WO2022030134A1 - Information processing device, information processing method, program, and recording medium - Google Patents

Abstract

This information processing device comprises: an input unit that, on the basis of received information, generates first data representing a situation that is grasped from the information; a processing unit that, on the basis of the first data, generates second data representing a concept related to the situation; and an output unit that, on the basis of the second data, generates and outputs language representing the concept. The output unit is configured such that, when the generated language does not match a solution in response to the information, the generated language is output to the input unit as new information.

Description

Information processing equipment, information processing methods, programs and recording media

The present invention relates to an information processing apparatus, an information processing method, a program and a recording medium.

In recent years, deep learning using a multi-layer neural network has been attracting attention as a machine learning method. Deep learning uses a large amount of training data to make a multi-layer neural network learn features.

Patent Documents 1 to 3 disclose a neural network processing apparatus capable of constructing a neural network with a small amount of labor and arithmetic processing by defining a large-scale neural network as a combination of a plurality of subnetworks. ing. Further, Patent Document 4 discloses a structure optimizing device that optimizes a neural network.

Japanese Unexamined Patent Publication No. 2001-051968 Japanese Patent Application Laid-Open No. 2002-251601 Japanese Patent Application Laid-Open No. 2003-317573 Japanese Unexamined Patent Publication No. 09-091263

There is a demand for information processing devices that aim to recognize external information and obtain solutions on their own. It is being considered to apply deep learning technology to such information processing devices. However, with conventional artificial information processing technology using deep learning, it is difficult to conceptualize information and give creativity, and the realization of information processing technology that can process information with an algorithm closer to human thinking has been sought. There is.

An object of the present invention is to provide an information processing device, an information processing method, a program, and a recording medium capable of processing information with an algorithm closer to human thinking.

According to one aspect of the present invention, an input unit that generates first data representing a situation grasped from the information based on the received information, and a concept related to the situation based on the first data. It has a processing unit that generates a second data to be represented, and an output unit that generates and outputs a word expressing the concept based on the second data, and the output unit has the information that the generated word is generated. An information processing apparatus configured to output the generated words as new information to the input unit when the solution does not match the corresponding solution is provided.

Further, according to another aspect of the present invention, the first step of generating the first data representing the situation grasped from the information based on the received information, and the first step based on the first data. It has a second step of generating a second data representing a situation-related concept, and a third step of generating and outputting a word representing the concept based on the second data. When the words generated in the third step do not match the solution corresponding to the information, an information processing method is provided in which the generated words are used as new information and repeated from the first step.

Further, according to still another aspect of the present invention, the computer is based on the information received, the means for generating the first data representing the situation grasped from the information, and the first data. When a means for generating a second data representing a situation-related concept, a word representing the concept is generated based on the second data, and the generated word does not match the solution according to the information. A program is provided that functions as a means for outputting the generated words as new information to the means for generating the first data.

According to the present invention, it is possible to realize an information processing device and an information processing method capable of information processing with an algorithm closer to human thinking.

FIG. 1 is a schematic diagram showing a configuration example of an information processing apparatus according to the first embodiment of the present invention. FIG. 2 is a schematic diagram showing a configuration example of a data processing module in the information processing apparatus according to the first embodiment of the present invention. FIG. 3 is a diagram illustrating the operation of the data processing module in the information processing apparatus according to the first embodiment of the present invention. FIG. 4 is a flowchart showing an information processing method using the information processing apparatus according to the first embodiment of the present invention. FIG. 5 is a sequence diagram showing a first application example of information processing using the information processing apparatus according to the first embodiment of the present invention. FIG. 6 is a diagram showing preconditions in a second application example of information processing using the information processing apparatus according to the first embodiment of the present invention. FIG. 7 is a sequence diagram showing a second application example of information processing using the information processing apparatus according to the first embodiment of the present invention. FIG. 8 is a schematic diagram (No. 1) showing a hardware configuration example of the information processing apparatus according to the first embodiment of the present invention. FIG. 9 is a schematic diagram (No. 2) showing a hardware configuration example of the information processing apparatus according to the first embodiment of the present invention. FIG. 10 is a diagram illustrating a data processing processor in the information processing apparatus according to the first embodiment of the present invention. FIG. 11A is a diagram (No. 1) illustrating a data processing processor in the information processing apparatus according to the first embodiment of the present invention. FIG. 11B is a diagram (No. 2) illustrating a data processing processor in the information processing apparatus according to the first embodiment of the present invention. FIG. 11C is a diagram (No. 3) illustrating a data processing processor in the information processing apparatus according to the first embodiment of the present invention. FIG. 12 is a schematic diagram showing a configuration example of the information processing apparatus according to the second embodiment of the present invention.

