JP2015130194A - autonomous knowledge analyzer - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an autonomous knowledge analyzer to build logic and knowledge from the inputted information.SOLUTION: An autonomous knowledge analyzer converts the language expressing human thoughts into patterns by human instructions and learning, analyzes the patterns and the relationship between patterns and autonomously generates a connection relationship between patterns corresponding to the type and characteristics of statements. Since the relation of connection is generated based on logical relationships, definitions, similarity, attribution, the definition of relationship, inheritance of attribute information rather than recorded by itself, each is constructed as a knowledge system with relationships. In addition, even when addition or change has occurred in the processing of the information, it locally changes the patterns and the connection of the patterns without addition and change of the program.

Description

  The present invention relates to artificial intelligence and software for autonomously constructing logic and knowledge from input information sequentially.

  Conventional artificial intelligence performs actions according to pre-programmed procedures. Implementations other than the processing set by the program are difficult, and it is difficult to analyze the input information and build logic and knowledge by the machine itself.

Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. In addition, it was difficult for the autonomous machine to carry out the expansion of knowledge and the construction of knowledge, collate with the constructed knowledge system, and to make the machine carry out actions in accordance with the truth, truth, facts, rules and common sense.
In the present invention, processing for analyzing and recording information and the structure of information by human instruction and learning is performed, and a relationship between information is generated by connection between patterns and processing between patterns. To make it autonomous. In addition, the input information is collated with the constructed knowledge system, and the machine performs actions in accordance with truth, truth, facts, rules and common sense.

(Corresponding to claim 1)

When the machine performs processing of input information, it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine. A program is created such that when a preset condition is detected, a corresponding operation is executed. If the detection of the condition and the corresponding action are not appropriate, the program installed on the computer is modified.
The input information is converted into patterns and recorded sequentially, and the processing of analyzing and recording information and the structure of information by human instruction and learning is performed. There is no artificial intelligence and software that can be built autonomously.

(Corresponding to claim 2)

When the machine performs processing of input information, it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine. A program is created such that when a preset condition is detected, a corresponding operation is executed. If the detection of the condition and the corresponding action are not appropriate, the program installed on the computer is modified.
The input information is converted into a pattern and sequentially recorded, and the process of analyzing and recording the information and the structure of the information is performed by human instructions and learning. Conventionally, there is no artificial intelligence and software that can be dealt with by adding and changing patterns and connection relations between patterns without adding and changing.

(Corresponding to claim 3)

When the machine performs processing of input information, it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine. A program is created such that when a preset condition is detected, a corresponding operation is executed. If the detection of the condition and the corresponding action are not appropriate, the program installed on the computer is modified.
The input information is converted into patterns and recorded sequentially, and the processing of analyzing and recording information and the structure of information by human instruction and learning is performed. Conventionally, there is no artificial intelligence and software that constructs autonomously and executes appropriate processing for input questions, commands, problems / issues.

(Corresponding to claim 4)

When the machine performs processing of input information, it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine. A program is created such that when a preset condition is detected, a corresponding operation is executed. If the detection of the condition and the corresponding action are not appropriate, the program installed on the computer is modified.
The input information is converted into a pattern and recorded sequentially, and information and the structure of the information are analyzed and recorded by human instruction and learning, and appropriate for the input question, command, problem / issue. In addition to executing processes and adding or changing information processing, it is possible to respond by adding and changing patterns and connection relationships between patterns without adding or changing programs. There is no intelligence and software in the past.

(Corresponding to claim 5)

When the machine performs processing of input information, it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine. A program is created such that when a preset condition is detected, a corresponding operation is executed. If the detection of the condition and the corresponding action are not appropriate, the program installed on the computer is modified.
The input information is converted into a pattern and recorded sequentially, and the value of the input information is evaluated, and the truth, truth, fact, definition, rule, common sense, explanation, hypothesis, prediction, opinion, and impression are identified, classified, and autonomous. Conventionally, there is no artificial intelligence and software for constructing a knowledge system.

(Corresponding to claim 6)

When the machine performs processing of input information, it is realized by incorporating and executing a program created in advance in a program language in a computer mounted on the machine. A program is created such that when a preset condition is detected, a corresponding operation is executed. If the detection of the condition and the corresponding action are not appropriate, the program installed on the computer is modified.
The input information is converted into a pattern and recorded sequentially, and the value of the input information is evaluated, and the truth, truth, fact, definition, rule, common sense, explanation, hypothesis, prediction, opinion, and impression are identified, classified, and autonomous. Conventionally, there is no artificial intelligence and software that constructs a knowledge system and collates input information with the constructed knowledge system and performs actions in accordance with truth, truth, facts, rules and common sense.

Problems to be solved by the invention

(Corresponding to claim 1)

  Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction. In addition, it has been difficult in the past to record the input information sequentially, and to analyze and record the information and the structure of the information by human instruction and learning to build the relationship between the information autonomously. It was.

Problems to be solved by the invention

(Corresponding to claim 2)

  Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction. In addition to sequentially recording the input information, a process for analyzing and recording the information and the structure of the information by human instruction and learning is performed, and additions and changes of programs are performed even when additions and changes occur in information processing It has been difficult in the past to cope with this problem.

Problems to be solved by the invention

(Corresponding to claim 3)

  Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction. In addition to sequentially recording the input information, the process of analyzing and recording the information and the structure of the information by human instruction and learning is performed, and the relationship between the information is autonomously constructed. Conventionally, it has been difficult to execute appropriate processing for problems and issues.

Problems to be solved by the invention

(Corresponding to claim 4)

  Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction. In addition to sequentially recording the input information, the processing to analyze and record the information and the structure of the information by human instruction and learning is performed, and appropriate processing is performed for the input question, command, problem / issue In addition, it has been difficult in the past to cope with the addition or change of information processing without adding or changing the program.

Problems to be solved by the invention

(Corresponding to claim 5)

  Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction. In addition, the entered information is recorded sequentially, and the value of the entered information is evaluated, and the knowledge is autonomously identified and classified into truth, truth, fact, definition, rule, common sense, explanation, hypothesis, prediction, opinion, and impression. Building a system has been difficult in the past.

Problems to be solved by the invention

(Corresponding to claim 6)

  Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction. In addition, the entered information is recorded sequentially, and the value of the entered information is evaluated, and the knowledge is autonomously identified and classified into truth, truth, fact, definition, rule, common sense, explanation, hypothesis, prediction, opinion, and impression. In the past, it was difficult to construct a system and collate the input information with the constructed knowledge system and perform actions in accordance with truth, truth, facts, rules and common sense.

Means for solving the problem

(Corresponding to claim 1)

