Also, knowledge data such as dictionary contents and sentences that have already been created are rewritten based on knowledge data entered later, and the contents are constantly changed in a meaningful manner in response to information changes. It is.

[0037]

Embodiment 2 FIG. 3 shows an embodiment of the system configuration of the present invention. In FIG. 3, reference numeral 11 denotes data input means. 4
Reference numeral 1 denotes an output unit such as a display, a printer, an audio output device, a video device, and the like. 511 is a computer. Reference numeral 52 denotes an input / output interface, which is an interface with the data input means 11, the output means, and the external storage device 71.

A data processing unit 53 processes input data (including the data processing means 10 and the data holding means 20 shown in FIG. 1). 61 is a CPU.
62 is a RAM. 63 is a ROM. 64 is an internal magnetic disk. (The above is the data processing unit 53) 71 is an external storage device. In the data input means 11, reference numeral 82 denotes a keyboard and a mouse. 83 is a voice input device.

A graphic data input device 84 inputs graphic data from an image reader or the like. 85
Is an electronic camera. 86 is a CD-ROM drive. Reference numeral 87 denotes a video input device such as a video camera. 1
10 is a storage medium, a magnetic disk or CDR
OM and the like.

Reference numeral 211 denotes a program for installing the program recorded on the storage medium 110 on the internal magnetic disk 6453. Reference numeral 121 denotes a knowledge data creation program, which is a program of a knowledge data creation unit. A decomposition element decomposition program 122 is a program for a decomposition element decomposition unit. Reference numeral 123 denotes a case recognition program, which is a program of a case recognition unit. Reference numeral 124 denotes a knowledge data reference program, which is a program of a knowledge data reference unit.

Reference numeral 131 denotes an editing program, which is a program of an editing unit. 132 is a registration program, which is a program of the registration unit. FIG. 4 shows an embodiment of the apparatus configuration of the present invention, and does not require, for example, all of the data input units.

In FIG. 4, reference numeral 11 denotes data input means. Reference numeral 12 denotes an input data holding unit which temporarily holds input data (a portion included in the internal magnetic disk 64 in FIG. 2). 13 'is a knowledge data creation program (stored on the internal magnetic disk 64 in FIG. 2). 14 'is a decomposition element decomposition program.
15 'is a case recognition program. 16 'is a knowledge data reference program.

Reference numeral 20 denotes a knowledge data holding unit (FIG. 2 is included in the internal magnetic disk 64, but may be included in the external magnetic disk device). 41 is an output means. Reference numeral 51 denotes an editing program for editing knowledge data to be displayed on a display (a program stored in the internal magnetic disk of FIG. 2). Reference numeral 55 denotes a registration program for registering knowledge data in the knowledge data holding unit 20 (a program stored in the internal magnetic disk 64 in FIG. 2).

Reference numeral 56 denotes an index creation program for automatically creating an index of data to be registered in the knowledge data holding means 20 (a program stored in the internal magnetic disk 64 in FIG. 2). The operation of the configuration in FIG. 4 will be described later. FIG. 5 shows the configuration of the case recognition unit and the editing unit of the present invention. In FIG. 5, reference numeral 15 denotes a case recognition unit (which recognizes a case of a syntax using a case recognition program and a CPU, a RAM, and the like).

Reference numeral 91 denotes a case setting unit which creates data such as name, age, height, and personality when an object is created for a person (for example, by a case setting program and a CPU, a RAM, etc.). Is composed). Reference numeral 92 denotes a wording symbol setting unit, which sets a case in a decomposed element such as sentence data, for example (comprising a wording setting program, a CPU, a RAM, and the like). Reference numeral 93 denotes a syntax wording symbol setting unit that sets a wording symbol (attribute name such as a case) for a syntax including a plurality of decomposable elements (consisting of a syntax wording symbol setting program, a CPU, a RAM, and the like). Is what is done).

Reference numeral 94 denotes a common part extraction object creating unit which creates a new object by extracting a common part from a plurality of objects (a common part extracting object creation program and a CPU, RAM, etc. Is). Reference numeral 51 'denotes an editing unit (configured by an editing program, a CPU, a RAM, and the like). Reference numeral 97 denotes a question creating unit which creates a question for editing knowledge data based on data including the case given by the case recognition unit. For example, when data such as "go to school." Is input, a question such as "who is?"
It is composed of a PU, a RAM, etc.).

Reference numeral 98 denotes an answer input unit for editing knowledge data by inputting an operator's answer to a question (question answer program, CPU, RAM).
Etc.). 6 and 7 show examples of the data configuration of the knowledge data of the present invention (examples of objects stored in OODB). The case data represents the descriptive content of the attribute name, and may be a description part or the like. Figure 6
In FIG. 7, reference numeral 111 denotes an example of an object to which case data is added. Name A, age 40, height 1
This is an object representing a person who is 68 cm, weighs 68 kg, and has a perfect personality.

Reference numeral 112 denotes an example of a phrase symbol object, which is an example of a phrase symbol "ha" representing a case.
The object name is "ha" and the attribute name is "representative case".
The attribute value is “Nominative”. Reference numeral 113 denotes an example of a wording symbol object, which is a wording symbol “Ni” representing a case.
This is an example. The object name is “ni”, the attribute name is “representative case”, and the attribute value is “objective case”. Reference numeral 114 denotes an example of an object provided with a wording symbol. It is an object that has been decomposed into sentence data of sentence data "I go to school." The nominative case is set as the case. After that, for example, when a sentence "The person gave me a book." Is input, a purpose is set and registered for "I" decomposed into decomposed elements.

Reference numeral 115 denotes an example of an object provided with a wording symbol. This is an object that sets a purpose for “school” in the sentence data “I go to school.” Reference numeral 116 denotes an example of an object in which case is assigned to sentence data. Attribute value "I" for attribute name "Nominative" for attribute name "Nominative", attribute value "School" for attribute name "Objective case", and attribute value for attribute name "Action case" The value “go” is set, and the ID of the object is set. Reference numeral 117 denotes an example of an object representing a behavioral case.
The behavioral case is set as the attribute value for the data of “coming to work”. Is set. And, for example, "I am going to work at 8am." Is input, "Where?" And "How?" Are displayed on the display.

FIG. 8 shows an example in the case of multimedia data. It is a coffee maker object. FIG.
, An object 218 of the multimedia data is an object of the coffee maker. The name of the object is the coffee maker, the ID of the video that describes how to use it (video 1), the model name representing the shape is cylindrical, and the ID of the image data of the shape is Figure 1, the inventor (author)
It holds that the ID of the image data of the photograph is photograph 1.

Reference numeral 219 denotes image data of a coffee maker, which has an ID shown in FIG. 220 is a photograph image data of the maker of the coffee maker,
This indicates that the photograph has an ID of photograph 1. 221
Shows a video of how to use a coffee maker. Video 1
ID indicates that a video of a method of using a coffee maker is recorded.

The operation of the configuration shown in FIG. 4 will be described. Data such as text data and multimedia data is input from the data input means 11. The input data holding unit 12 temporarily holds input data. The knowledge data creation program 13 'decomposes input data into decomposers. Decomposition into decomposed elements is performed by the editing program editing the input data decomposed by the decomposed element decomposition program and displaying it on the display, prompting the operator to determine whether the data is correct or not, and inputting confirmation data of correctness or correct decomposition data by the operator. Prompt. Alternatively, the user is asked whether the data decomposed by voice is correct or not, and is prompted to input a correct decomposed element by voice. Then, the operator checks the contents and, if necessary, makes corrections on the display. The contents are reflected in the decomposition element decomposition program through the editing program,
The correctly decomposed data is passed to each recognition program 15 '.

Then, the knowledge data reference program 16 '
Refers to the knowledge data holding unit 20, refers to the syntax of the object of the knowledge data already created, the case data assigned to the object, and the like, and obtains the attribute to be added to the decomposed element. The case recognition program 15 'sets the case (attribute name) obtained for the decomposition element. Then, for example, when no case is found (for example, when the target case “ni” is not registered in the object, etc.), the fact that the case is not registered is displayed on the output unit 41 by the editing program 51, and Prompt for case. When the case is entered, the case is registered in the case object (or table or the like). Alternatively, if the input data has a missing case (for example, if the number of cases (the number of attribute names) recorded in the object obtained from the disassembly of the input data is larger), the editing program 51 outputs the knowledge. A question of the content represented by the attribute name of the object-rich portion of the data holding unit is created, and output means 4
1 is displayed. Alternatively, the sound is output. For example, if the input data has no measure and the object selected in the knowledge data has a measure of "bicycle", "is it a bicycle?" Is displayed on the display and the operator is asked.
Alternatively, the question is output from the voice output device as voice data.

According to the operator's answer, if the operator inputs "high" or another attribute value by keyboard input or voice input by a microphone, knowledge data supplementing each attribute value is created, and the registration program 55 An index is generated for the knowledge data, and the knowledge data holding unit 20 holds the index. When the operator inputs “no”, an index is generated without supplementing the case and registered in the knowledge data holding unit 20.