[First Embodiment]
The information processing apparatus according to the first embodiment of the present invention will be described with reference to FIGS. 1 to 3. FIG. 1 is a schematic diagram showing a configuration example of an information processing apparatus according to the present embodiment. FIG. 2 is a schematic diagram showing a configuration example of a data processing module in the information processing apparatus according to the present embodiment. FIG. 3 is a diagram illustrating the operation of the data processing module in the information processing apparatus according to the present embodiment.

As shown in FIG. 1, the information processing apparatus 1000 according to the present embodiment may be composed of an input unit 100, a processing unit 200, and an output unit 300.

The input unit 100 has a function of generating status information data (first data) based on the received information. The situation information data is a data of the situation grasped from the received information in a predetermined format. The format of the situation information data is not particularly limited, but for example, bitmap format data or vector representation format data can be applied.

The information received by the input unit 100 includes information received from the outside of the information processing apparatus 1000 and information output by the output unit 300. The information received by the input unit 100 from the outside is not particularly limited, and examples thereof include auditory information (words, sounds, etc.), visual information (images, etc.), tactile information (sensible temperature, texture, etc.). Be done. For example, when voice input or text input is made to the input unit 100 as information from the outside, the input unit 100 can convert the input data into the situation information data by using a known word identification AI system or the like. Alternatively, when an image is input to the input unit 100 as information from the outside, the input unit 100 can convert the input data into situation information data using a known image identification AI system or the like. The information received by the input unit 100 from the output unit 300 is information related to "words" (thoughts) generated by the output unit 300.

The processing unit 200 performs predetermined conceptual processing on the status information data received from the input unit 100, and generates conceptual information data (second data) representing a concept related to the situation represented in the status information data. , It has a function to output the generated conceptual information data to the output unit 300. The processing unit 200 may be configured to include a recognition processing unit 210 and a concept processing unit 220.

The recognition processing unit 210 has a function of converting the situation information data received from the input unit 100 into conceptual information data (third data). The conceptual information data generated by the recognition processing unit 210 is data that conceptualizes the situation represented by the situation information data received from the input unit 100. The conceptual information data may be data in the same format as the situation information data received from the input unit 100. Data in the same format is data having the same data structure and size. However, the situation information data and the conceptual information data do not necessarily have to be in the same format, and may be in different formats as long as there is at least a partially overlapping portion.

The recognition processing unit 210 performs conceptual processing such as identification and classification on the status information data received from the input unit 100, and generates conceptual information data. The recognition processing unit 210 may include a plurality of types of recognition modules depending on the type of status information data. For example, the recognition processing unit 210 may include a word recognition module that processes status information data received from the word identification AI system, an image recognition module that processes status information data received from the image identification AI system, and the like. The classification performed by the recognition processing unit 210 may include, for example, classification of objects, requirements, environments, and the like shown in the situation information data.

The concept processing unit 220 performs conceptual processing on the conceptual information data received from the recognition processing unit 210, and converts the conceptual information data received from the recognition processing unit 210 into conceptual information data (second data) of another concept. It has a function to do. The conceptual processing performed on the conceptual information data is a process of grasping the situation from the received conceptual information data, thinking about the grasped situation, and deriving the optimum solution for the grasped situation.

For example, a certification support system such as a Coq system can be applied to the concept processing unit 220. If there is no answer (solution) from the concept processing unit 220 (incomplete), the output data is fed back to the input unit 100 via the output unit 300, which will be described later, so that the self-questioning self-answer is repeated and a more appropriate answer is given. It will be possible to obtain.

The output unit 300 has a function of generating "words" (thoughts) based on the solution (conceptual information data) output from the processing unit 200. The "word" is not particularly limited, but as an example, a sentence including at least a subject and a predicate expressing an idea corresponding to the concept information data output from the concept processing unit 220 can be mentioned. The "word" generated by the output unit 300 can be output as character information or voice information, and can be output to the input unit 100. The method of converting a concept into a "word" is not particularly limited. As a simple method, for example, a method of generating a sentence by inserting a word extracted from a concept into a sentence of a fixed frame can be mentioned.

The output unit 300 may further include a function of generating an "action" based on the solution (conceptual information data) output from the processing unit 200. In this case, the output unit 300 can be configured to output, for example, an instruction regarding the generated "behavior". This instruction is not particularly limited, but may include, for example, a predetermined instruction for performing a task according to the generated "behavior", for example, execution of an application.

The concept information data received by the output unit 300 may be the concept information data output by the recognition processing unit 210 or the concept information data output by the concept processing unit 220.