Human thought is expressed in language, but the information expressed in this language is converted into what is called a pattern. When converted into a pattern in this way, each human thought can be expressed as an individual pattern, and the transition of human thought can be viewed as a change from pattern to pattern. Further, the pattern is not only a language expression, but can express image information, for example, and can process information. Further, it is a concept that has a very wide range that can be handled, for example, it is possible to generate a signal for driving the driving device to perform the operation.
If information is converted into such a pattern and human thought is regarded as a pattern, human thought can be expressed as a change from pattern to pattern. When we convert the language that expresses human thoughts into patterns and analyze the patterns and the relationships between patterns, the types of sentences (normal sentences, question sentences, imperative sentences, etc.) and features (truths, truths, facts, definitions, rules, Common sense, explanation, hypothesis, prediction, opinion, impression).
Sentence types can be trained by machines so that humans can identify the types and characteristics of sentences. The sentence type can be identified as a normal sentence, a question sentence, a command sentence, etc. by analyzing words included in the sentence. Among the features of the sentence, truth, truth, facts, definitions, rules, and common sense are taught by humans for each individual and learned by the machine. This can be done by identifying and recording each type of information (truth, truth, fact, definition, rule, common sense) as each information is entered into the machine. Identification of other types of information (explanation, hypothesis, prediction, opinion, impression) is that the sentence type is not recorded as (truth, truth, fact, definition, rule, common sense) and the word ( I think, think, think, etc.) can be implemented by analyzing.
The sentence can be expressed in the following form:
Main part (what + subject) + predicate (how (when, where, what, why, how) + predicate (how))
Using symbols, it can be expressed as follows.
Main part (M + S) + Predicate ((When: Where: What: Why: How) + V)
M: What S: What, Who Who: When Where: Where What: What What: Why: How: How V: What, What How This way, analyzing the structure of a sentence constitutes a sentence A pattern constituting a sentence can be generated from a pattern for a word. If the language is converted into a pattern and expressed as a set of patterns corresponding to the words that make up the sentence, and the semantic relationship between the words is expressed by the connection relationship between the patterns, the semantic relationship of the sentence It can be expressed by the connection relation of the pattern.
In addition, this connection relationship can express various relationships such as logical relationship, similarity relationship, reciprocal relationship, analogy, cause and result, event and the reason why the event occurred, inheritance of attributes, The definition of relationship can also be expressed flexibly.
Example sentence:
(M1 + S1) + (H1 + V1)
(M2 + S1) + (H2 + V2)
(M3 + S2) + (H3 + V3)
... etc. Newly entered information can be collated with the already recorded knowledge system and evaluated for consistency and novelty. Information that is identified as truth, truth, facts, definitions, rules and common sense and is not consistent with the information recorded is likely to be incorrect. On the other hand, if it is not possible to judge whether the information is consistent or not, it is considered that knowledge sufficient to determine the information has not yet been stored, and therefore human judgment is required. If humans can judge that the information is correct, the result will be notified to the machine. The machine records the information as an explanation or hypothesis, and obtains other information in the future, so that it can be logically expanded with a combination of truth, truth, facts, definitions, rules and common sense as evaluated information. It shall be identified and recorded.
In this way, information is converted into patterns, information types and characteristics are identified, classified and recorded, and information-information relationships (logical relationships, similar relationships, reciprocal relationships, analogies, etc.) Information can be constructed as a knowledge system by expressing the connection relationship between patterns.

Means for solving the problem

(Corresponding to claim 2)

In the present invention, information processing is expressed by transition from pattern to pattern. The pattern is not only a language expression, but can also express image information, for example, and can process information. Further, it is a concept that has a very wide range that can be handled, for example, it is possible to generate a signal for driving the driving device to perform the operation. In the information processing, each processing expression can be expressed by a pattern, and the contents of processing (information retrieval, collation, conditional processing, etc.) can be expressed by a pattern and pattern processing. For example, information can be retrieved by irradiating a recording area with a pattern of information to be retrieved and exciting a correlated pattern.
Information verification can be performed by irradiating the verification source pattern to the verification target pattern and confirming the presence or absence of excitation. Conditional processing can be performed by irradiating a pattern corresponding to a condition with a pattern indicating a state and branching a pattern to be executed depending on the presence or absence of excitation. In this way, it is possible to implement processing defined in a normal programming language by processing between patterns. The contents of the process can be realized by defining a connection between a pattern defining the process and a pattern to be processed. Since processing by patterns can be realized by defining the patterns and connecting the patterns, a very flexible system can be constructed. In the case of processing using a programming language, addition or change of processing requires redesigning the program and restructuring the entire system, such as recompilation. However, in processing by patterns in the present invention, addition of patterns and patterns This can be dealt with by locally changing the connection between them.

Means for solving the problem

(Corresponding to claim 3)

According to the present invention, the information is the main part (what, what (subject)) + predicate (how (when, where, what, why, how), how (predicate)). Since they are organized and recorded in a form, it is possible to easily identify whether or not they have an answer to a question. A sentence representing a question is converted into a pattern, and a pattern related to the content of the question is searched. By confirming whether or not information corresponding to the question content is included as a pattern in the retrieved pattern, an answer can be generated appropriately.
Also, analyze the meaning of the command for the command statement, and collate it with the knowledge area in which past experience records and common sense related to whether or not the response to the command can be implemented, the impact assessment when it is implemented, etc. Therefore, it is possible to carry out appropriate judgment and response.

Means for solving the problem

(Corresponding to claim 4)

According to the present invention, the information is the main part (what, what (subject)) + predicate (how (when, where, what, why, how), how (predicate)). Since they are organized and recorded in a form, it is possible to easily identify whether or not they have an answer to a question. A sentence representing a question is converted into a pattern, and a pattern related to the content of the question is searched. By confirming whether or not information corresponding to the question content is included as a pattern in the retrieved pattern, an answer can be generated appropriately.
Also, analyze the meaning of the command for the command statement, and collate it with the knowledge area in which past experience records and common sense related to whether or not the response to the command can be implemented, the impact assessment when it is implemented, etc. Therefore, it is possible to carry out appropriate judgment and response.
In the present invention, it is possible to implement processing defined in a normal programming language by processing between patterns. The contents of the process can be realized by defining the pattern defining the process and the pattern to be processed by connecting the patterns. As described above, the pattern processing can be realized by defining the patterns and connecting the patterns, so that a very flexible system can be constructed. In the case of processing using a programming language, addition or change of processing requires redesigning the program and restructuring the entire system, such as recompilation. However, in processing by patterns in the present invention, addition of patterns and patterns This can be dealt with by locally changing the connection between them.

Means for solving the problem

(Corresponding to claim 5)

If information is converted into patterns and human thoughts are taken as patterns, human thoughts can be expressed as changes from patterns to patterns. Transform the language that expresses human thought into patterns, analyze patterns and relationships between patterns, sentence types (normal sentences, question sentences, imperative sentences, etc.) and features (truths, truths, facts, definitions, rules, Common sense, explanation, hypothesis, prediction, opinion, impression).
Sentence types can be trained by machines so that humans can identify the types and characteristics of sentences. The sentence type can be identified as a normal sentence, a question sentence, a command sentence, etc. by analyzing words included in the sentence. Of the features of the sentence, truth, truth, facts, definitions, rules, and common sense are taught to the machine for each individual. This can be done by identifying the type of information (truth, truth, fact, definition, rule, common sense) as each information is entered into the machine. Identification of other types of information (explanation, hypothesis, prediction, opinion, impression) is that the sentence type is not recorded as (truth, truth, fact, definition, rule, common sense) and the word ( I think, think, think ... etc.) and analyze it.
Consistency of input information with already recorded information and new information are included by comparing input information with knowledge already recorded (truth, truth, fact, definition, rule, common sense) It can be evaluated whether or not there is. As the relationship between recorded information and information, various relationships such as logical relationships, similar relationships, conflicting relationships, analogy, causes and results, events and the reasons for the occurrence of such events are structured as knowledge. When collating with the input information, the knowledge already recorded is excited and it is possible to detect consistency with the new information and whether the input information includes the new information. If the source of the input information is reliable and the input information is related to the field or theme of interest, and if it contains novelty, it is determined that the input information is valuable, the field of knowledge system, Record in appropriate locations according to the theme and type of sentence. As described above, in the present invention, the information system is constructed as a knowledge system while evaluating the reliability and value of information, so that the problems of information reliability and redundancy can be solved. (Non-credible information or low-value information can be ignored.)

Means for solving the problem

(Corresponding to claim 6)

  In the present invention, since it is constructed as a knowledge system while evaluating the reliability and value of information, the problems of information reliability and redundancy are solved. Since the action decision for the input information is carried out by collating with a reliable knowledge system, it is possible to make a judgment based on truth, truth, facts, rules and common sense.