FIG. 9 is a flowchart of case recognition of the knowledge data creation program of the present invention. S1 The decomposition element decomposition unit decomposes input data into decomposition elements. S2 Determine the disassembly of the phrase (case) (“ha”,
"Ni" etc.). S3 Search the knowledge data holding unit to find a case. S4 It is determined whether there is a case. If there is a case, the process of S5 is performed, and if not, the process of S6 is performed. S5: A case is given to the decomposition element before the wording symbol (for example, the nominative case is given to the decomposition element before "wa" and the "objective case" is given to the decomposition element before "ni"). S6 Since there is no rating, a question is asked to the operator (displayed on the display). S7 The case input by the operator is added to the case object (or data table or the like).

FIG. 10 is a flowchart of a program for giving a symbol representing a main behavior. "go",
"Who", "When", and "Who"
An object (or a data table or the like) having a case representing "what", "where", "why" or the like is provided. If the input data does not have the case, an inquiry is made to the operator. S1 The main behavior symbol is extracted from the knowledge data holding unit. S2, S3 It is determined whether there is a primitive of the knowledge data (data decomposed into decomposed elements). S4 The case (“Who”, “When”, “What”, “Where”, “Why”, etc.) of the symbol (“Go”, “See,” etc.) that represents the main behavior, and if each has an attribute value, If there is no such thing as "at 8:30?" Or "by car", display it on the display and ask a question to create primitive data ("at 8:30", "by car", etc.).

FIG. 11 is an explanatory diagram of the syntax wording symbol of the present invention. Since "I go to school." (Syntax A) includes a nominative case, a purpose case, and a behavioral case, "symantic case, purpose case, and behavioral case" are set as syntax wording symbols. 32
1 is a syntax wording symbol of the A syntax. 322 is an example of a syntax wording symbol of another syntax (B syntax). When the A syntax is input, reference is made to the syntactic phrasing symbol of the B syntax 322 having a configuration similar to that of the case included in the A syntax, and it is detected that the A syntax does not include a companion case ("who and?" Then, "Who is it?" Is displayed, a question is asked, and the operator is prompted for input.
At this time, if an attribute value such as “Yamada-kun” is set in the accompaniment, a question is also asked together with “Yamada-kun?”. If the objects of a plurality of syntaxes are similar to the syntax of the input data, the objects are selected according to the frequency or the probability. For example, when the attribute names are represented as 1, 2, 3, 4, and 5, the attribute configuration of the input data is [1, 2, 3], and knowledge data similar to the input data among the stored knowledge data. Are assumed to be [1, 2, 3, 4] and [1, 2, 3, 5]. In such a case, the number of times or probability at which the number of times has been selected is obtained beforehand, and the one with the higher number of times or probability is selected first.

FIG. 12 is a flowchart of a program for assigning an attribute using a syntax wording symbol. S1 Determine the syntax wording symbol. S2 Object syntax or syntax phrasing table (table with syntax phrasing symbol (case) for each syntax)
For similar syntax. S3 If there is no match and the case of the input data is not in the object or table, the process proceeds to S4. If there is no match and the case of the input data is small, the process proceeds to S5. If not, the process proceeds to S6. The match / mismatch is determined by comparing only the contents of the included cases, and the order is not important. S4 If there is no object or table, ask a question to enter case data. S5: Add a new case to the input data or register the syntax of the input data as a new syntax. S6 It is determined whether the order is different. If not (NO), the process ends. S7 If the order is different (YES), the leading disassembly is understood as coordination.

FIG. 13 is an explanatory diagram of a case where the common parts of the knowledge data objects according to the present invention are combined to create a new knowledge data object, and an explanatory diagram of the hierarchical structure of the objects. FIG. 13A is an explanatory diagram. FIG.
In FIG. 13A, reference numeral 421 denotes a state before the common part is extracted, and FIG. 13B shows a state after the common part is extracted. It is assumed that object 1 has attributes 1, 2, 3, and 4, and object 2 has attributes 1, 2, 5, and 6. Then, a new object 3 having the common attributes 1 and 2 is generated, and its index (ID) is set to K. And
The attributes 1 and 2 of the object 1 and the attributes 1 and 2 of the object 2 are represented by K which is the ID of the object 3.

FIG. 13B is an explanatory diagram of hierarchization of objects according to the present invention. In the present invention, every time a new object is generated as shown in FIG. 13A, an object having a common attribute is newly generated, and the objects are hierarchically managed by an index. FIG. 13 (b) shows the hierarchical structure of an object structured in a hierarchical manner by the index. 431 is an object of the highest hierarchy. 432, 433, and 434 are objects (metadata) of the next layer. 435,43
Reference numeral 6,437 denotes an object of the lowest hierarchy, which has primitive data. FIG.
6 is a flowchart for performing the operation of FIG.

S1 Search for a part common to a plurality of knowledge data objects. S2 Create a new object by grouping out common parts. S3 Remove the attribute of the common part from the original object,
Create an object with only the common part and assign a new ID. In the common part of the original object, the content of the data is identified by the ID of the generated new object.

FIGS. 15 and 16 are flow charts for creating an index of an object of knowledge data according to the present invention. When new knowledge data is registered, the program of the next flowchart operates to automatically generate an index (including some of the same contents as the flowchart of FIG. 13). S1 Determines whether new data is registered. If not, the process ends. If there is, the process of S1 'is performed. S1 'Search for data similar to the newly registered data. S2 It is determined whether the newly registered data is only the decomposition element or the case recognition data. If the data is only the decomposition element or the case recognition data (attribute), the process proceeds to S3. S3: It is determined whether there is an index of the same data as the newly registered data. If there is, the process proceeds to S4.

S4 The index is registered in the new data. S5 Create a new index for the newly registered data. S6 Search all data (knowledge), and if there is the same data as newly registered data, replace it with a new index. S7 If it is not only the decomposition element or the case recognition data in S2, it is determined whether or not there is an index of data of the case already recognized with the same structure (having the same attribute name or the like). If there is an index, the process proceeds to S8; otherwise, the process proceeds to S9.

S8: Register in the index. S9 Search for similar case recognition data (objects having a similar attribute configuration). S10: Search for similar case recognition data; if there is, go to S11; otherwise go to S12. S11 A new index is created in the common part, the common part of the case recognition data is registered with the newly created index, and similar object data is compressed. S12 Create a new index. In the present invention, the index of the object data is represented in a hierarchical structure, and the search starts from the object of the highest index.

FIG. 17 is a flowchart of the retrieval process according to the present invention. In the search, the search is started from the highest object of the object hierarchized by the index in FIG. S1 There is a search request. S2 Search for the highest-order data (index). S3 Determine whether there is an index. If found, the process ends, and if there is, the process proceeds to S4. S4 Search for data of the next hierarchy. S5 If there is next data, the next data is searched for in S4. If not, the process ends.

[0066]

Embodiment 3 FIG. 18 shows a basic configuration of another embodiment of the present invention. In FIG. 18, reference numeral 10 denotes data processing means for giving meaning to input data. 1
Reference numeral 1 denotes data input means for inputting text data such as text, multimedia data such as images and voices. Reference numeral 12 denotes an input data holding unit for temporarily holding input knowledge data.

In the data processing means 10, reference numeral 13 denotes a knowledge data generating unit which decomposes input data, obtains necessary attributes by referring to the knowledge data holding unit 21, assigns the necessary attributes to the decomposed elements, and obtains the meaning of the input data. (Attributes are attribute names that represent general names such as the name of a person, personality, etc., such as the case of a sentence (object case, nominative case, etc.), or the particles “ha”, “Yamada” of the nominative case Etc. means the attribute value such as a specific name). The knowledge data creation unit is configured by a CPU, a RAM, a knowledge data creation program, and the like.

Reference numeral 14 denotes a decomposing element decomposition unit that decomposes input data into decomposing elements having the smallest meaning.
The decomposition element decomposition unit includes a CPU, a RAM, a decomposition element decomposition program, and the like. Reference numeral 15 denotes a case recognizing unit that adds attributes such as a name to the decomposed element of the input data if it is a nominative case, a purpose case, or a person's name and gives meaning. The case recognition unit includes a CPU, a RAM, a case recognition program, and the like. Reference numeral 16 denotes a knowledge data reference unit for referring to the knowledge data held in the knowledge data holding unit. The knowledge data reference section
It comprises a CPU, a RAM, a knowledge data reference program, and the like.

Reference numeral 18 denotes an input requesting section again, which compares the format created by the attribute (case etc.) of the disassembly element of the input data decomposed into the disassembly element with the format created by the attribute of the already created knowledge data. A similar format is found, compared with input data, and an operator is requested to input a missing attribute. The input requesting unit is constituted by a CPU, a RAM, a requesting program again, and the like. Again, the input requesting unit makes the operator determine whether the decomposed element is correct or not. Reference numeral 19 denotes a format determination unit which finds a format that is the same as or similar to the format of the attribute of the disassembly element of the input data. The format determination unit is a CPU,
It is composed of a RAM, a format determination program, and the like. The format determining unit also compares the decomposed element decomposed by the decomposed element decomposing unit with the format and judges whether the decomposed element is correct or not.