Next, the basic data processing modules constituting the recognition processing unit 210 and the concept processing unit 220 will be described with reference to FIG. FIG. 2 is a diagram showing a schematic configuration of a data processing module in the information processing apparatus according to the present embodiment. Although an example of application to the recognition processing unit 210 and the concept processing unit 220 is shown here, the same data processing module can also be applied to the input unit 100 and the output unit 300. Further, one functional block may include a plurality of data processing modules 400. For example, one functional block can be configured to use different data processing modules 400 depending on the input data.

As shown in FIG. 2, the data processing module 400 may include a data acquisition unit 410, a storage unit 420, an identification unit 430, and a data output unit 440.

The data acquisition unit 410 has a function of acquiring status information data and conceptual information data. In one example, the situation information data and the conceptual information data may be data in a bitmap format having a predetermined size, as described above.

The storage unit 420 stores a learning model including a plurality of models. In each of the plurality of models, a pattern representing a specific information or concept is associated with a value representing another information or concept corresponding to the pattern. For example, each model has a pattern that maps the relationship between multiple elements that represent a particular situation or concept and their element values, and multiple elements that represent different information or concepts that correspond to that particular situation or concept. It may include a value that maps the relationship to those element values.

The identification unit 430 compares the pattern of the data acquired by the data acquisition unit 410 with each pattern of the plurality of models stored in the storage unit 420, and selects the model having the most suitable pattern among the plurality of models. It has a function to select the associated value. The patterns and values possessed by each of the plurality of models may be data in the same format as the situation information data and the conceptual information data, or data in different formats, as in the relationship between the situation information data and the conceptual information data. You may. The value associated with the pattern does not necessarily have to be bitmap format data, and may be data including predetermined index information. In this case, this index information can be used to configure the data corresponding to the index information to be acquired from the outside.

The identification unit 430 may have a learning function of a learning model. For example, when there is no model in the learning model whose degree of conformity is equal to or higher than a predetermined threshold, a model corresponding to the data acquired by the data acquisition unit 410 is added to the learning model, and the learning model is updated. can do. For the construction and update of the learning model, for example, the technique described in Japanese Patent Application No. 2019-144121 by the same applicant can be applied.

The data output unit 440 has a function of outputting the value (data in bitmap format) selected by the identification unit 430 as output data. The data output unit 440 typically outputs the selected value to the next-stage processing unit, but returns the selected value to the data acquisition unit 410 and performs recursive processing in the data processing module 400. It may be configured to do so.

Next, the data processing in the data processing module 400 will be specifically described with reference to FIG. FIG. 3 is a schematic diagram showing the operation of the data processing module 400 in the information processing apparatus according to the present embodiment.

First, the data acquisition unit 410 acquires status information data from the input unit 100. Alternatively, the data acquisition unit 410 acquires conceptual information data from the processing unit 200. The data acquisition unit 410 outputs the acquired data (input data) to the identification unit 430.

The identification unit 430 compares the pattern of the data received from the data acquisition unit 410 with each pattern of the plurality of models of the learning model stored in the storage unit 420. Then, the identification unit 430 extracts a model having the pattern having the highest degree of conformity to the pattern of the data received from the data acquisition unit 410 from the plurality of models included in the learning model. It can be said that a model having a pattern having a high degree of conformity with respect to the pattern of the received data is a model in which the information distance is close to the received data.

The method of determining the goodness of fit between the input data (situation information data or conceptual information data) and the training model is not particularly limited, but for example, an internal product value of the input data pattern and the training model pattern is used. The method can be mentioned.

The method of determining the goodness of fit between the input data and the learning model using the inner product value of the input data pattern and the learning model pattern will be described below. Here, for the sake of brevity, each of the situation information data, the conceptual information data, the pattern of the learning model, and the value of the training data is in a bit map format including nine cells arranged in a 3 × 3 matrix. It shall be a pattern. The value of each cell is 0 or 1.

The internal product value of the input data pattern and the learning model pattern is calculated by multiplying the values of cells with the same coordinates and adding up the multiplication values of each coordinate. For example, as shown in FIG. 3, the values of the cells constituting the input data pattern are A, B, C, D, E, F, G, H, and I, and the pattern of the learning model to be compared is configured. It is assumed that the value of each cell is 1,0,0,0,1,0,0,0,1. In this case, the internal product value of the input data pattern and the learning model pattern is A × 1 + B × 0 + C × 0 + D × 0 + E × 1 + F × 0 + G × 0 + H × 0 + I × 1. The inner product value calculated in this way is normalized by dividing by the number of cells having a value of 1 among the cells included in the input data. The calculation and normalization of the inner product value for the input data is performed for each of the plurality of models included in the training model.