Embodiment 1 of the Invention

(Corresponding to claim 1)

FIG. 1 is a diagram showing the configuration of an autonomous knowledge analyzer in one embodiment of the present invention.
In FIG. 1, reference numeral 1 denotes a pattern converter that converts information into a pattern. Reference numeral 2 denotes a pattern recorder that records patterns, connection relations between patterns, relations between patterns, and pattern excitation history. Reference numeral 3 denotes a pattern register that registers and changes a pattern and a connection relation between the patterns by a human instruction or autonomously. Reference numeral 4 denotes a pattern controller for controlling pattern processing. Reference numeral 5 denotes a pattern inverse converter for converting a pattern into information. Reference numeral 6 denotes a pattern analyzer that analyzes patterns and relationships between patterns.

Next, the operation will be described.
In FIG. 1, the pattern converter converts information into a pattern. The converted pattern is analyzed and divided into a main part and a predicate. The main part is divided into a part that modifies the subject and the subject, and the predicate is divided into a part that modifies the predicate and the predicate. Further, the part that modifies the predicate is identified and arranged as information corresponding to 5W1H. This arrangement can be performed by analyzing the words included in the sentence and the attributes of the words. FIG. 2 shows an example of typical arrangement. FIG. 2 shows how the information included in the sentence is organized and recorded as a pattern in the pattern information recording unit.
Using symbols, it can be expressed as follows.
Main part (M + S (Who, What)) + Predicate ((When: Where: What: What: How) + V)
M: What S: What, Who H1 (When): When H2 (Where): Where H3 (What): What H4 (Why): Why H5 (How): How V: What, How However, this sort of arrangement clarifies the type of information contained in the sentence, and also clarifies the structure of the sentence when analyzing the meaning of the sentence. The set of sentence patterns converted and arranged into patterns is sequentially recorded in the recording module of the second pattern recorder.
FIG. 3 shows an example of arrangement by sentence analysis. Each recording module records each pattern corresponding to the organized information. FIG. 4 shows how each pattern shown in FIG. 3 is recorded in each recording module of the pattern recorder.

The pattern analyzer 6 analyzes sentence types and characteristics. Sentence types can be trained by machines so that humans can identify the types and characteristics of sentences. The sentence type can be identified as a normal sentence, a question sentence, a command sentence, etc. by analyzing words included in the sentence. Of the features of the sentence, truth, truth, facts, definitions, rules, and common sense are taught to the machine for each individual. It also identifies and records the type of information (truth, truth, fact, definition, rule, common sense) as each information is entered into the machine. Identification of other types of information (explanation, hypothesis, prediction, opinion, impression) is that the sentence type is not recorded as (truth, truth, fact, definition, rule, common sense) and the word ( I think, think, think, etc.) can be implemented by analyzing.
The pattern registering device 3 verifies the input information with the recording module of the pattern recorder, and checks whether the same pattern as the input information is recorded. If recorded, the corresponding pattern is excited, and if not recorded, it is recorded and excited as a new pattern in the recording module. Excitation refers to a state in which the recording module is activated by collating information. The history of the excited recording module is sequentially recorded in the pattern recorder.
Next, another information is input, and it is confirmed whether or not to collate with the already recorded pattern in the same manner as described above. When the recording module is excited, the history of the recording module that was excited retroactively from the point in time when the recording module was excited is called from the location where the excitation pattern history of the pattern recorder was recorded, and the recording module that was excited this time Transfer the connection information to the connection information recording part. By transferring, the coupling coefficient indicating the strength of the connection relationship with the recording module considered to be related to the excitation of the recording module is sequentially increased. Thereby, the connection relationship with the recording module related to exciting the recording module excited this time is strengthened. When the recording module excited this time is excited again, the history of the recording module related to the excitation of this recording module is transferred back to the past of the period set at that time. At this time, the transfer is accumulated and accumulated on the previously transferred information. In addition, the method of increasing the coupling coefficient indicating the strength of the transfer relationship and the connection relationship of the related recording module detects the correlation with the already recorded history information for each excitation history in addition to simple superposition, A method for increasing the appropriate coupling coefficient shall be included. Although this method is a little complicated in processing, it is possible to avoid an erroneous determination due to an increase in coupling coefficient due to the history of the recording module that is not directly related to the excitation of the recording module included in the history information.
The history of the excited pattern is recorded, and every time a pattern is excited, the excitation history before the pattern is excited is transferred to the connection information recording section of the pattern, so that the pattern and the pattern reach the excitation. The scenario up to this point will be recorded. Usually, a thought pattern changes according to a number of conditions, but the transition follows a past context (context). The recording, transcription, and irradiation (collation) of each recording module described above correspond to transition of thought in consideration of the context.

The pattern controller 4 controls pattern processing. When the type of sentence is analyzed by the pattern analyzer 6, a pattern corresponding to each sentence and a connection relation between the patterns are generated. This connection relationship can be generated by human instruction or autonomously. The generation of the connection relationship can be performed by a human instruction by designating each pattern and designating which pattern is to be connected to which pattern. The generation of the autonomous connection relationship can be performed by specifying the type of connection relationship corresponding to the analysis result regarding the type of sentence and the structure of the sentence.
FIG. 5 and FIG. 6 show how various patterns and patterns are recorded in the inter-pattern relationship recording unit.
Once the sentence type and sentence structure are analyzed, the pattern can be converted to an equivalent pattern. Since sentences are composed of words, sentence patterns are constituents of word patterns. Since the sentence expresses a certain concept, the pattern corresponding to the sentence also corresponds to the certain concept. Since the sentence concept is composed of words, the sentence concept inherits the word concept. Similarly, sentence patterns inherit the concept inherited by word patterns. Analysis of the sentence reveals the following structure, for example. (See remarks in Figure 7)
Definition: Main part 2 ((S2) is) + Predicate 2 ((H2).)
Attribution relationship: Main part 7 ((S7) is) + Predicate 7 ((H7))
Relationship definition: main part 4 ((S4-1) and (S4-2) are) + predicate 4 (assuming V4),
Main part 5 ((S5) is) + Predicate 5 (V5)
S2: acid, H2: substance that dissolves in water to produce H + S7: hydrochloric acid, H7: acid S4-1: acid, S4-2: base, V4: reaction S5: salt and water, V5: resulting definition Since the main part 2 and the predicate 2 are in an equivalent relationship, they can be expressed as follows.
Main part 2 述 Predicate 2
In the attribution relationship, the main part 7 has an attribution relationship with the predicate 7, and can be expressed as follows.
Main part 7 ⇒ Predicate 7
In the relationship definition, various cases can be expressed. In this example, it can be expressed as follows.
Main part 4 ((S4-1) and (S4-2)) + predicate 4 ⇒ main part 5 + predicate 5

By analyzing the sentence in this way, the pattern of the sentence can be expressed by the connection relation of the patterns constituting the sentence. The pattern constituting the sentence represents the position in the sentence depending on the connection method. Since the sentence component is a word, it inherits the characteristics of the word, but the sentence component composed of the word has a multi-layered structural relationship in which the positioning is expressed by the connection relationship described above. Can be expressed.
FIG. 8 shows how the text shown in FIG. 7 is recorded in the pattern recorder and the pattern is equivalently converted.
In this way, the pattern expresses a sentence, expresses the relationship (positioning) between the components of the sentence, and also the relationship between the sentence and the sentence (logical relationship, cause and result, purpose and means, etc., various relationships other than the logical relationship Can be expressed. In addition, since the feature defined between the patterns has a feature that it is inherited by a higher-order pattern, various concepts (thinking) including abstract concepts can be expressed by connecting the patterns.
FIG. 9 shows a state in which input information is analyzed, and equivalent pattern conversion and pattern-pattern connection relations are generated. When information is recorded in such a form, this recording system corresponds to a knowledge system. In the example of the text in FIG. 7, when the question “when hydrochloric acid and sodium hydroxide react” is input to the knowledge obtained by inputting the information of NO1-NO8, “salt and water are generated. The pattern "." Is excited, and information that is not input can be generated by logical development between the patterns.
Examples of what types of connection relations are designated for sentence types and sentence structures are shown in FIGS. FIG. 10 shows logical relationships, definitions, attribution relationships, and similar relationships, FIG. 11 shows relationship definitions, and FIG. 12 shows inference development, inheritance of common features, and identification of individual features. 13 to 15 show a state in which the type of sentence and the type of connection relation corresponding to the sentence structure for FIGS. 10 to 12 are designated.
The connection relationship between patterns is autonomously generated according to a human instruction or an analysis result of the pattern analyzer, and is constructed by connection information autonomous generation in the connection information recording unit of the pattern recorder by a series of information inputs.
The information input from the above is converted into a pattern, and syntactic analysis, 5W1H organization, inter-pattern connection corresponding to the sentence type and structure are performed and recorded in the pattern recorder. Since the pattern recorded in this way has a connection relationship with other patterns, it is possible to simulate an action corresponding to the transition of human thought.