Reference numeral 21 denotes a data holding unit (hereinafter, referred to as a knowledge data holding unit) which holds input data with meaning. For example, in the case of the input data “I go to school.”, The decomposition factors are “I”, “Ha”, “School”, “Nii”, “Go”, “.”. Then, an ID is assigned to the sentence "I will go to school." (Syntax A), and "I" is a nominative case. "" And "School" for the attribute name and attribute value "Ni", "Go" for the action case, and "." For the punctuation mark attribute name and attribute value. is there.

Reference numeral 31 'is an example of case recognition data, which holds an attribute name and an attribute value, and corresponds to the attribute name (primary case, purpose case, name, age, etc.) set in the input data. Attribute value (decomposition element “ha” representing the nominative case, “wo” of the decomposition case representing the objective case, etc.). Reference numeral 32 denotes an example of case data having attribute names such as “name” and “age”, and the attribute value of the attribute name “name” is “XXX” and the attribute value of the attribute name “age” "40 years old" and so on. The case data represents the descriptive contents of the attribute name, and includes a description and the like.

Reference numeral 31 denotes an output means, which is an audio output device including a display or a speaker for outputting audio, an audio amplifier, and the like. The editing unit 51 edits data for displaying data necessary for meaning on an output unit. The editing unit includes a CPU, a RAM, an editing program, and the like. Reference numeral 55 denotes a registration unit for registering a case set as a decomposition element in the dictionary unit. The registration unit includes a CPU, a RAM, a registration program, and the like.

The operation of the configuration shown in FIG. 18 will be described (see FIG. 19). Data such as text data of a sentence and multimedia data is input to the data input means 11.
The input data is temporarily stored in the input data storage unit 12.
In the knowledge data creating unit 13, a decomposition element decomposition unit 14 decomposes input data into decomposition elements. For example, in the case of the input data "I am going." The decomposed element is edited by the editing unit as display data to be displayed on the display unit, and is edited on the display unit. Then, the operator judges the correctness and corrects the decomposed element when correction is necessary. The decomposed decomposed element is compared with the format by a format determination unit. If the decomposed element is different, a format determined to be correct is displayed, and the operator is prompted to input the correctly decomposed element.

The data decomposed into decomposed elements is sent to the case recognition unit 1
In step 5, the case is recognized (initially specified by the operator. If the information is already held in the knowledge data holding unit, it is referred to and automatically set). Then, an attribute value (me, go, etc.) is set in the knowledge data holding unit corresponding to the case (nominative case, action case, etc.). Therefore, FIG.
As shown in FIG. 9, for example, it is assumed that data "I go to school." The decomposition element decomposition unit converts the syntax A (41) into “I”, “ha”, “school”,
Decompose into decomposed elements like "ni", "go", "."
The attribute values that represent the nominative case for "I" and the case of "wa" are "nominative case", the purpose case for "school", the purpose case for "ni" as the attribute value representing that case, and "go" , An attribute name and an attribute value representing punctuation are set as the case of action and the case of “.”. Attribute settings are processed statistically and the one with the highest frequency is selected. For example, when recognizing “ha”, statistics of the frequency at which “wa” is used as an attribute name are obtained, and those having a high use frequency or a high probability are set as attributes.

Therefore, the knowledge data reference unit 16 converts the format of the syntax data A (the structure of the attribute value (“I”,
A search is made from the knowledge data holding unit for knowledge data having a format similar to the structure of the attribute name ("school", "go", etc.)) or the attribute name ("nominative", "objective", "behavioral", etc.). As a result, knowledge data 42 as shown is found. Then, for example, the attribute values of the two are compared, and there is no purpose case “school” included in the syntax A in the existing knowledge data 43. Therefore, the input requesting unit 18 and the editing unit 51 again ask “Do you go to school? Is created and displayed on the output means 31. The operator looks at the display and, if correct, inputs that fact through the data input means 11.
Then, the existing knowledge data 43 is rewritten to the knowledge data 42. For example, if "at school" is not correct,
When the operator inputs that fact through the data input unit, the knowledge data 43 is left as it is. Alternatively, for example, if the operator recognizes that the data in which the syntax of the input data is added to the syntax of the knowledge data 43 is correct, and in that state, it is determined that the data to which the attribute value is further added is not correct, Then, both the knowledge data 43 and the knowledge data 43 to which the attribute value is added are stored (see the description of the embodiment in FIG. 22).

FIG. 20 is a flowchart of the re-input request unit of the present invention. S1 Data input is performed.
S2 The input data is decomposed into decomposed elements and applied to the format of the knowledge data holding unit. S3 It is determined whether the case recognition data and the case data (decomposition elements, attributes, etc.) are insufficient. If there is no shortage, the process ends. If there is, go to S4. The screen displays the lack of knowledge data in S4 format,
Confirm the addition of knowledge data. Knowledge data (case recognition data, case data) is added to the S5 format. S6 Generate a new format.

Although the above has described case recognition data,
The present invention is also applicable to case data (see the description of the embodiment in FIG. 23). As described above, according to the present invention, the meaning of the input data is given based on the stored knowledge data, and if the format of the knowledge data is insufficient, and the meaning cannot be provided, a syntax that compensates the lack is output. The information is displayed on the means and the operator judges whether it is correct or not. In this way, the content of the knowledge data can be enhanced and a database of meaningful knowledge data can be constructed in an interactive manner. In the above, in the case of decomposition into decomposed elements or case recognition, the result was displayed on the display and the format for interacting with the operator was explained. However, the output unit is equipped with a voice output device, and the voice is used to judge whether the operator is correct or not. It is also possible to ask a question.

[0077]

Embodiment 3 FIG. 21 shows Embodiment 3 of the present invention, in which knowledge data is enriched by comparing knowledge data with input data based on an attribute name (case, etc.) and its attribute value. Assume that syntax A (41) is input. In the knowledge data creating unit 13, the syntax A is decomposed into decomposed elements. The decomposition element decomposition unit 14 decomposes input data into decomposition elements. At this time, the format determination unit 19 compares the format created by the plurality of decompositions with its own format.
The editing unit 51 edits the decomposed element decomposed by the decomposed element decomposing unit 14 and the questions to the operator based on the comparison result of the format judging unit 19, and displays them on the output unit. Alternatively, the content of the question is edited into audio data and output from the audio output device. Based on the result, the operator inputs data from the data input means 11 using a keyboard, a mouse, or the like so as to be correctly disassembled. Alternatively, input is made by voice, and the input data is correctly decomposed so as to make sense.

The decomposed element thus decomposed is recognized as a case and given meaning. For example, "I", "ha",
In the case of data decomposed into "school", "he", "go", and ".", Nominative case is given to "me", action case is given to "go", and purpose case is given to "school". And knowledge data 4
2 is created. At this time, the knowledge data reference unit searches the knowledge database holding unit (not shown), and finds the attribute value (I, school, going) and the attribute name (Nominative, Objective,
Behavioral data), and obtain knowledge data having a common pattern. In the case of FIG. 4, there is no data representing the exact same data configuration (I go to school), but there is the knowledge data 43 of "I" of the nominative, "go" of the behavioral, and "bicycle of the means". It is assumed that it is closest to the format of the input data. Again, the input requesting unit 18 obtains the
Is compared with the attribute name and attribute value. As a result, knowledge data 4
2 does not have the purpose “at school”, the editing unit 51 displays “where?” And “at school” on the output means 31 and asks the operator a question. Alternatively, a question is output to a voice output device as voice data, and the operator is prompted to input correct data. In this case, the question is correct,
That case recognition is correct. Then, since the knowledge data 42 has the measure "bicycle", the data in which the objective case "school has a nomination of" school "is added to the knowledge data 42. The editing unit 51 displays the question “what?” And “by bicycle?” On the output unit 31.

Therefore, if the operator determines that the question is correct, the knowledge data 43 indicates that the purpose is “school”.
Is replaced with the knowledge data 43 as new data. At this time, if incorrect, the knowledge data 43 is left as it is,
Knowledge data obtained by adding the target case “school” to 3 is added as new data (the configuration of this knowledge data is not shown). The data input by the operator responding to the questioned content may be a voice input using a microphone or the like in addition to the data input using the keyboard or the like.

[0080]

Embodiment 4 FIG. 22 shows Embodiment 4 of the present invention, in which the format of input data and the format of knowledge data are compared based on case recognition data (attribute names of cases such as nominative case and objective case). In a state where the existing knowledge data 53 'and the knowledge data 54' exist in the knowledge data holding unit, the syntax A (4
Assume that 1) is input. The knowledge data creation unit uses syntax A
(41) is decomposed into decomposed elements to recognize the nominative case “I”, the purpose case “school”, and the action case “go”. Next, the format of the case recognition data of the knowledge data is referred to by referring to the knowledge data holding unit. Knowledge data 53 'and syntax A (4
1) matches two kinds of cases, and the knowledge data 54 'and the syntax A (41) only match one kind of case.
The knowledge data 53 'is selected as knowledge data of a similar format. Therefore, the knowledge data 53 'and the syntax A
By comparing (41), there is no behavioral case “go” of the syntax A in the knowledge data 53 ′, so the data B (“B”) added with the nominative case “I”, the purpose case “school”, and the behavioral case “go” that match both. (Not shown). Then, based on the measure “walk” in the knowledge data 53 ′, the editing unit 51 determines “what?”
"On foot?" Is displayed on the output means 31 to ask the operator a question. If the operator inputs that question is correct, the new knowledge data 55 is registered in place of the knowledge data 53 '. When there are a plurality of pieces of knowledge data having similar formats, the number of times of use or the higher probability of selection is selected.