Next, the model with the maximum normalized internal product value is extracted from the multiple models of the learning model. The normalized internal product value represents the likelihood, and the larger the value, the higher the goodness of fit to the input data. Therefore, by extracting the model with the maximum normalized internal product value, the model with the highest goodness of fit to the input data can be extracted. Here, for example, as shown in FIG. 3, a model having a pattern in which the value of each cell is 1,0,0,0,1,0,0,0,1 has the highest goodness of fit to the input data. It is assumed that it is extracted as.

Next, the identification unit 430 outputs the value associated with the extracted model to the data output unit 440 as another conceptual information data converted from the input data. For example, as shown in FIG. 3, the extracted model has a pattern in which the value of each cell is 1,0,0,0,1,0,0,0,1 and the value of each cell is 1,1. , 0,0,1,0,0,1,1 and the value. In this case, the value whose value in each cell is 1,1,0,0,1,0,0,1,1 becomes the output data to be output to the data output unit 440.

Next, the data output unit 440 outputs the data acquired from the identification unit 430 to the next-stage processing unit.

With this configuration, the input data in which the value of each cell is A, B, C, D, E, F, G, H, I can be input, and the value of each cell is 1,1,0,0,1. It can be converted into output data of, 0, 0, 1, 1.

By configuring the recognition processing unit 210 using such a data processing module 400, it is possible to convert the situation information data received from the input unit 100 into conceptual information data. Further, by configuring the concept processing unit 220 using such a data processing module 400, the concept information data received from the recognition processing unit 210 can be converted into the concept information data of another concept.

Next, the information processing method using the information processing apparatus 1000 according to the present embodiment will be described with reference to FIG. FIG. 4 is a flowchart showing an information processing method according to the present embodiment.

First, the input unit 100 acquires information from the outside (step S101). The information acquired by the input unit 100 is not particularly limited, and examples thereof include auditory information (words, sounds, etc.), visual information (images, etc.), and tactile information (sensible, temperature, texture, etc.). ..

The input unit 100 converts the information received from the outside into data in a predetermined format and generates status information data (step S102). For example, when voice input or text input is made to the input unit 100 as information from the outside, the input unit 100 converts the input data into the situation information data by using a known word identification AI system or the like. Alternatively, when an image is input to the input unit 100 as information from the outside, the input unit 100 converts the input data into situation information data using a known image identification AI system or the like.

Next, the input unit 100 outputs the generated status information data to the recognition processing unit 210 of the processing unit 200.

Next, the recognition processing unit 210 uses, for example, the above-mentioned data processing module 400 to convert the situation information data received from the input unit into the concept information data and output it to the concept processing unit 220 (step S103).

Next, the concept processing unit 220 uses, for example, the above-mentioned data processing module 400 to convert the concept information data received from the recognition processing unit 210 into concept information data representing a concept different from the received concept information data (. Step S104).

Next, the concept processing unit 220 outputs the generated concept information data to the output unit 300.

Next, the output unit 300 generates "words" and "actions" based on the conceptual information data received from the processing unit 200 (step S105).

Next, the output unit 300 determines whether or not the generated "word" is sufficient as an answer (solution) (step S106).

As a result of the determination, if the generated "word" is insufficient (incomplete) as a solution (“No” in step S106), the process proceeds to step S102, and the generated “word” is fed back to the input unit 100. Then, the process of step S102 to step S106 is repeated. That is, repeat the self-questioning and self-answer using the generated "words". The "word" generated by the output unit 300 represents, so to speak, a result based on past experience. Output unit 300 By processing the generated "words" again, it is possible to make a judgment based on past experience.

On the other hand, when the generated "word" is sufficient as a solution ("Yes" in step S106), the output unit 300 outputs the generated "word" and / or "action" and ends a series of processing. (Step S107). The output of "words" is not particularly limited, but for example, the generated "words" can be output as character information or voice information. The output of the "behavior" is not particularly limited, but for example, a predetermined instruction for executing the task corresponding to the generated "behavior" can be output.

As described above, all the processing in the information processing apparatus according to the present embodiment is data-to-data conversion, that is, data-driven processing.

One of the hypotheses that explain human thinking is that the generation of words is thinking. This hypothesis explains that conscious activity is performed by repeating self-questioning and self-answering with the generated words as the mental language. It can be said that feeding back the "words" generated by the output unit 300 to the input unit 100 corresponds to the self-questioning self-answer in this hypothesis. This enables information processing by an algorithm that is closer to human thinking.