Next, the realization means of the present invention is completely different from the expert system, finite automaton, neural network, and programming language (Prolog, LISP, object-oriented language (C, C ++, Java, etc.)) used in conventional artificial intelligence, robots, etc. A description will be given of differences in configuration and function, and comparison with conventional artificial intelligence technology is shown in FIGS.
FIG. 16 shows a comparison with the expert system.
The expert system is a program composed of rules, and is composed of a fixed inference engine and a variable knowledge database. The inference engine performs inference using rule groups, and propositional logic, predicate logic, recognition logic, modal logic, temporal logic, fuzzy logic, etc. are used as rule groups as inference logic. A knowledge database is generally constructed by inputting expert knowledge, but when formulated, problems such as contradictions among rules have been pointed out in the literature.
On the other hand, the processing in the present invention is not fixed like a program, and the processing is expressed by a change from pattern to pattern. The process can be changed by changing the connection between patterns. In the expert system, only logical expansion (propositional logic, predicate logic, recognition logic, modal logic, temporal logic, fuzzy logic, etc.) by inference using rule groups is the object of processing, but in the present invention, processing other than logical expansion is performed. (Relevance definition, novelty detection, evaluation, etc.) are also possible, making the system more flexible than the expert system. In addition, while the expert system is a half-data driven system such as an inference engine (fixed) and a knowledge database (variable), in the present invention, functions and processes can be handled as data, and a complete data driven system. It has become.
Since an expert system is a program composed of a group of rules, it is also subject to program language restrictions. Since the present invention is a complete data driven system that can handle data in terms of processing, it is possible to easily change or add processing during execution of processing. This is performed by changing the connection between the pattern expressing the processing and the pattern. Normally, in the case of programming processing in a programming language, reprogramming and recompilation are required when processing changes are implemented. However, in the present invention, processing changes, additions, etc. are made by changing the connection relationship between corresponding patterns locally. Since it can be implemented by changing, there is no need to rebuild the entire system like recompilation.
FIG. 17 shows a comparison with a finite automaton.
A finite automaton has a restriction that only one state can be obtained at a certain point in time. Due to this restriction, it is difficult to perform an operation of simultaneously taking a plurality of states that can be realized in the present invention. Of human thoughts, it is appropriate to express what corresponds to consciousness with the action of taking one state (excited thoughts express actions that suppress other thoughts), but correspond to human unconsciousness. As an expression of the action, the restriction that only one state can be taken is considered inappropriate. (Because unconscious thinking has weak suppression between thoughts, and multiple thoughts can work in parallel)
In the present invention, it is not necessary to restrict the pattern excitation to one, so there is no restriction like a finite automaton. (Thinking corresponding to consciousness and unconsciousness can be expressed as a pattern.) Parallel processing by multiple pattern excitation is possible. (The problem of interference between patterns in parallel processing can be solved by separating the processing lines.)
FIG. 18 shows a comparison with a neural network.
In the present invention, the connection between the recording modules can be handled very flexibly. Patterns and patterns can be connected by specifying them individually. Any combination such as many-to-many coupling, multi-stage coupling, feedforward coupling, feedback coupling, etc. can be specified and set. On the other hand, in a neural network, artificial neurons in the input layer, intermediate layer, and output layer are generally connected in series, and the connection between artificial neurons is performed by synaptic learning. As the synapse learning, an error back propagation method (back propagation) is generally used. However, back propagation has the following drawbacks pointed out in the literature.
• In backpropagation, the transfer function used in artificial neurons must be differentiable.
-Convergence in learning by backpropagation is very slow.
・ Back-propagation learning does not always converge.
・ In many cases, it converges to a local minimum error point instead of a global optimal solution. On the other hand, the present invention does not have the above-mentioned drawbacks because the method of repetitive optimum value search such as back propagation is not used for the connection between the recording modules.
FIG. 19 shows a comparison with a programming language.
Since the present invention is a complete data driven system that can handle data in terms of processing, it is possible to easily change or add processing during execution of processing. Normally, in the case of programming processing using a programming language, reprogramming and recompilation are required when processing changes are implemented. Since it can be implemented by changing, there is no need to rebuild the entire system like recompilation.
In programming languages, the names of variables, constants, functions, etc. used in the program will not work unless they are strictly specified. On the other hand, in the present invention, since processing is performed by connecting patterns, it is possible to construct a system that operates even if there is some ambiguity, such as enabling processing with similar patterns. In other words, it is possible to construct a system that can be processed even if there is some ambiguity and that is closer to human thought processing.
The implementation method of the present invention not only covers the functions of expert systems, finite automata, neural networks and programming languages employed in conventional artificial intelligence, robots, etc., but also the functions and flexibility they do not have (completely Data-driven system, flexibility of function expansion (functions can be added by local changes), support for systems that allow ambiguity, unconscious (multiple states) simulation functions, support for parallel processing, variety of connections The system has the characteristics (feed forward, feedback, many-to-many coupling, etc.), there is no restriction of differentiability, and there is no problem of convergence in backpropagation).

Embodiment 2 of the Invention

(Corresponding to claim 2)

FIG. 20 and FIG. 21 relate to claim 2 by adding and changing patterns and connection relationships between patterns without adding and changing programs even when additions and changes occur in information processing. 1 illustrates an example of artificial intelligence and software that can be used.
In FIG. 20, it is assumed that the transition from the following pattern to the pattern is performed as the baseline processing.
Baseline processing AS → A10 → A20 → A30 → A40 → AE
(Here, AS is a pattern indicating the start of the process, A10, A20, A30, and A40 are patterns indicating the contents of the process, and AE is a pattern indicating the end of the process.)
In the above process, when the pattern AS indicating the start of the process is excited first, A10, A20, A30, and A40 indicating the contents of the process are sequentially excited, finally reaching the pattern AE indicating the end, and the process is performed. Complete.
Next, suppose you want to change the contents of the baseline processing to the following.
Contents of processing to be changed A10 → A20 → A30 → A40 ⇒ A11 → A21 → A31 → A41
In this case, as shown in FIG. 21, A11, A21, A31, and A41 are registered in the pattern recording, and the connection is recorded in the inter-pattern relation recording unit so that the process proceeds between A11 → A21 → A31 → A41. To do. Next, in the baseline processing, the pattern AS excites the pattern A10 at the start of processing, but this connection is changed so that the pattern AS is connected to the pattern A11. Further, the connection is changed so as to excite the pattern AE indicating the end after the end of the processing pattern A41.
When the above changes are completed, when the pattern AS is excited, the processes A11, A21, A31, and A41 are continuously executed, and finally the pattern AE indicating the end is reached and the process is completed.
As described above, it is possible to greatly change the contents of processing by registering a new processing pattern and changing the connection between partial patterns. Further, since this change can be suppressed to a local change, it is not necessary to perform a system-wide reconfiguration such as a large program change or recompilation like processing using a programming language.