If "by bicycle" is answered in response to "what?" And "on foot?"
The new knowledge data in which the attribute value of the measure is "bicycle" is created. At this time, if only "no" is answered, the knowledge data 53 'is stored as it is,
Data B in which behavioral status “go” is added to knowledge data 53 ′
Is registered as new knowledge data.

FIG. 23 shows a third embodiment of the present invention, in which a format is applied to case data.
FIG. 23 shows a case where, when there is no appropriate format in the knowledge data holding unit, a new format that compensates for the lack of the knowledge data holding unit is generated, the knowledge data is enriched, and the meaning of the input data is given. FIG. 23 shows knowledge data 61 and knowledge data 62 as illustrated as case data.
Assume that exists. In such a case, case data A
It is assumed that (60) has been input. The knowledge data creating unit 13 assigns the attribute “name” to the name “XXX”, assigns the attribute “age” to the age attribute value “40 years old”, and assigns the attribute “age” to the attribute value “Enma”. The attribute “character” is assigned to the attribute (this assignment is performed by searching the knowledge data holding unit and, for example, based on the knowledge data 61).

Then, the input requesting unit 18 again compares the knowledge data 61 and the knowledge data 62 with the case data A (60), and if the knowledge data 61 has the common attributes “name” and “age”, Because there are many cases, they are extracted as similar cases. Then, since the attribute “character” is “aggressive”, the editing unit displays “Is character aggressive?” On the display to urge the operator to correct the data. Then, if the attribute value of the character of the case data A (60) is to be changed to "aggressive" to "aggressive", "high" is inputted by the data input means, and if not, "no" is inputted. . At the time of the high, the knowledge data 6 in which the character “Enjoy” of the case data A (60) is changed to the character “Aggressive”
1 is saved. In the case of "No", the character "Enmitsu"
New data 63 is created. If another attribute value “honest” or the like is input as an answer to the question, the data rewritten with the character “honest” or the like is held as new data (not shown).

[0084]

Embodiment 5 FIG. 24 shows an embodiment of the system configuration of the present invention. In FIG. 24, reference numeral 11 denotes data input means. Reference numeral 31 denotes an output means for outputting a moving image such as an audio output device including a display, a speaker, and an audio amplifier, and a video device. 50 is a computer. Reference numeral 52 denotes an input / output interface, which includes data input means 11, data output means (display, printer, audio output device, video device, etc.), and external storage device 71.
Interface.

Reference numeral 53 denotes a data processing section for processing input data (including the data processing means 10 and the data holding means 20 in FIG. 18). 61 is a CPU. 62 is a RAM. 63 is a ROM. 64 is an internal magnetic disk. (The above is the data processing unit 53) In the data input unit 11, reference numeral 81 denotes a keyboard and a mouse. 83 is a voice input device. Numeral 84 denotes a graphic data input device for inputting graphic data from an image reader or the like.

Reference numeral 85 denotes an electronic camera. 86 is CD-R
OM drive. Reference numeral 87 denotes a video camera for inputting image data such as a moving image. A microphone for inputting voice. Reference numeral 91 denotes a storage medium, such as a magnetic disk or a CDROM. An installation program 92 is a program for installing the program recorded on the storage medium 91 on the internal magnetic disk 6453. Reference numeral 121 denotes a knowledge data creation program, which is a program of a knowledge data creation unit. A decomposition element decomposition program 122 is a program for a decomposition element decomposition unit. Reference numeral 123 denotes a case recognition program, which is a program of a case recognition unit.
Reference numeral 124 denotes a knowledge data reference program, which is a program of a knowledge data reference unit.

Reference numeral 125 denotes an input request program again.
Reference numeral 131 denotes an editing program, which is a program of the editing unit. 132 is a registration program, which is a program of the registration unit. FIG. 247 shows an example of the device configuration of the present invention, and does not require, for example, all of the data input units.

FIG. 25 shows an example of a method for compensating for incomplete data of a decomposition element according to the present invention. According to the present invention, the knowledge data is created as described above. If the data of the decomposed element is incomplete, the data is completed as shown in FIG. Assume that syntax A has been entered. The case recognition data creating unit decomposes the case as shown in FIG.
At this time, for the input data “comporter” of the decomposition element,
With reference to the similar dictionary 72 of the knowledge data holding unit, the input request unit 18 and the editing unit 51 again display words “computer”, “concrete”, “comment”, etc. similar to “component” on the display unit. Then, the operator selects a computer and corrects the input data “comporter” of the decomposition element to “computer”.

[0090]

Embodiment 6 FIG. 26 is a flowchart of an embodiment of the present invention. S1 The case recognition data of the input data is obtained. S2 The input data (knowledge data) is applied to the case recognition data. S3 Determine whether the primitive data is missing.
If found (NO), proceed to S11; if not found (YES), proceed to S4. S4 Find similar data and make an inquiry. S5 It is determined whether there is correct data in the similar data. If there is (YES), proceed to S6, otherwise proceed to S7. S6 Recognize as similar data. S7: Request for input again. S8 It is determined whether there is correct data in the similar data. If there is (YES), proceed to S10, otherwise proceed to S9.

S11 It is determined whether primitive data is insufficient. If there is not enough (YES), the process proceeds to S12, and if not, the process proceeds to S13 (NO). S1
2 Have the user enter primitive data. S13 It is determined whether there is too much. If too much (YES) S
Proceed to 10, and if not too much (NO), end the process.

[0092]

Embodiment 7 FIG. 27 shows a basic configuration of the present invention. In the present invention, the meaning is to analyze the input data by decomposing it into decomposed elements to perform syntactic analysis such as case recognition, and to classify into syntactic patterns (to understand sentence understanding such as positive sentences, negative sentences, and question sentences). Or when talking with personal habits or bosses, or classifying things that change with the situation, such as spoken or written language),
To give meaning to the input data, such as giving specific meaning to the content of the data (for example, giving the proper noun "Yamada" an attribute of hobby baseball, accurate warmth, etc.). .

In FIG. 27, reference numeral 1 denotes data input means, which is provided with text data input means, voice data input means, graphic data input means, etc., and includes natural data such as text data, graphic data, voice data, and multimedia data. Is entered. Reference numeral 2 denotes text data input means for inputting text data such as a character code, such as a keyboard. Reference numeral 3 denotes voice data input means, such as a microphone. Reference numeral 4 denotes a graphic data input means for inputting still image data or moving image data such as characters and figures such as an image scanner, a digital camera and a video. Alternatively, it is a means for inputting vector data for drawing a figure such as a pointing device.

Reference numeral 10 denotes data input processing means for determining the type of data input by the data input means 1;
Data representing the type (text data, voice data, image data, etc.) is added to the input data and input to the knowledge data processing means. Reference numeral 11 denotes a data type determination unit that determines the type of input data. 14
Denotes data output processing means for determining the type of data to be output and selecting an output device. Reference numeral 15 denotes an output means determination unit. Reference numeral 20 denotes a knowledge data processing means for giving meaning to input data while interacting with a user. 22 is a semantic unit which parses the input data with reference to the knowledge data holding means 30,
It gives meaning to the input data.

Reference numeral 24 denotes a knowledge domain determination unit which creates an index to related knowledge data, and determines a knowledge domain such as what the technical field of input data belongs to or general knowledge. , Which knowledge data and how they are related to each other, and create an index. 27 is a dialogue processing unit,
In the process of giving meaning to the input data, based on the determination result of the knowledge processing unit, the input of the input data is urged, the result of the syntax analysis or the result of the meaning, or the voice output of the search result of the knowledge data holding means 30 is output. Or perform interactive processing with the user by display or the like.

Reference numeral 28 denotes a re-input requesting unit for requesting a supplementary input of data in response to an instruction from the knowledge processing unit 21 or asking a user a question. Reference numeral 29 denotes a registration unit for registering the knowledge data given by the knowledge processing unit 21 in its own knowledge data holding means 30. Numeral 30 denotes own knowledge data holding means, which is the own (current user) knowledge data holding means. 3
Reference numeral 1 denotes a knowledge domain index holding unit, which indicates a personal information index indicating personal information in the knowledge data area, a specialty domain index indicating knowledge data in a technical field in the knowledge data area, and general knowledge in the knowledge data area. This holds the general knowledge index.

Reference numeral 32 denotes a personal information index, which represents an index to personal information such as personal preference.
33 is an index of a specialty field. 34 is an index of general knowledge. Numeral 35 is a knowledge data area, which is an area for retaining meaningful knowledge data. 55 is a data output means. Reference numeral 56 denotes an image output means such as a display. Reference numeral 57 denotes a print output unit, such as a printer. Reference numeral 58 denotes an audio output unit such as a speaker.