Next, FIGS. 5 to 7 will be described with respect to an application example of information processing using the information processing apparatus 1000 according to the present embodiment. Here, two application examples will be given to more specifically explain the information processing in the present embodiment.

The first application example is an application example to a license issuing business process for the purpose of acquiring a license number in response to a license request from a user and reporting the acquired license number to the user. It is assumed that the information processing apparatus 1000 has knowledge of acquiring a license number by operating the numbering ledger.

FIG. 5 is a sequence diagram showing an information processing method in the first application example.

It is assumed that the user (Mr. A) has input to the input unit 100 an instruction requesting the issuance of a license. The input unit 100 converts the input information into status information data and outputs it to the recognition processing unit 210.

Next, the recognition processing unit 210 converts the status information data received from the input unit 100 into conceptual information data. Here, it is assumed that the input information is converted into conceptual information data including the concept (concept A) indicating the situation of "Mr. A", "license", and "issue". The recognition processing unit 210 outputs the converted conceptual information data to the output unit 300.

Next, the output unit 300 generates "words" based on the conceptual information data received from the recognition processing unit 210. For example, for the concept A indicating the situation of "Mr. A", "license", and "issuance", a word such as "think" is added to generate a "word". As a result, for example, the "word" "Consider issuing Mr. A's license" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, it is assumed that the data is converted into conceptual information data including the concept (concept B) indicating the problems of "Mr. A", "license", and "issuance" according to the word "thinking". The recognition processing unit 210 outputs the converted concept information data to the concept processing unit 220.

Next, the concept processing unit 220 converts the input concept information data into another concept information data. Here, it is assumed that the data is converted into conceptual information data including a concept (concept C) indicating an action recalled from the problems of "Mr. A", "license", and "issuance". For example, the method of "numbering ledger operation" is recalled from the information of "license" and "issuance", and the input conceptual information data is used as the concept of "Mr. A", "license", and "numbering ledger operation". Convert to include conceptual information data. The concept processing unit 220 outputs the converted concept information data to the output unit 300.

Next, the output unit 300 generates "words" based on the concept information data received from the concept processing unit 220. For example, a word such as "do" is added to the concept C indicating the behavior of "Mr. A", "license", and "numbering ledger operation" to generate "words". As a result, for example, the "word" of "operate the license numbering ledger of Mr. A" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, it is assumed that the data is converted into conceptual information data including the concept (concept D) indicating the work for "Mr. A", "license", and "numbering ledger operation" according to the word "do". The recognition processing unit 210 outputs the converted conceptual information data to the output unit 300.

Next, the output unit 300 generates an "action" based on the conceptual information data received from the recognition processing unit 210. Here, as work for the information "license" and "numbering ledger operation", for example, a schedule application for issuing a license number based on the information "license" and "numbering ledger operation" shown in the conceptual information data. Output instructions to execute. The schedule application outputs the license number obtained by execution to the input unit 100.

Next, the input unit 100 converts the license number received from the schedule application as it is or into status information data, and outputs it to the recognition processing unit 210. The recognition processing unit 210 identifies that the received information is the result of the execution of the schedule application, and outputs the received information to the output unit 300.

Next, the output unit 300 generates "words" based on the information received from the recognition processing unit 210. For example, a word such as "desu" is added to the license number "123-456-7890" output by the schedule application to generate a word. As a result, for example, the "word" "123-456-7890." Is generated.

Next, the output unit 300 reports the generated "words" to the user by displaying them on a display device, outputting them by voice, or the like, and ends a series of processes.

The second application example is a schedule adjustment process for the purpose of comparing the schedules of two users (Mr. A and Mr. B) for 5 days from Monday to Friday, adjusting the free time of both users, and reporting. It is an application example to. Here, it is assumed that the schedules of Mr. A and Mr. B from Monday to Friday are as shown in FIG. In the table of FIG. 6, it is assumed that the "am" and "afternoon" time zones are working hours, and the "lunch break" is outside working hours. Further, "0" indicates a free time zone without reservation, and "1", "2", and "3" indicate a time zone with reservation. The values of "1", "2", and "3" indicate the priority of the reservation, "1" is the reservation that requires attendance, "2" is the reservation that attendance is preferable, and "3" is the reservation that can be absent. , Suppose. The information processing apparatus 1000 is provided with knowledge of adjusting the schedule during working hours, but is not provided with knowledge of "lunch break" and "priority" in the initial state.

FIG. 7 is a sequence diagram showing an information processing method in the second application example.