Embodiment 3 of the Invention

(Corresponding to claim 3)

FIG. 22 is a diagram showing the configuration of an autonomous knowledge analyzer in one embodiment of the present invention.
In FIG. 22, 1 is a pattern converter for converting information into a pattern. Reference numeral 2 denotes a pattern recorder that records patterns, connection relations between patterns, relations between patterns, and pattern excitation history. Reference numeral 3 denotes a pattern register that registers and changes a pattern and a connection relation between the patterns by a human instruction or autonomously. Reference numeral 4 denotes a pattern controller for controlling pattern processing. Reference numeral 5 denotes a pattern inverse converter for converting a pattern into information. Reference numeral 6 denotes a pattern analyzer that analyzes patterns and relationships between patterns. Reference numeral 7 denotes a pattern processor that executes appropriate processing in accordance with patterns and analysis results between patterns.

Next, the operation will be described.
In FIG. 22, the pattern converter converts information into a pattern. The converted pattern is analyzed and divided into a main part and a predicate. The main part is divided into a part that modifies the subject and the subject, and the predicate is divided into a part that modifies the predicate and the predicate. Further, the part that modifies the predicate is identified and arranged as information corresponding to 5W1H. This arrangement can be performed by analyzing the words included in the sentence and the attributes of the words.
By such arrangement, the type of information included in the sentence becomes clear, and the structure of the sentence becomes clear in analyzing the meaning of the sentence. The set of sentence patterns converted and arranged into patterns is sequentially recorded in the recording module of the second pattern recorder.
The pattern analyzer 6 analyzes sentence types and characteristics. Sentence types can be trained by machines so that humans can identify the types and characteristics of sentences. The sentence type can be identified as a normal sentence, a question sentence, a command sentence, etc. by analyzing words included in the sentence. Of the features of the sentence, truth, truth, facts, definitions, rules, and common sense are taught to the machine for each individual. This can be done by identifying and recording each type of information (truth, truth, fact, definition, rule, common sense) as each information is entered into the machine. Identification of other types of information (explanation, hypothesis, prediction, opinion, impression) is that the sentence type is not recorded as (truth, truth, fact, definition, rule, common sense) and the word ( I think, think, think, etc.) can be implemented by analyzing.
The pattern registering device 3 verifies the input information with the recording module of the pattern recorder, and checks whether the same pattern as the input information is recorded. If recorded, the corresponding pattern is excited, and if not recorded, it is recorded and excited as a new pattern in the recording module. Excitation refers to a state in which the recording module is activated by collating information. The history of the excited recording module is sequentially recorded in the pattern recorder.
The history of the excited pattern is recorded, and every time a pattern is excited, the excitation history before the pattern is excited is transferred to the connection information recording section of the pattern, so that the pattern and the pattern reach the excitation. The scenario up to this point will be recorded. Usually, a thought pattern changes according to a number of conditions, but the transition follows a past context (context). The recording, transcription, and irradiation (collation) of each recording module described above correspond to transition of thought in consideration of the context.
The pattern controller 4 controls pattern processing. When the type of sentence is analyzed by the pattern analyzer 6, a pattern corresponding to each sentence and a connection relation between the patterns are generated. This connection relationship can be generated by human instruction or autonomously. The generation of the connection relationship can be performed by a human instruction by designating each pattern and designating which pattern is to be connected to which pattern. The generation of the autonomous connection relationship can be performed by specifying the type of connection relationship corresponding to the analysis result regarding the type of sentence and the structure of the sentence.
Once the sentence type and sentence structure are analyzed, the pattern can be converted to an equivalent pattern. Since sentences are composed of words, sentence patterns are constituents of word patterns. Since the sentence expresses a certain concept, the pattern corresponding to the sentence also corresponds to the certain concept. Since the sentence concept is composed of words, the sentence concept inherits the word concept. Similarly, sentence patterns inherit the concept inherited by word patterns.
By analyzing the sentence, the pattern of the sentence can be expressed by the connection relation of the patterns constituting the sentence. The pattern constituting the sentence represents the position in the sentence depending on the connection method. Since the sentence component is a word, it inherits the characteristics of the word, but the sentence component composed of the word has a multi-layered structural relationship in which the positioning is expressed by the connection relationship described above. Can be expressed.
In this way, the pattern expresses a sentence, expresses the relationship (positioning) between the components of the sentence, and also the relationship between the sentence and the sentence (logical relationship, cause and result, purpose and means, etc., various relationships other than the logical relationship Can be expressed. In addition, since the feature defined between the patterns has a feature that it is inherited by a higher-order pattern, various concepts (thinking) including abstract concepts can be expressed by connecting the patterns.
The connection relationship between patterns is autonomously generated according to a human instruction or an analysis result of the pattern analyzer, and is constructed by connection information autonomous generation in the connection information recording unit of the pattern recorder by a series of information inputs.
Reference numeral 7 denotes a pattern processor that performs appropriate processing according to the analysis result of the pattern analyzer 6.
The pattern analyzer analyzes the type of input information, identifies a normal sentence, a question sentence, a command sentence, and the like, and performs an appropriate process according to the type of each sentence.
Appropriate processing means that the value of information is evaluated if it is a normal sentence, and if it is important information or information of a field or theme of interest, it is organized and recorded in the knowledge system. Information that is determined to have no value is discarded, or stored after a certain period of time and then discarded.
If it is a question sentence, the meaning of the question is analyzed, and what kind of information is appropriate as an answer is analyzed. Whether or not information corresponding to the answer to the question is recorded in the knowledge system is irradiated to the recording area of the knowledge system using a search key. When related information that can be a candidate for an answer is retrieved, the characteristics, attributes, etc. of the information are analyzed to determine whether the answer is appropriate. If an appropriate answer is included, an answer to the question is generated, and if not included, a process for notifying that is performed.
If it is a command sentence, the meaning of the command is analyzed, and what kind of treatment is required is analyzed. Next, we will evaluate the feasibility of the treatment and the impact of doing so. Whether or not it can be implemented is determined in consideration of whether or not it has been implemented in the past. Regarding the effects of the treatment, the results of the past implementation and the results to be brought are compared with the common sense included in the knowledge system, and judged to be appropriate.
FIG. 23 to FIG. 25 show processing examples of normal sentences, question sentences, and command sentences.
FIG. 23 shows a processing example of a normal sentence. In this example, processing is divided into 10 stages. (Process division and synthesis can be implemented as appropriate.)
Each process will be described below.
Similar to the pattern recording module, the pattern processing module can record patterns, relationships between patterns, and connection information, and has a function as a processor. The functions of the processor include pattern capture, pattern irradiation, inter-pattern processing (logical operation, difference detection, synthesis, other operations), pattern recording, pattern output, and the like.
(1) Input information capture When a normal sentence is detected as input information, it is input to the pattern processing module of the pattern processor 7.
(2) Field / theme analysis of input information In order to analyze what field and theme the input information relates to, the pattern of field / theme analysis is excited.
(3) Irradiation of input information & field / theme analysis request The pattern recorder is irradiated with the pattern of input information and field / theme analysis request. In response to the input information and the pattern of the field / theme analysis request in the pattern recorder, the pattern corresponding to the corresponding field / theme is excited.
(4) Input information field / theme analysis result storage A pattern corresponding to the excited corresponding field / theme is stored.
(5) Irradiation of input information & field / theme analysis result The pattern recorder is irradiated with a pattern corresponding to the field / theme corresponding to the input information and the input information.
(6) Related information excitation Patterns related to input information are excited in fields / themes related to input information.
(7) Input information / related information comparison The input information and the already recorded related pattern are sequentially compared to evaluate whether the input information is novel. Also, consistency (consistency, inconsistency) with already recorded information is evaluated. If there is an inconsistency with the information already recorded, the reliability of the information (evaluable by comparing the reliability of the source information group that is the basis of the information) is evaluated, and it is already recorded with the input information Determine which information is correct.
(8) Comparison result storage The evaluation result of the value (novelty, reliability) of input information is stored.
(9) Determination of treatment of input information The value of the input information is judged, and it is determined whether additional information is recorded as related information related to the relevant field / theme or whether the input information is discarded.
(10) Execution of treatment A treatment or discarding additionally recorded as related information related to the relevant field / theme is executed.
FIG. 24 shows an example of question text processing. In this example, processing is divided into 10 stages. (Process division and synthesis can be implemented as appropriate.)
Each process will be described below.
Similar to the pattern recording module, the pattern processing module can record patterns, relationships between patterns, and connection information, and has a function as a processor. The functions of the processor include pattern capture, pattern irradiation, inter-pattern processing (logical operation, difference detection, synthesis, other operations), pattern recording, pattern output, and the like.
(1) Interrogation of question text When a question text is detected as input information, it is input to the pattern processing module of the pattern processor 7.
(2) Field / theme analysis of question sentence Examine the pattern of field / theme analysis in order to analyze what field and theme the question sentence relates to.
(3) Irradiation of question / field / theme analysis request The pattern recorder is irradiated with the pattern of the question / field / theme analysis request. In response to the question sentence and the pattern of the field / theme analysis request in the pattern recorder, the pattern corresponding to the corresponding field / theme is excited.
(4) Field of question text / theme analysis result storage Patterns corresponding to the excited relevant fields / themes are stored.
(5) Irradiation of question sentence / field / theme analysis result The pattern recorder is irradiated with a pattern corresponding to the field / theme corresponding to the question sentence and the question sentence.
(6) Excitation of related information Patterns related to question sentences are excited in fields / themes related to question sentences.
(7) Questionable text / related information comparison When a pattern related to a question text is excited in a field / theme related to the question text, it is stored in a pattern processing module and analyzed. Attributes required for answers corresponding to questions (for example, what is required for answers corresponding to 5W1H? Subject (person, thing), time, place, subject (person, thing), reason, means?) And related information Compare the attributes of the information contained in and evaluate whether the necessary information is included as an answer.
(8) Evaluation result storage If necessary information is included in the related information, it is extracted.
(9) Answer generation An answer to the question sentence is generated from the information extracted as the answer and the question sentence.
(10) Answer output Generated answer is output.
FIG. 25 shows a processing example of a command statement. In this example, processing is divided into 10 stages. (Process division and synthesis can be implemented as appropriate.)
Each process will be described below.
Similar to the pattern recording module, the pattern processing module can record patterns, relationships between patterns, and connection information, and has a function as a processor. The functions of the processor include pattern capture, pattern irradiation, inter-pattern processing (logical operation, difference detection, synthesis, other operations), pattern recording, pattern output, and the like.
(1) Command sentence fetching When a command sentence is detected as input information, it is input to the pattern processing module of the pattern processor 7.
(2) Intention analysis of command sentence In order to analyze what kind of intention a command sentence includes, an intention analysis pattern is excited.
(3) Irradiation of command sentence and intention analysis request The pattern recorder is irradiated with the pattern of the command sentence and the intention analysis request. In response to the command statement and the pattern of the intention analysis request in the pattern recorder, the pattern corresponding to the corresponding intention is excited.
(4) Intention analysis result storage of command sentence A pattern corresponding to the excited intention is stored.
(5) Irradiation of command sentence and intention analysis result The pattern recorder is irradiated with a command sentence and a pattern corresponding to the intention of the command sentence.
{Circle around (6)} Excitation record of related instruction Excitation of the pattern corresponding to the execution record related to the intent of the instruction sentence and the instruction sentence is excited.
(7) Evaluation storage at the time of command execution The pattern of evaluation (execution availability, influence, etc.) related to command execution is extracted from the related execution record pattern and stored.
(8) Determining whether or not the command is executed Judge whether or not the command is executed based on the evaluation result regarding the command execution.
(9) Instruction execution procedure generation When instruction execution is determined, the instruction execution procedure is extracted from the instruction execution record.
(10) Execution of instruction (or notify of reason for refusal)
FIG. 26 shows an example of problem / task processing. In this example, processing is divided into 10 stages. (Process division and synthesis can be implemented as appropriate.)
Each process will be described below.
Similar to the pattern recording module, the pattern processing module can record patterns, relationships between patterns, and connection information, and has a function as a processor. The functions of the processor include pattern capture, pattern irradiation, inter-pattern processing (logical operation, difference detection, synthesis, other operations), pattern recording, pattern output, and the like.
(1) Problem / Problem Capture When a problem / problem is detected as input information, it is input to the pattern processing module of 7 pattern processor.
(2) Problem / Problem Type Analysis Examine the pattern of problem / problem type analysis requests in order to analyze the type of problem / problem.
(3) Irradiation of problem / problem & mold analysis request Irradiate the pattern recorder with the pattern of problem / problem and mold analysis request. In response to the pattern of the problem / problem and the type analysis request in the pattern recorder, the pattern corresponding to the corresponding type is excited.
(4) Type analysis result storage of problem / problem A pattern corresponding to the excited type is stored.
(5) Irradiation of problems / problems & mold analysis results Irradiate the pattern recorder with patterns corresponding to problems / problems and molds.
(6) Excitation record of similar problems / tasks Excitation patterns corresponding to problem / problem and type similar problem / task records.
(7) Evaluation and storage of solutions / results Extract patterns of solution / result evaluation (executability / results, etc.) from patterns of implementation records of similar problems / issues and store them.
(8) Extracting the best solution From the solutions to the problem / problem, extract the one that can be implemented and has the best evaluation of the result.
(9) Generate procedure of solution Extract the procedure from the determined solution.
(10) Execution of the solution procedure The solution procedure is sequentially executed.