The operation of the basic configuration of the present invention shown in FIG. 27 will be described. Data is input from the data input unit 1. In the input data processing means 10, the data type determination unit 11 determines the media type of the input data, adds data representing the media type to the input data, and sends the data to the knowledge data processing means 20. In the knowledge data processing means 20, the knowledge data processing means 20 parses the input data and refers to the knowledge database of the knowledge data holding means 30 to give meaning to the input data. In the syntax analysis, for example, the input data is decomposed into decomposers which are the minimum units of a meaningful chunk of data, and the knowledge data reference unit 26 refers to the knowledge data holding unit 30. Then, a format that matches the syntax pattern of the input data is determined based on the format of the knowledge data that has already been given and held in the knowledge data holding means 30, and is given a meaning.

For example, when data “I go to school.” Is input, the data is decomposed into decomposers. In this case, "I", "Ha", "School", "Nii", "Go", "."
Decompose into Then, it is compared with the pattern of the syntax format of "I" and "ha". The syntactical format "I am" is already meaning that "I" in "I" is the subject and "ha" is a particle representing the nominative case, so the data input "I" is the subject, "I" Is determined to be a particle representing the nominative case. Further, referring to the knowledge data holding means 30, based on the syntactic pattern or the definition information of the words which have been already registered, the purpose case and the
"Ni" is given a meaning such as a particle case representing the purpose, "go" is given an "action case", and "." Is given a final case.

The result with the meaning is converted into dialog data by the dialog processing unit 27 and output to the data output unit 55. For example, the user's judgment is requested by audio output and / or display on a display such as "Is the meaning of the data just input what is it now?" Also, for example, when the data “I am going at 8 o'clock” is input, the knowledge data processing unit 20 determines that there is not enough purpose to indicate where to go as a result of parsing, and performs interactive processing. Through the unit 27, the user is asked to ask "Where are you going?" Or the like by voice display and / or display on the display to urge the user to supplement the input data.

Further, in the process of giving meaning through such dialogue processing, if there is a habit of using the word of the person, for example, a habit such as “Ehto” or “And Ne”, the habit of the word is used as the habit of the word. It is registered in the knowledge data holding unit together with the person's name. Then, if there is a data entry such as "Look for a person who speaks," you can search for that person at high speed using that habit as an index. Or, when the person inputs data by voice saying "Um, I go to school." And converts the input data into written words, etc., "Um," is a habit of the person's speech. "I go to school."

Alternatively, in the process of assigning meanings, the knowledge domain determination unit 24 determines that the contents of the input data are a specialized area such as medicine or law, and obtains a knowledge domain. Then, the search process can be performed at high speed by referring to the knowledge data using the index of the knowledge domain. Further, in the knowledge data processing means 20, the knowledge processing unit 21 determines how the registered knowledge data is used (for example, obtains statistical data on the usage) and is indexed as expert knowledge. However, if it is used as general knowledge, an index is created for general knowledge, and this is used preferentially. Or, on the contrary, what was indexed to specialized knowledge is re-indexed to general knowledge.

Further, according to the present invention, a word type is determined and converted to another word type. For example, when data is input in a spoken language but not registered in the knowledge data in a spoken language, in such a case, it is determined that the input data is a spoken language, and the input data is converted to a written language. The word types are written words, business words, fellow words, infant words, and the like. Further, in the present invention, the reference to the knowledge data holding means refers not only to the knowledge data of the individual user but also to the knowledge database of another person. At that time, for example, there is a case where the knowledge data of the other party is created based on a dialect. In such a case, the input data is translated into the dialect so that the data can be searched at high speed.

Further, if it is not possible to make a meaning, search, etc. in the knowledge database of another person, a knowledgeable person who can understand the knowledge of the problem can be obtained and the knowledgeable person information can be provided to the user. Alternatively, the intellectual can be automatically called, or the contents of the question can be automatically transmitted over the Internet. As described above, according to the present invention, the system asks the user about the input data, solves it interactively with the system on the spot, and furthermore, if the knowledge is still insufficient, the knowledge of another person is obtained. By borrowing data and making inquiries to intellectuals, the knowledge can be filled immediately. Therefore, according to the present invention, it is possible for the user to easily give an appropriate meaning according to the personality of the user. Therefore, it is possible to easily create an optimal knowledge database for each user.

At this time, knowledge domains such as personal information, specialized fields, and general knowledge are obtained so that the target knowledge can be efficiently found from a large amount of knowledge data.
Since search areas can be narrowed down, knowledge data can be created efficiently. In addition, since accurate parsing can be performed by interactive processing by accurately determining the lack of knowledge, the user can create the knowledge database at a high speed with a simple operation.

[0106]

Eighth Embodiment FIG. 28 shows a configuration of an eighth embodiment of the present invention. In FIG. 28, reference numeral 1 denotes data input means.
Details of the data input means are shown in FIG. Reference numeral 5 denotes an input / output interface. In input / output interface 5, 1
0 is a data input processing means. Reference numeral 11 denotes a data type determination unit. 14 is a data output processing means. Reference numeral 15 denotes an output means determination unit. Reference numeral 16 denotes a network interface. 20 is a knowledge data processing means. 21 is a knowledge processing unit. 26 is a knowledge data reference unit.

In the knowledge data reference section 26, reference numeral 41 denotes a process of referring to own knowledge data, which is a means for referring to the user's own knowledge data holding means 30 '. 42
Is a process of referring to another person's knowledge data. The other person's knowledge data holding means A (40), the other person's knowledge data holding means B
This is means for referring to (41). Reference numeral 43 denotes processing of inquiries of intellectuals, for example, making a list of intellectuals, making a telephone call to intellectuals, making a question on the Internet or the like.

Reference numeral 27 denotes a dialog processing unit. 28 is a re-input request unit. 29 is a registration unit. Reference numeral 30 denotes a knowledge data holding unit. 30 'is own knowledge data holding means. 32 'is an index of a knowledge domain of personal information such as personal preference and habit. 33 'is an index of the knowledge domain of the specialized field. 34 'is an index of a knowledge domain of general common sense. Numeral 35 'is a knowledge data file which is used for holding its own knowledge data. Or, on the contrary, what was indexed to specialized knowledge is re-indexed to general knowledge.

Further, according to the present invention, a word type is determined and converted to another word type. For example, when data is input in a spoken language but not registered in the knowledge data in a spoken language, in such a case, it is determined that the input data is a spoken language, and the input data is converted to a written language. The word types are written words, business words, fellow words, infant words, and the like. Further, in the present invention, the reference to the knowledge data holding means refers not only to the knowledge data of the individual user but also to the knowledge database of another person. At that time, for example, there is a case where the knowledge data of the other party is created based on a dialect. In such a case, the input data is translated into the dialect so that the data can be searched at high speed.

Further, if it is not possible to make a meaning or search in another person's knowledge database, a knowledgeable person who can understand the knowledge of the problem is sought, and the knowledgeable person information 53 is provided to the user. The network to access. 55 is a data output means (details are shown in FIG. 4). Reference numeral 61 denotes an application program recognizing unit, which is a means for cooperating with another system, for recognizing another application program and establishing a connection. 6
2, 63, 64 and 65 are other application programs. The operation of the configuration in FIG. 28 will be described later.

FIG. 29 shows the configuration of the embodiment of the configuration of the knowledge processing section of the present invention. In FIG. 29, reference numeral 21 denotes a knowledge processing unit. Reference numeral 22 denotes a meaning assigning unit. In the meaning assignment unit 22, reference numeral 71 denotes a decomposition element decomposition unit. Reference numeral 72 denotes a syntax analysis unit. Reference numeral 73 denotes a format determination unit which compares a pattern of arrangement of decomposed elements of the input data with a syntax format stored in the knowledge data storage unit.

Reference numeral 74 denotes a personal information determination unit for determining personal information from input data. Reference numeral 75 denotes a language classification unit for determining the type of the user's language (spoken language, written language, business language, and the like). 7
Reference numeral 5 'denotes a format determination unit for comparing the input word type with the word type pattern held in the knowledge data holding unit. 75 ”is a word type conversion unit that converts spoken words into written words,
And the like.

Reference numeral 76 compares a pattern peculiar to oneself with a pattern peculiar to another person, and converts the pattern if necessary. For example, when there is Kansai accent in the user and the knowledge data of another person to be referred to is a standard language, it is converted into a standard language. Reference numeral 76 'denotes a format determination unit which compares a pattern unique to the user with a pattern unique to another person. Reference numeral 76 "denotes a pattern conversion unit which converts a dialect into a standard language or the like.
Reference numeral 78 denotes a specialized field determining unit for determining the specialized field of the input data. Reference numeral 79 denotes a general knowledge determination unit that determines general knowledge of input data.

Personal information judging section 74, specialized field judging section 7
8. The general knowledge determination unit 79 is included in the knowledge domain determination unit 24 (see FIG. 27). Reference numeral 80 denotes a statistics unit for calculating statistics on how words are used. Based on the statistics of the statistics section 80, the general knowledge determination section 79 changes, for example, the knowledge data indexed to the specialized field to general knowledge or vice versa. It performs processing such as performing

[0115]

Embodiment 9 FIG. 30 shows an embodiment of the system configuration of the present invention. In FIG. 30, reference numeral 1 denotes data input means. In the data input means 1, reference numeral 111 denotes a keyboard and a mouse. 112 is a pen input device. 1
Reference numeral 13 denotes a voice input device. 114 is an OCR,
This is for reading characters or the like of a document optically. Reference numeral 115 denotes an electronic camera which captures a document such as a character or a figure and records the document on a storage medium.