It is assumed that the user (Mr. A) has input to the input unit 100 an instruction requesting schedule adjustment with Mr. B. The input unit 100 converts the input information into status information data and outputs it to the recognition processing unit 210.

Next, the recognition processing unit 210 converts the status information data received from the input unit 100 into conceptual information data. Here, it is assumed that the input information is converted into conceptual information data including the concept (concept A) indicating the situation of "Mr. A", "Mr. B", and "schedule adjustment". The recognition processing unit 210 outputs the converted conceptual information data to the output unit 300.

Next, the output unit 300 generates "words" based on the conceptual information data received from the recognition processing unit 210. For example, a word such as "think" is added to the concept A indicating the situation of "Mr. A", "Mr. B", and "schedule adjustment" to generate "words". As a result, for example, the "word" "Consider adjusting the schedule of Mr. A and Mr. B." is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, it is assumed that the data is converted into conceptual information data including the concept (concept B) indicating the tasks of "Mr. A", "Mr. B", and "schedule adjustment" according to the word "thinking". The recognition processing unit 210 outputs the converted concept information data to the concept processing unit 220.

Next, the concept processing unit 220 converts the input concept information data into another concept information data. Here, it is assumed that the data is converted into conceptual information data including a concept (concept C) indicating an action recalled from the tasks of "Mr. A", "Mr. B", and "schedule adjustment". For example, the method of "working" and "searching with the schedule app" is recalled from the information "schedule adjustment", and the input conceptual information data is searched for "Mr. A", "Mr. B", "working", and "schedule app". It is converted into conceptual information data including the concept showing the behavior. The concept processing unit 220 outputs the converted concept information data to the output unit 300.

Next, the output unit 300 generates "words" based on the concept information data received from the concept processing unit 220. For example, for the concept C indicating the behaviors of "Mr. A", "Mr. B", "working", and "searching with the schedule application", words such as "do" are added to generate "words". As a result, for example, the "word" "Search with Mr. A and Mr. B's working schedule application" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, it is converted into conceptual information data including the concept (concept D) indicating the work for "Mr. A", "Mr. B", "working", and "searching with the schedule application" according to the word "do". And. The recognition processing unit 210 outputs the converted conceptual information data to the output unit 300.

Next, the output unit 300 generates an "action" based on the conceptual information data received from the recognition processing unit 210. Here, as work on the information shown in the conceptual information data, an instruction to execute a schedule application for adjusting the schedule during work of Mr. A and Mr. B is output. The schedule application outputs the result obtained by execution to the input unit 100. In the example of FIG. 6, since there is no common free time in the working schedules of Mr. A and Mr. B, it is assumed that the result of "no free time" is output.

Next, the input unit 100 converts the result received from the schedule application as it is or into status information data, and outputs it to the recognition processing unit 210. The recognition processing unit 210 identifies that the received information is the result of the execution of the schedule application, and outputs the received information to the output unit 300.

Next, the output unit 300 generates "words" based on the information received from the recognition processing unit 210. For example, as a result of output by the schedule application, a word such as "is" is added to "no space" to generate a word. As a result, for example, the "word" "There is no space" is generated.

Next, the output unit 300 reports the generated "words" to the user by displaying them on a display device, outputting them by voice, or the like, and ends a series of processes.

Next, it is assumed that the user (Mr. A) who received the report has input an instruction to the input unit 100 to request the schedule adjustment of the lunch break with Mr. B. The input unit 100 converts the input information into status information data and outputs it to the recognition processing unit 210.

Next, the recognition processing unit 210 converts the status information data received from the input unit 100 into conceptual information data. Here, it is assumed that the input information is converted into conceptual information data including the concept (concept A) indicating the situation of "Mr. A", "Mr. B", "lunch break", and "schedule adjustment". The recognition processing unit 210 outputs the converted conceptual information data to the output unit 300.

Next, the output unit 300 generates "words" based on the conceptual information data received from the recognition processing unit 210. For example, for the concept A indicating the situation of "Mr. A", "Mr. B", "lunch break", and "schedule adjustment", words such as "think" are added to generate "words". As a result, for example, the "word" "Mr. A and Mr. B consider adjusting the lunch break schedule" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, it is assumed that the data is converted into conceptual information data including the concept (concept B) indicating the tasks of "Mr. A", "Mr. B", "lunch break", and "schedule adjustment" according to the word "thinking". The recognition processing unit 210 outputs the converted concept information data to the concept processing unit 220.