Embodiment 4 of the Invention

(Corresponding to claim 4)

27 and 28 relate to claim 4 and can cope with addition and change of patterns and connection relations between patterns without adding and changing programs even when additions and changes occur in information processing. It is shown that.
The correspondence regarding the recording module of the pattern recorder 2 is described with reference to FIGS. 20 and 21 in the previous section. Even when additions and changes occur in the processing of information, it is possible to cope with the addition and change of patterns and connection relationships between patterns without adding and changing programs.
Below, the response | compatibility regarding the pattern processing module of 7 pattern processors is demonstrated.
In FIG. 27, it is assumed that the processing in the pattern processor is set in (1) to (4). At this time, when the pattern processing module (1) in the pattern processor is excited, the process proceeds from (1) → (2) → (3) → (4) and is completed. Next, when it is desired to expand the processing, as shown in FIG. 28, pattern processing modules (5) to (10) are added to the pattern processing module of the pattern processor, and the pattern processing module (4) is excited. After that, when the pattern processing module {circle over (1)} is excited by adding a connection to excite the pattern processing module {circle over (5)}, the processing proceeds from {circle over (1)} → {2} → {3} → {4} → {5} The processing function is expanded in the order of ▼ → ▲ 6 ▼ → ▲ 7 ▼ → ▲ 8 ▼ → ▲ 9 ▼ → ▲ 10 ▼.
Even when the processing is added and changed in this way, it is possible to cope with it by adding and changing the pattern and the connection relation between the patterns without adding and changing the program.