Reference numeral 116 denotes a video camera for photographing a moving subject and recording moving image data on a storage medium. Reference numeral 117 denotes a graphic data input device, which is an image scanner or the like for reading characters, figures, etc. of a document to obtain image data. 55 is a data output means. In the data output means 55, reference numeral 121 denotes a display. Reference numeral 122 denotes a printer. 123 is a fax.

Reference numeral 125 denotes a video device. Reference numeral 5 denotes an input / output interface. In the input / output interface 5, reference numeral 16 denotes a network interface.
A peripheral device interface 136 includes a data input processing unit, a data output processing unit, and the like.
Reference numeral 130 denotes an external storage device, which is a storage device that holds knowledge data. 140 and a storage medium, such as a floppy disk that stores a program for executing the present invention.

Reference numeral 142 denotes an installation program for installing various programs for executing the system of the present invention. An input / output processing program 143 is a program for controlling data input means and data output means. Reference numeral 145 denotes a knowledge data creation program which performs knowledge data processing. A decomposition element decomposition program 146 is a program for decomposing input data into decomposition elements. Reference numeral 147 denotes a syntax analysis program which performs syntax analysis and the like to recognize the case of a disassembly element.

Reference numeral 148 denotes a personal information determination program, which is a program for determining personal information. The program includes a program for determining individual word types and a program for mutually converting between different individual word types. 14
Reference numeral 9 denotes a specialized field determination program, which is a program for determining a knowledge domain. Reference numeral 150 denotes a general knowledge determination program, which is a program for determining whether general knowledge or specialized knowledge exists. 151 is a knowledge data reference program for searching for knowledge data holding means.

A registration program 152 is a program for registering meaningful knowledge data in the knowledge data holding means. Reference numeral 153 denotes a translation program for the pattern of the self and that of another person, which determines the difference between the format of the self and that of the other person and performs mutual conversion. 160 is a data processing unit. In the data processing unit 160, 1
61 is a CPU. 162 is a RAM. 16
3 is a ROM. 164 is an internal magnetic disk.

Reference numeral 165 denotes a knowledge data creation program, which is a knowledge data creation program held on the internal magnetic disk 164. 170 is an individual or common or other person's knowledge data holding means. FIG. 31 is a diagram for explaining the operation of the embodiment of FIG. Data is input by the data input unit 1. The data type of the input data is determined, the data type is added, and the data is input to the knowledge data processing means 20. In the knowledge processing section 21, the knowledge domain determination section 24 determines the knowledge domain of the input data.

The input data is decomposed into decomposed elements by a decomposed element decomposing unit 71, and the parsing unit 72 parses the input data with reference to its own knowledge data holding unit 30 '. here,
The knowledge data holding means 30 'or another person's data holding means 40' is intercepted and parsed. For example, when "I go to school by bicycle" is entered,
"I", "Ha", "School", "Ni", "Bicycle",
Decompose into "in", "go", ".", Etc. Then, the syntactical format that has already been registered is deepened, and the subject is the subject for "I", the noun particle for "ha", the objective case for "school", The particle case, the device case for "bicycle", the particle case for "de", the action case for "go", and the end case for "." I do.

Then, it is checked whether or not there is a missing case in the input data by the syntax analysis (processing of S1). For example, if the input data has a missing part such as "I go by bicycle" or the like, "where?" Transfer to 55. Ask the user "Where?" Or display 2
22, "Where?" Is displayed to prompt the user to input data thinly.

When the user inputs “school” by voice input or keyboard input, the decomposition element decomposition section 71 decomposes the input data to which the supplementary input data is added into decomposition elements, and The analysis unit 72 performs syntax analysis as described above to give meaning. When it is determined in the determination process of S1 that there is no missing case or that an instruction to ignore the missing case has been input by the user, the syntax analysis result is displayed on the display 222, and the user is confirmed.

In the determination process of S2, when the user inputs that the user is satisfied with the syntax analysis result obtained by analyzing the apparatus, the registration unit 29 stores the meaningful sentence data in its own knowledge data holding unit. 30 'and perform the following format determination processing (S3). In the processing of S3, the own knowledge data holding means 30 'is searched, the format of the input data is compared with the format held in the knowledge data holding means 30' while performing interactive processing, and the format determination processing is performed. For example, processing such as comparing the word type of the input data with the registered word type and performing conversion processing is performed. Then, the result of the processing is displayed on the display 222 via the interactive processing unit 27.

If the user is satisfied in the judgment processing of S4,
The registration unit 29 registers the format determination result in its own knowledge data holding unit 30. In FIG. 31, the processing for obtaining a syntax by referring to the decomposition element decomposition unit 71, the syntax analysis unit 72, and the own knowledge data holding unit 30 'is repeatedly performed a plurality of times while interacting with the user. In addition, the format determination is performed on the result of the syntax analysis with reference to the own knowledge data holding means 30 '.
The process 3 is also performed a plurality of times while interacting with the user. Further, the determination of whether the knowledge domain is the best, the change of the knowledge domain, and the like are also included in the processing of S3. Further, the change of the index from the specialized knowledge to the general knowledge or the change from the general knowledge to the specialized knowledge is also included in the processing of S3.

If it is determined in step S4 that the user is not satisfied, or if the user's own pattern is to be translated into another person's pattern, another person's knowledge data holding means 40 'is searched in step S5. The pattern of the input data is compared with the format held in the knowledge data holding means 40 'of another person. Then, the comparison result is displayed on the display 222 via the interactive processing unit 27. Here also, the process of S5 is performed a plurality of times while interacting with the user. The determination and change of the knowledge domain are also included in the processing of S5. Further, the change of the index from the specialized word to the general knowledge, the change from the general knowledge to the specialized knowledge, and the like are also included in the processing of S3.

If it is determined in step S6 that the user is satisfied, the registration unit 29 registers the result of the meaning in the own knowledge data holding unit 30 '. If it is determined in step S6 that the user is not satisfied, a list of intellectuals is created and displayed on the display in steps S7 and 93. Then, the user contacts the intellectual by the telephone number or the like of the intellectual displayed on the display. Or automatically call an intellectual from the device,
Alternatively, a question is asked to the intellectual by computer communication such as the Internet via a network.

Therefore, if an answer can be obtained from a knowledgeable person,
Based on the answer, meaning is given (syntax analysis and format determination), and displayed on the display 222. If the user is satisfied, the result given by the registration unit 29 is registered in the own knowledge data holding means 30 '. If the user is not satisfied, the user presented to the user creates a meaning based on his / her own judgment or knowledge obtained from the intellectual, and registers it in his / her own knowledge data holding means 30 '.

FIG. 32 is a flowchart of the input data processing section of the present invention. The operation will be described according to the order of the illustrated steps. S1 Input data. S2 Determine whether there is a media attribute. If there is a media attribute, the process ends. If not, check the input device (input device). First, it is determined whether the device is a text device (keyboard or the like). If the device is a text-based device, a text-based media ID is assigned, and the process ends. If it is not a text device, it is determined whether or not the device is a voice device (voice input device).

If the device is an audio device, an audio media ID is assigned, and the process is terminated. If it is not an audio device, it is determined whether or not it is an image device. If the device is an image-based device (such as a scanner), an image-based media ID is assigned, and the process ends. If it is not an image device, it is determined whether it is a vector device (mouse, pointing device, etc.).

If the device is a vector device, a vector media ID is assigned, and the process is terminated. If it is not a vector device, it is determined that the device is a moving image device. Video-based media ID for video-based devices (such as video cameras)
Is added, and the process ends. If it is not a moving image device, the following processing (S3) is performed.

S3 Since the device is not any registered device, an inquiry is made to the user. S4, S4 ', S5
Determine whether the user has entered a media type. If there is a media input, the media ID specified in S5 is assigned. If there is no instruction of the media type, the process ends.

FIG. 33 is a flowchart of the output data processing section. S1 Receive output data. S2 Determine whether the data is text data. If the data is text data, a text device (display or the like) is selected, and the process is terminated. If the data is not text data, it is determined whether the data is audio data.

If the data is audio data, an audio device (audio output device or the like) is selected, and the process is terminated. If it is not audio data, it is determined whether or not it is image data. If it is image data, an image device (display or the like) is selected, and the process is terminated. If it is not image data, it is determined whether or not it is vector data.

If it is vector data, a vector device (display or the like) is selected, and the process is terminated. If it is not vector data, it is determined that the data is moving image data.
For moving image data, moving image devices (video equipment, etc.)
Is selected to end the processing. If it is not moving image data, the following processing (S3) is performed. S3 Inquires the user because it is not any registered device. S4, S4 ', S5 It is determined whether the user has designated the media type. If there is an input of the media type, the device designated in S5 is selected. If there is no instruction of the media type, the process ends.

FIG. 34 is a flow chart for classifying word types according to the present invention. S1 A word to be determined is captured. S2 It is determined whether it is a spoken language. For spoken language, refer to spoken language format data. If it is not a spoken language, determine if it is a written language.