Next, the concept processing unit 220 converts the input concept information data into another concept information data. Here, it is assumed that the data is converted into conceptual information data including a concept (concept C) indicating an action recalled from the tasks of "Mr. A", "Mr. B", "lunch break", and "schedule adjustment". At this stage, the information processing apparatus 1000 does not have "lunch break" as knowledge, but the information "schedule adjustment" recalls the methods of "lunch break" and "search with the schedule application". Then, the input conceptual information data is converted into conceptual information data including concepts indicating actions such as "Mr. A", "Mr. B", "lunch break", and "search with the schedule application". The concept processing unit 220 outputs the converted concept information data to the output unit 300.

Next, the output unit 300 generates "words" based on the concept information data received from the concept processing unit 220. For example, to the concept C indicating the actions of "Mr. A", "Mr. B", "lunch break", and "search with the schedule application", words such as "do" are added to generate "words". As a result, for example, the "word" "Search with Mr. A and Mr. B's lunch break schedule application" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, it is converted into conceptual information data including the concept (concept D) indicating the work for "Mr. A", "Mr. B", "lunch break", and "search with the schedule application" according to the word "do". do. The recognition processing unit 210 outputs the converted conceptual information data to the output unit 300.

Next, the output unit 300 generates an "action" based on the conceptual information data received from the recognition processing unit 210. Here, as work on the information shown in the conceptual information data, an instruction to execute a schedule application for adjusting the schedule of lunch breaks of Mr. A and Mr. B is output. The schedule application outputs the result obtained by execution to the input unit 100.

Next, the input unit 100 converts the result received from the schedule application as it is or into status information data, and outputs it to the recognition processing unit 210. The recognition processing unit 210 identifies that the received information is the result of execution of the schedule application, and outputs the received information to the output unit 300. In the example of FIG. 6, since the common free time in the lunch break schedule of Mr. A and Mr. B exists on Tuesday, Thursday, and Friday, it is assumed that the result of "Tuesday, Thursday, Friday lunch break" is output. ..

Next, the input unit 100 converts the result received from the schedule application as it is or into status information data, and outputs it to the recognition processing unit 210. The recognition processing unit 210 identifies that the received information is the result of the execution of the schedule application, and outputs the received information to the output unit 300.

Next, the output unit 300 generates "words" based on the information received from the recognition processing unit 210. For example, as a result of the schedule application output, a word such as "desu" is added to "Tue, Thu, Fri lunch break" to generate a word. This produces, for example, the "word" that says, "Tuesday, Thursday, and Friday lunch break."

Next, the output unit 300 reports the generated "words" to the user by displaying them on a display device or outputting them by voice.

Further, the output unit 300 generates "words" based on the concept information data received from the recognition processing unit 210 in parallel with the generation of the "action" for the concept D. For example, to the concept D indicating the work of "Mr. A", "Mr. B", "lunch break", and "search with the schedule application", words such as "done" are added to generate "words". As a result, for example, a "word" such as "Mr. A and Mr. B searched with the lunch break schedule application" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, we refer to the result of "action" according to the word "done", and "remember the lunch break" according to the result obtained even though we do not have "lunch break" as knowledge. Generate a concept (concept E) that indicates learning.

Next, the concept processing unit 220 learns the learning model stored in the storage unit 420 of the data processing module 400, and makes the "lunch break" memorized as knowledge. In this way, a series of processes is completed.

Next, it is assumed that the user (Mr. A) who received the report further inputs an instruction requesting the input unit 100 to adjust the schedule of priority 3 with Mr. B. The input unit 100 converts the input information into status information data and outputs it to the recognition processing unit 210.

Next, the recognition processing unit 210 converts the status information data received from the input unit 100 into conceptual information data. Here, it is assumed that the input information is converted into conceptual information data including the concept (concept A) indicating the situation of "Mr. A", "Mr. B", "Priority 3", and "Schedule adjustment". The recognition processing unit 210 outputs the converted conceptual information data to the output unit 300.

Next, the output unit 300 generates "words" based on the conceptual information data received from the recognition processing unit 210. For example, for the concept A indicating the situation of "Mr. A", "Mr. B", "Priority 3", and "Schedule adjustment", words such as "think" are added to generate "words". As a result, for example, the "word" "Mr. A and Mr. B consider the priority 3 schedule adjustment" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, it is converted into conceptual information data including the concept (concept B) indicating the tasks of "Mr. A", "Mr. B", "Priority 3", and "Schedule adjustment" according to the word "thinking". do. The recognition processing unit 210 outputs the converted concept information data to the concept processing unit 220.