Embodiment 5 of the Invention

(Corresponding to claim 5)

Examples of artificial intelligence and software that evaluates the value of input information, identifies and classifies truths, truths, facts, definitions, rules, common sense, explanations, hypotheses, predictions, opinions, and impressions and autonomously builds a knowledge system Will be described.
If information is converted into patterns and human thoughts are taken as patterns, human thoughts can be expressed as changes from patterns to patterns. When we convert the language that expresses human thoughts into patterns and analyze the patterns and the relationships between patterns, the types of sentences (normal sentences, question sentences, imperative sentences, etc.) and features (truths, truths, facts, definitions, rules, Common sense, explanation, hypothesis, prediction, opinion, impression).
Sentence types can be trained by machines so that humans can identify the types and characteristics of sentences. The sentence type can be identified as a normal sentence, a question sentence, a command sentence, etc. by analyzing words included in the sentence. Of the features of the sentence, truth, truth, facts, definitions, rules, and common sense are taught to the machine for each individual. This can be done by identifying and recording each type of information (truth, truth, fact, definition, rule, common sense) as each information is entered into the machine. Identification of other types of information (explanation, hypothesis, prediction, opinion, impression) is that the sentence type is not recorded as (truth, truth, fact, definition, rule, common sense) and the word ( I think, think, think, etc.) can be implemented by analyzing.
Consistency of input information with already recorded information and new information are included by comparing input information with knowledge already recorded (truth, truth, fact, definition, rule, common sense) It can be evaluated whether or not there is. As the relationship between recorded information and information, various relationships such as logical relationships, similar relationships, conflicting relationships, analogy, causes and results, events and the reasons for the occurrence of such events are structured as knowledge. When collating with the input information, the knowledge already recorded is excited and it is possible to detect consistency with the new information and whether the input information includes the new information. If the source of the input information is reliable and the input information is related to the field or theme of interest, and if it contains novelty, it is determined that the input information is valuable, the field of knowledge system, Record in appropriate locations according to the theme and type of sentence. As described above, in the present invention, the information system is constructed as a knowledge system while evaluating the reliability and value of information, so that the problems of information reliability and redundancy can be solved.
FIG. 29 and FIG. 30 show an example in which input information is checked against knowledge already recorded (truth, truth, fact, definition, rule, common sense) to confirm whether or not it matches. A processing example will be described below.
(1) Information acquisition When information is input, it is input to the pattern processing module of 7 pattern processor.
(2) Fields of information / theme analysis In order to analyze the fields and themes related to information, the field / theme analysis request pattern is excited.
(3) Information & Field / Theme Analysis Request Irradiation The pattern recorder is irradiated with the information and field / theme analysis request pattern. In response to the information and the pattern of the field / theme analysis request in the pattern recorder, the pattern corresponding to the corresponding field / theme is excited.
(4) Information field / theme analysis result storage A pattern corresponding to the excited corresponding field / theme is stored.
(5) Irradiation of information & field / theme analysis result The pattern recorder is irradiated with information and a pattern corresponding to the field / theme corresponding to the information.
(6) Excitation of truth, rules, common sense, etc. In fields / themes related to information, patterns of truth, rules, common sense, etc. related to information excite.
(7) Comparison of information with truth, rules, common sense, etc. When patterns of truth, rules, common sense, etc. related to information are excited in fields / themes related to information, they are stored in a pattern processing module and analyzed. An analysis is performed to determine whether the information contained in the information is consistent with the relevant truths, rules, common sense, etc. If it is consistent with the truth, rules, and common sense related to the information content, it is judged as consistent. If the content of information contradicts the truth, rules, or common sense, it is judged as inconsistent. If neither matching nor mismatching can be determined, matching / mismatching is not detected.
(8) Verification result Outputs verification results with truth, rules, common sense, etc.
FIG. 29 shows a case where the input information is consistent with truth, rules, common sense, and the like. In this case, the pattern stored in the recording module storing the truth, rules, common sense, etc. recorded by the pattern corresponding to the input information is excited, and the truth, rules already recorded by the input pattern are recorded. It is detected that it is consistent with the pattern storing common sense and the like.
FIG. 30 shows a case where input information is not consistent with truth, rules, common sense, and the like. In this case, the pattern corresponding to the input information excites the related pattern stored in the recording module storing the recorded truth, rule, common sense, etc. as the related pattern, but does not match. On the other hand, the pattern corresponding to the input information excites a pattern having the opposite meaning to the recording module such as truth, rule, and common sense. Since the pattern of input information excites a pattern having the opposite meaning to truth, rules, common sense, etc., it is detected that the matching result is inconsistent.
As shown in the above example, it can be evaluated whether the input information is consistent with truth, truth, fact, definition, rule, or common sense. Truth, truth, facts, definitions, rules, and common sense can be extended sequentially by human instruction. In addition, the patterns belonging to truth, truth, facts, definitions, rules, and common sense are between patterns developed by relationships such as logic, definition, similarity, attribution, and relationship definitions as described in [0027]. It can be determined that the pattern derived from the connection relationship is reliable and correct. This is because correct development (derived from relationships such as definition of logic, definition, similarity, attribution, relationship, etc.) is premised on the premise of truth, truth, facts, definitions, rules, and common sense.
A connection system between patterns that is correctly developed (derived from relationships such as definition of logic, definition, similarity, attribution, and relationship) based on truth, truth, facts, definitions, rules, and common sense can be judged as a correct knowledge system. Therefore, the input information is irradiated to the recording module that builds this knowledge system, and as a result, the excited pattern group is a correct knowledge group derived using the input information as a trigger.

Embodiment 6 of the Invention

(Corresponding to claim 6)

Describes artificial intelligence and software implementations that collate input information with the constructed knowledge system and perform actions in accordance with truth, truth, facts, rules and common sense.
As described in the section [0033], by irradiating the input information to the area of the pattern recording module constructed by the pattern group of truth, truth, fact, rule, and common sense, the input information becomes truth, truth, In fact, it can be judged whether it is consistent with the rules and common sense. Artificial intelligence and software that performs actions in accordance with truth, truth, facts, rules and common sense by allowing actions consistent with truth, truth, facts, rules and common sense and prohibiting actions that do not match Can be realized.

Effect 1 of the Invention

(Corresponding to claim 1)

According to the first invention, the process of analyzing and recording information and the structure of the information by human instruction and learning is performed, and the relationship between the information is expanded and the knowledge is connected by the connection relation between the patterns and the process between the patterns. Can be autonomously implemented.
Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction.
According to the present invention, it is not necessary to design and implement a program or the like for information processing, so that development effort can be greatly reduced.

Effect 2 of the invention

(Corresponding to claim 2)

According to the second invention, the process of analyzing and recording information and the structure of the information is performed by human instruction and learning, and the relationship between the information is expanded and the knowledge is increased by the connection relation between the patterns and the process between the patterns. Can be autonomously implemented. In addition, even when additions and changes occur in information processing, the program can be implemented without additions and changes.
Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction.
According to the present invention, it is not necessary to design and implement a program or the like for information processing, so that development effort can be greatly reduced.

Effect 3 of the invention

(Corresponding to claim 3)

According to the third invention, the process of analyzing and recording information and the structure of the information is performed by human instruction and learning, and the relationship between the information is autonomously constructed by the connection relation between the patterns and the process between the patterns. At the same time, it is possible to execute appropriate processing for the input question, command, problem / issue.
Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction.
According to the present invention, it is not necessary to design and implement a program or the like for information processing, so that development effort can be greatly reduced.

Advantage 4 of the Invention

(Corresponding to claim 4)

According to the fourth aspect of the invention, the process of analyzing and recording information and the structure of information by human instruction and learning is performed, and the relation between information is autonomously constructed by the connection relation between patterns and the process between patterns. At the same time, it is possible to execute appropriate processing for the input question, command, problem / issue. In addition, even when additions and changes occur in information processing, the program can be implemented without additions and changes.
Conventionally, when a machine performs processing of input information, etc., it has been necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction.
According to the present invention, it is not necessary to design and implement a program or the like for information processing, so that development effort can be greatly reduced.