In the case of a written language, format data of a written language is referred to. If it is not a written word, judge it as a business word. If it is a business language, the format data of the business language is referred to. If it is not a business language, judge it as a fellow language. If the format data is for fellow words, the format data of fellow words is referred to. If it is not a fellow language, the infant language is judged. If it is an infant language, refer to the format data of the infant language system. If it is not an infant language, the following processing (S3) is performed.

S3 Inquire the user for words for the new word group! S4, S4 ', S5 If the user indicates a word, the user will try to use the word type specified in S5. If there is no instruction from the user, the process is terminated.

FIG. 36 is a flowchart of the mutual conversion of word types. S1 Import the word to be converted S2 Determine whether it is a spoken word. If it is a spoken language, it is recognized as a spoken language, and the process ends. If it is not a spoken word, determine whether it is a written word. If it is a business word, it is recognized as a business word, and the process of S6 is performed. If it is not a business language, judge it as a companion language.

If it is a fellow language, it is recognized as a fellow language, and S6
Is performed. If it is not a fellow language, it is judged as an infant language. If it is an infant language, it is recognized as an infant language, and the process of S6 is performed. If it is not an infant language, the process of the next process (S3) is performed.

S3 The user is inquired of the word type for the new word type. S4, S4 ', S5 If the user indicates the word type, the specified word type in S5 is adopted. If there is no instruction of the word type from the user, the process of S6 is performed. S6 Select a corresponding format among the word types to be converted.

FIG. 36 is a flowchart of the translation between one's own expression pattern and another's expression pattern according to the present invention. S1 Incorporate words to be translated. S2 Search for past translation examples. S3: It is determined whether there is the corresponding knowledge data. If there is, the process of S6 is performed.

S4, S4 'Inquire the user about the translation method. S5, S5 'The user judges the input of the instruction of the translation method. If an instruction is input, the process of S6 is performed,
If not, end.

S6 References the relevant (instructed translation example). For example, the flowchart of FIG. 10 will be specifically described with reference to a case where “I like apples” is translated. In S1, "I like apples" is imported. In S2, Mr. B searches for how to express “I like apples”. In S3,
He has the relevant knowledge, and Mr. B is American and says “I l
ike an apple. If it is linked to the knowledge ", use that. In S3, if there is no relevant knowledge, "I" is "I" in Mr. B's words,
"I like" instructs "like" in the word of Mr. B, "apple" instructs "apple" in the word of Mr. B, and the arrangement method is such as "I like an apple."

FIG. 37 is a flowchart of automatic setting of a knowledge domain according to the present invention. S1, S1 'It is determined whether the work is a new work or a work in progress. If it is a new work, the process of S2 is performed, and if it is a work in progress, the process of S3 is performed.
S2: The most relevant knowledge domain is selected from the directory name and file name immediately before. S3 Select a knowledge domain in process. S4, S4 'Check if there is an explicit domain instruction by the user. If there is a domain instruction, the process of S5 is performed, and if not, the process of S9 is performed.

S5, S5 'The given knowledge domain is compared with the knowledge domain determined by the device. If they match, S
6 is performed. If they do not match, the process of S7 is performed.
S6: adopt the knowledge domain. S7 Inquire the user. S7 ', S7 "Input the knowledge domain. S8 Dynamically perform the statistical processing used by the user, and monitor the most appropriate knowledge domain as the knowledge domain to which this word belongs. If it is, present it to the user and wait for the user's instruction.

S9 The statistical processing of the words used by the user is dynamically performed, and the most appropriate knowledge domain as the knowledge domain to which the words belong is monitored. If another optimal knowledge domain is found, it is presented to the user and waits for the user's instruction.

FIG. 38 is a flowchart of a process of referring to knowledge data and a method of acquiring knowledge of a knowledgeable person. S1 Grasp a knowledge domain closely related to the required knowledge. S2 Refers to its own knowledge data and displays the most relevant knowledge in this knowledge domain. S3 Waits for the user to input whether the displayed knowledge is satisfactory or not, and if satisfied, performs the processing of S3.
If not satisfied, the process of S5 is performed. S4, S5 If there is an input that the user is not satisfied, the knowledge data of another person is referred to, and the knowledge most closely related to this knowledge domain is displayed.

S4 ', S3 If there is an input of satisfaction, the knowledge is adopted in S3. S5: Refer to the intellectual knowledge table, present the telephone and e-mail number of an appropriate expert, and confirm with the user whether to make an inquiry. S5 ', S6 The user does not make an inquiry, and the user investigates the related data offline. S5 ', S6 The device automatically checks and selects and displays from among a plurality of answers. S6 ', S7 If the user is not satisfied with the displayed selection result, the user investigates the related data offline.

S6 ', S3 If there is an input from the user that the user is satisfied with the displayed selection result, the knowledge is adopted in S3.

[0161]

According to the present invention, a user (operator) is referred to a database in which various data such as dictionaries and sentences which are already labeled and stored for input information are stored. Let them know the result,
Based on the result, the user can easily and interactively perform accurate labeling intended by the user through input using a keyboard or the like or voice input using a microphone or the like.
Therefore, it is possible to register data in which an accurate label is assigned to the input information (case recognition, attribute setting) in the database. Also, during the labeling process,
Rewrites the contents of already created dictionaries and various data, such as sentences, based on the information entered later, and constantly changes the contents in a meaningful way while learning according to changes in information. be able to. Therefore, according to the present invention, the content of the database to be stored is abundant.

Further, referring to various data such as dictionaries and sentences which are already labeled and registered with respect to the input information, and performing the labeling in an interactive manner, the format and the input information of the data which has already been registered and stored are referred to. Since the format of the created data is compared and applied to determine the lack of data labels, labeling can be performed efficiently. If labeling cannot be performed due to lack of necessary information in data created from input information,
Insufficient information is displayed or displayed on a display or a sound output device, and the user (operator) is prompted to input the missing information, thereby enriching the contents of the registered and stored data. In addition, the lack of data of the data obtained from the input information can be further compensated based on the data supplemented in such a manner. Therefore, according to the information change, the data created based on the contents of the accumulated data and the input information is prepared. The data is correctly labeled data.

Similarly, the user can easily perform the data creation, update, or search without requiring a specialized engineer. Also, when labeling cannot be performed due to lack of data in the data selected from the input information, or when it is difficult to understand the meaning of the term, specialized field, or the like in the search, the device asks the user a question, and the data of the input information is deleted. It encourages supplements and allows the user and the device to interact and label the data obtained from the input information. In addition, the knowledge database can be searched quickly and efficiently. Further, the present invention provides a method for inputting meaning data to knowledge data that has been registered with meaning.
While the user and the device interact with the latest contents, the contents can be learned and the knowledge data can be updated. Therefore, the content of the knowledge data becomes rich while being used, and the knowledge data that can respond to the latest situation can be always maintained. Therefore, if the present invention is applied to an information retrieval system or the like, necessary information can be easily retrieved in an interactive manner, and information to be stored can be automatically enriched by ordinary use.
And the database which can respond to the latest situation can always be maintained.

[Brief description of the drawings]

FIG. 1 is a diagram showing a basic flow of the present invention.

FIG. 2 is a diagram illustrating a basic configuration of a first embodiment of the present invention.

FIG. 3 is a diagram illustrating a system configuration according to a first embodiment of the present invention.

FIG. 4 is a diagram showing a second embodiment of the device configuration of the present invention.

FIG. 5 is a diagram illustrating a configuration of a case recognition unit and an editing unit according to the present invention.

FIG. 6 is a diagram showing an example of a data configuration of the present invention.

FIG. 7 is a diagram showing an example of a data configuration of the present invention.

FIG. 8 is a diagram showing an example in which the present invention is based on multimedia data.

FIG. 9 is a diagram showing a flowchart of case recognition according to the present invention.

FIG. 10 is a diagram showing a flowchart for assigning a symbol representing a main behavior of the present invention.

FIG. 11 is an explanatory diagram of a syntax wording symbol of the present invention.

FIG. 12 is a diagram showing a flowchart of a program for giving an attribute by a syntax wording symbol of the present invention.

FIG. 13 is an explanatory diagram showing the generation of new knowledge data objects by grouping the objects of knowledge data according to the present invention, and the layering of the objects.

FIG. 14 is a diagram showing a flowchart for collecting knowledge data objects and generating new knowledge data objects according to the present invention.

FIG. 15 is a diagram showing a flowchart of index creation of an object of knowledge data according to the present invention.

FIG. 16 is a diagram showing a flowchart of index creation of an object of knowledge data according to the present invention.

FIG. 17 is a diagram showing a flowchart of a search process of the present invention.

FIG. 18 is a diagram illustrating a basic configuration of a third embodiment of the present invention.

FIG. 19 is an explanatory diagram of the operation of the basic configuration according to the third embodiment of the present invention.

FIG. 20 is a flowchart of the input requesting unit according to the third embodiment of the present invention.

FIG. 21 is a diagram showing a fourth embodiment of the present invention.

FIG. 22 is a diagram showing a fifth embodiment of the present invention.

FIG. 23 is a diagram showing a sixth embodiment of the present invention.

FIG. 24 is a diagram showing a system configuration of the present invention.