Effect 5 of the invention

(Corresponding to claim 5)

According to the fifth invention, the value of the input information is evaluated, and the knowledge system is autonomously constructed by identifying and classifying the truth, truth, fact, definition, rule, common sense, explanation, hypothesis, prediction, opinion, and impression. can do.
Traditionally, it is difficult to evaluate the value of information input to a machine and identify and classify it into truth, truth, facts, definitions, rules, common sense, explanations, hypotheses, predictions, opinions, and impressions and autonomously build a knowledge system. Met.
In addition, when making the machine process input information, it is necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction.
According to the present invention, it is not necessary to design and implement a program or the like for information processing, so that development effort can be greatly reduced.

Effect 6 of the invention

(Corresponding to claim 6)

According to the sixth invention, the value of the input information is evaluated, and the knowledge system is autonomously constructed by identifying and classifying the truth, truth, fact, definition, rule, common sense, explanation, hypothesis, prediction, opinion, and impression. At the same time, the entered information can be checked against the constructed knowledge system, and actions in accordance with truth, truth, facts, rules and common sense can be implemented.
Traditionally, the value of information input to the machine is evaluated, and the knowledge system is autonomously constructed by identifying and classifying it into truth, truth, facts, definitions, rules, common sense, explanations, hypotheses, predictions, opinions, and impressions. It was difficult to check against the knowledge system and to carry out actions in accordance with truth, truth, facts, rules and common sense.
In addition, when making the machine process input information, it is necessary to set a program in the computer in advance. It was necessary to create a program for judging the situation from the input information and a program for causing the machine to operate according to individual conditions, and to install and execute it on a computer mounted on the machine. The program had to be created in a dedicated programming language, and there was a demerit that it took a lot of time for development. If the detection of the condition and the corresponding operation are not appropriate, there is a demerit that it is necessary for a person to correct the program installed in the computer, and it takes a long time for the correction.
According to the present invention, it is not necessary to design and implement a program or the like for information processing, so that development effort can be greatly reduced.

Configuration example of autonomous knowledge analyzer (Claim 1) Information is organized and recorded as a pattern Example of organizing by sentence analysis How patterns are recorded in the pattern recording module How the relationship between patterns is recorded in the pattern relationship recording part (part 1) How the relationship between patterns is recorded in the pattern relationship recording part (Part 2) Examples of relationships between patterns A sentence is recorded in the pattern recording part and equivalently converted Information is analyzed and equivalent transformations and connection relationships are generated Connection relationship type (logical relationship, definition, attribution relationship) Connection relationship type (definition of relationship) Connection type (development of inference, inheritance of common features and identification of individual features) Connection relationship type specification (logical relationship, definition, attribution relationship) Connection relationship type specification (relationship definition) Connection type designation (development of inference, inheritance of common features and identification of individual features) Comparison with conventional artificial intelligence technology (Comparison with expert system) Comparison with conventional artificial intelligence technology (Comparison with finite automaton) Comparison with conventional artificial intelligence technology (comparison with neural network) Comparison with conventional artificial intelligence technology (comparison with programming language) How to add / change processing (part 1) How to add / change processing (part 2) Configuration example of autonomous knowledge analyzer (Claim 3) Example of normal sentence processing Sample question processing Example statement processing Example of problem / issue processing Realization method of addition / change of processing in pattern processor (part 1) Realization method of addition / change of processing in pattern processor (part 2) Example of collation with knowledge (truth, truth, fact, definition, rule, common sense) (Part 1) Example of collation with knowledge (truth, truth, fact, definition, rule, common sense) (part 2)

1 pattern converter 2 pattern recorder 3 pattern register 4 pattern controller 5 pattern inverse converter 6 pattern analyzer 7 pattern processor

Claims (6)

  A pattern converter that converts information into a pattern, a pattern, a connection between patterns, a pattern recorder that records the relationship between patterns and the history of pattern excitation, and a human instruction or autonomy A pattern registerer that automatically registers and changes, a pattern controller that controls pattern processing, a pattern inverse converter that converts patterns into information, and a pattern analyzer that analyzes patterns and relationships between patterns. Artificial intelligence and software that performs processing to analyze and record information and the structure of information by instructions and learning, and autonomously constructs relationships between information by connection between patterns and processing between patterns.   A pattern converter that converts information into a pattern, a pattern, a connection between patterns, a pattern recorder that records the relationship between patterns and the history of pattern excitation, and a human instruction or autonomy A pattern registerer that automatically registers and changes, a pattern controller that controls pattern processing, a pattern inverse converter that converts patterns into information, and a pattern analyzer that analyzes patterns and relationships between patterns. The process of analyzing and recording information and the structure of information by instructions and learning of patterns, and even when additions and changes occur in information processing, patterns and connections between patterns without adding or changing programs Artificial intelligence and software that can respond by adding and changing A.   A pattern converter that converts information into a pattern, a pattern, a connection between patterns, a pattern recorder that records the relationship between patterns and the history of pattern excitation, and a human instruction or autonomy A pattern registerer that automatically registers and changes, a pattern controller that controls pattern processing, a pattern inverse converter that converts patterns into information, a pattern analyzer that analyzes patterns and relationships between patterns, patterns and Equipped with a pattern processor that performs appropriate processing according to the analysis results between patterns, performs processing to analyze and record information and the structure of information by human instruction and learning, and connects relationships between patterns In addition to constructing autonomously by processing between relationships and patterns, input questions, commands, problems / issues Artificial intelligence and software to perform the appropriate processing to.   A pattern converter that converts information into a pattern, a pattern, a connection between patterns, a pattern recorder that records the relationship between patterns and the history of pattern excitation, and a human instruction or autonomy A pattern registerer that automatically registers and changes, a pattern controller that controls pattern processing, a pattern inverse converter that converts patterns into information, a pattern analyzer that analyzes patterns and relationships between patterns, patterns and Equipped with a pattern processor that performs appropriate processing according to the analysis results between patterns, and performs processing to analyze and record information and the structure of information by human instruction and learning. Appropriate processing is performed on the issue, and when additions and changes occur regarding processing of information Without the addition and change the program, pattern and add a connection relation between patterns and changes artificial intelligence and software that can correspond by.   A pattern converter that converts information into a pattern, a pattern, a connection between patterns, a pattern recorder that records the relationship between patterns and the history of pattern excitation, and a human instruction or autonomy A pattern registerer that automatically registers and changes, a pattern controller that controls pattern processing, a pattern inverse converter that converts patterns into information, a pattern analyzer that analyzes patterns and relationships between patterns, patterns and Equipped with a pattern processor that performs appropriate processing according to the analysis results between patterns, evaluates the value of input information, truth, truth, fact, definition, rule, common sense, explanation, hypothesis, prediction, opinion, impression Artificial intelligence and software that identifies, classifies and autonomously builds a knowledge system.   A pattern converter that converts information into a pattern, a pattern, a connection between patterns, a pattern recorder that records the relationship between patterns and the history of pattern excitation, and a human instruction or autonomy A pattern registerer that automatically registers and changes, a pattern controller that controls pattern processing, a pattern inverse converter that converts patterns into information, a pattern analyzer that analyzes patterns and relationships between patterns, patterns and Equipped with a pattern processor that performs appropriate processing according to the analysis results between patterns, evaluates the value of input information, truth, truth, fact, definition, rule, common sense, explanation, hypothesis, prediction, opinion, impression The knowledge system is autonomously identified and classified, and the information entered is checked against the constructed knowledge system, and the truth, truth, facts Artificial intelligence and software to implement the action along the rules and common sense.