FIG. 25 is a diagram showing an example of a method for setting data of a decomposition element according to the present invention.

FIG. 26 is a diagram showing a flowchart of the embodiment of the present invention.

FIG. 27 is a diagram illustrating a basic configuration of a seventh embodiment of the present invention.

FIG. 28 is a diagram showing a configuration of an example of the present invention.

FIG. 29 is a diagram showing an embodiment of a knowledge processing unit of the present invention.

FIG. 30 is a diagram showing a system configuration example of the present invention.

FIG. 31 is an explanatory diagram of the operation of the basic configuration of the present invention.

FIG. 32 is a diagram showing a flowchart of an input data processing unit.

FIG. 33 is a view illustrating a flowchart of the output data processing unit.

FIG. 34 is a diagram showing a word type classification flowchart.

FIG. 35 is a diagram showing a flowchart for mutually converting word types.

FIG. 36 is a diagram showing a translation flowchart between a user's own pattern and another person's pattern.

FIG. 37 is a diagram showing a flowchart for automatically setting a knowledge domain.

FIG. 38 is a flowchart of a method of referring to knowledge data and acquiring knowledge of a knowledgeable person.

Claims (28)

[Claims] 1. A database creation method for inputting information and saving the information as data in a database, a recording medium holding a program for creating a database for inputting information and saving the information as data in a database, and inputting information. In a recording medium in which a data group processed by processing the information is recorded A: Information input means B: Input information processing means C: Decomposition element decomposition means for decomposing data obtained by the information processing means into decomposition elements D: A labeling means for performing a labeling process of the decomposition element; E: a labeling dialogue means for performing a labeling dialogue process; F: a registration means for registering the labeled data in an information holding device; A storage medium storing a database creation program and A recording medium on which a data group characterized by being formed is recorded. 2. The information input means includes a discriminating means for discriminating whether the form of information is digital or analog, and when the information is analog, the information is converted into digital by the input information processing means. 2. A method for creating a database according to claim 1, a recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 3. The database creation method according to claim 1, wherein said information input means includes a discrimination means for discriminating a type of information (media type).
A recording medium storing a database creation program and a recording medium recording a data group created by the steps. 4. The information type is at least one information selected from characters, voices, sounds, images (still images, moving images), vectors and cases, and specifies an input device corresponding to each type of information. A method for creating a database according to claims 1 and 2, a recording medium storing a database creation program, and a recording recording a data group created by the steps. Medium. 5. The information processing means converts information which is not data which can be processed by the decomposing means for decomposing into decomposable data into data which can be processed and sends it to the decomposing means for decomposing means. The method for creating a database according to claim 1, wherein the data is sent as it is to the decomposing means, and the recording medium storing the database creating program and the data created by the steps are used. A recording medium having recorded thereon a data group. 6. The database creation method according to claim 1, wherein said decomposing element decomposing means decomposes into a minimum decomposable unit or a unit that can be labeled with a set of said minimum units. A recording medium storing a database creation program and a recording medium recording a data group created by the steps. 7. A method according to claim 1, wherein said decomposing means has reference means for referring to the decomposed elements registered in said database. A recording medium and a recording medium on which a data group created by the step is recorded. 8. A method according to claim 1, wherein said decomposing means has means for interacting with a user.
Item 8, a method for creating a database according to Item 6 and Item 7, a recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 9. The decomposing element dialogue means: A: The decomposing element is compared with the format to determine whether or not the decomposing element is appropriately decomposed. If it is not appropriate, the output means indicates to the user that it is not appropriate. B: Ask the user whether the decomposition is correct. C: If it is correct, use the decomposition as it is. D: If it is not a correct decomposition, output it. 9. A method for creating a database according to claim 8, further comprising: a decomposition element dialogue means comprising the steps of: (a) prompting a user to input a correct decomposition element to the user and making the inputted object a correct decomposition element. , A recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 10. The labeling means obtains a label for the obtained decomposed element by referring to an already registered database, and assigns a label. When a label to be assigned is not found, the fact is output to an output means. The method for creating a database according to claim 1, wherein the user is prompted to input a correct label and the input object is labeled, a recording medium storing a database creation program, and a program created by the steps. A recording medium having recorded thereon a data group. 11. The labeling dialogue means includes: A: a means for determining whether a label is appropriately labeled with respect to a label given to the obtained decomposed element, and output if the label is not appropriately labeled. Means is output in the means. B: Ask the user whether the labeling is correct. C: If it is correct, the label is used as is. D: If the labeling is not appropriate, the fact is output to the output means to prompt the user to input a correct label. Let the input thing be a correct new label E: If the label in the reference database and the label input by the user are different, let X be the newly input label and Y be the already registered label, The data having a configuration in which X is added to the common label is displayed on the output means, and the user is asked for correctness. The data to which the data X has been added is additionally registered to the data already registered in the database, and if the data to which the label Y has been added is not correct, the part already registered in the database is kept as it is. The labeling comprising the above steps 11. The database creation method according to claim 10, further comprising a dialogue means, a recording medium storing a database creation program, and a recording medium recording a data group created by said step. 12. The labeling means refers to a database registered for the decomposed element and, if there is no appropriate label, outputs the fact to the output means to the user and prompts the user to input a label. 12. The method for creating a database according to claim 1, wherein the object is a label, a recording medium storing a database creation program, and the method created by the steps. A recording medium on which a data group to be recorded is recorded. 13. The method according to claim 1, wherein the minimum decomposed element is a character when the information is a language.
2. A method for creating a database according to claim 1, a recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 14. A syntax composed of a plurality of decomposers is defined as 1
Two types of parsers, compare the types of attributes included in their own parsers with the types of attributes included in other parsers, and find other parsers that have all the types of their own parser. The method for creating a database and the database creation program according to claim 13, characterized in that an attribute which is not present in the parsing element which is not present is displayed on the output means, and the user is asked whether the label is assigned correctly or not. A recording medium storing the stored recording medium and a data group created by the steps. 15. Attributes include particles, auxiliary verbs, conjunctions, adverbs,
15. The method for creating a database according to claim 14, wherein the phrase is a phrase including a part of speech such as a punctuation mark and a symbol such as a space, a recording medium storing a database creation program, and a fact that the database is created by the steps. A recording medium on which a characteristic data group is recorded. 16. The wording symbols include who, when, where,
If the question contains a question word such as why and how, and if any of the decomposed parts of the input data is missing, the user displays the missing question word on the output means and inputs the attribute. A method for creating a database according to claim 15, a recording medium storing a database creation program, and a recording medium recording a data group created by said step. 17. A system according to claim 17, further comprising a similar word dictionary, wherein when the data of the decomposed element is incomplete, the incomplete data of the decomposed element is converted into correct data according to the result of the user's judgment. 14. A method for creating a database according to claim 13, a recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 18. An interactive means for editing a decomposition element or data necessary for labeling and displaying the result on an output means, and a user selecting or newly inputting the data based on the data indicated on the output means. Claims 1, 8, characterized in that the result is a formal decomposition or label
Item 13. A database creation method according to Item 9, Item 11, and Item 12, a recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 19. A method according to claim 1, wherein the data holding means for holding data in the database is an object database means, and the data is object data. , A recording medium storing a database creation program, and a recording medium recording a data group created by the above steps 20. When there are a plurality of common labels among objects, only a common part X is extracted to generate a new object and an ID is assigned to the object. 10. The database creation method according to claim 19, a recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 21. The database creation method according to claim 1, wherein said data registration means classifies the data according to the field and registers the data according to the field. And a recording medium recording a data group created by the steps. 22. The database creation method according to claim 1, wherein said registration means has a function of classifying the data according to a field and registering the data according to a field, and further comprising a data processing means registered in said database. , A recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 23. The data processing means of the database refers to the data holding means to change the contents of the registered data referred to in the labeling and labeling processing of the data by the interactive processing with the user, and in the labeling process. 23. Creation of a database according to claim 22, wherein a field of data is obtained, data of the determined field is referred to, labeling is performed on the data, and contents of registered data are updated at the same time. A method, a recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 24. The method according to claim 1, wherein the user first refers to his / her own database holding his / her own data, and refers to another person's data / base when labeling cannot be performed by his / her own database. 24. A method for creating a database according to items 23 to 23, a recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 25. A database according to claim 24, wherein if labeling is not possible even by referring to another person's database, a list of intellectuals having knowledge enabling labeling is obtained and displayed. , A recording medium storing a database creation program, and a recording medium recording a data group created by the steps. 26. The method according to claim 13, wherein said processing means recognizes the input word type and converts the input word type into a different word type. A recording medium storing a database creation program and a recording medium recording a data group created by the steps. 27. A difference between a pattern of own input information and a pattern of another person's database is identified, and the pattern of own input information is replaced with a pattern of another person's database.
The method for creating a database according to claims 21, 24 and 25, wherein the database created by said step is created by searching a database of another person. A recording medium on which a characteristic data group is recorded. 28. To determine in which field the input information is similar, calculate the similarity with each field, use the field with the closest similarity, 26. A database creation method according to claim 21, wherein the field is determined based on a directory name or a file name used immediately before the threshold value is not reached. A recording medium storing a creation program and a recording medium recording a data group created by the steps.