Next, the concept processing unit 220 converts the input concept information data into another concept information data. Here, it is assumed that the data is converted into conceptual information data including a concept (concept C) indicating an action recalled from the tasks of "Mr. A", "Mr. B", "Priority 3", and "Schedule adjustment". At this stage, the information processing apparatus 1000 does not have "priority 3" as knowledge, but the information "schedule adjustment" recalls the methods "priority 3" and "search by schedule application". Then, the input conceptual information data is converted into conceptual information data including concepts indicating actions such as "Mr. A", "Mr. B", "Priority 3", and "Search by schedule application". The concept processing unit 220 outputs the converted concept information data to the output unit 300.

Next, the output unit 300 generates "words" based on the concept information data received from the concept processing unit 220. For example, for the concept C indicating the actions of "Mr. A", "Mr. B", "Priority 3", and "Search by schedule application", words such as "do" are added to generate "words". As a result, for example, the "word" "Search with Mr. A and Mr. B priority 3 schedule application" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, it was converted into conceptual information data including the concept (concept D) indicating the work for "Mr. A", "Mr. B", "Priority 3", and "Search with the schedule application" according to the word "do". It shall be. The recognition processing unit 210 outputs the converted conceptual information data to the output unit 300.

Next, the output unit 300 generates an "action" based on the conceptual information data received from the recognition processing unit 210. Here, as work on the information shown in the conceptual information data, an instruction to execute a schedule application for adjusting the schedule of priority 3 of Mr. A and Mr. B is output. The schedule application outputs the result obtained by execution to the input unit 100.

Next, the input unit 100 converts the result received from the schedule application as it is or into status information data, and outputs it to the recognition processing unit 210. The recognition processing unit 210 identifies that the received information is the result of the execution of the schedule application, and outputs the received information to the output unit 300. In the example of FIG. 6, since the free time or the time zone of priority 3 common to the schedules of Mr. A and Mr. B exists on Monday afternoon and Friday morning, "Monday afternoon, Friday morning priority 3". It is assumed that the result is output.

Next, the input unit 100 converts the result received from the schedule application as it is or into status information data, and outputs it to the recognition processing unit 210. The recognition processing unit 210 identifies that the received information is the result of the execution of the schedule application, and outputs the received information to the output unit 300.

Next, the output unit 300 generates "words" based on the information received from the recognition processing unit 210. For example, as a result output by the schedule application, a word such as "is" is added to "priority 3 on Monday afternoon and Friday morning" to generate a word. As a result, for example, the "word" "Monday afternoon, Friday morning has priority 3" is generated.

Next, the output unit 300 reports the generated "words" to the user by displaying them on a display device or outputting them by voice.

Further, the output unit generates "words" based on the concept information data received from the recognition processing unit 210 in parallel with the generation of the "action" for the concept D. For example, to the concept D indicating the work of "Mr. A", "Mr. B", "Priority 3", and "Search by schedule application", words such as "done" are added to generate "words". As a result, for example, a "word" such as "Mr. A and Mr. B searched with the priority 3 schedule application" is generated. The output unit 300 outputs the generated "word" to the input unit 100.

Next, the input unit 100 converts the input "word" into status information data and outputs it to the recognition processing unit 210. The recognition processing unit 210 converts the situation information data received from the input unit 100 into conceptual information data. Here, the result of "action" is referred to according to the word "done", and "priority 3" is obtained according to the result obtained even though "priority 3" is not provided as knowledge. Generate a concept (concept E) that shows the learning of "remembering".

Next, the concept processing unit 220 learns the learning model stored in the storage unit 420 of the data processing module 400, and makes the "priority 3" memorized as knowledge. In this way, a series of processes is completed.

After learning "Lunch break" and "Priority 3", in response to the instruction from the user (Mr. A) to request schedule adjustment with Mr. B, "There is no space. Lunch break on Tuesday, Thursday, and Friday. It is also possible to get reports such as "It is priority 3 on Monday afternoon and Friday morning."

Next, a hardware configuration example of the information processing apparatus 1000 according to the present embodiment will be described with reference to FIGS. 8 to 11C. 8 and 9 are schematic views showing a hardware configuration example of the information processing apparatus according to the present embodiment. FIG. 10 is a diagram illustrating a data processing processor in the information processing apparatus according to the present embodiment. 11A, 11B and 11C are diagrams showing pattern examples of input data and a learning model.

The information processing device 1000 can be realized by a hardware configuration similar to that of a general information processing device, as shown in FIG. 8, for example. For example, the information processing apparatus 1000 may include a CPU (Central Processing Unit) 500, a main storage unit 502, a communication unit 504, and an input / output interface unit 506.

The CPU 500 is a control / arithmetic unit that controls the overall control and arithmetic processing of the information processing apparatus 1000. The main storage unit 502 is a storage unit used for a data work area or a data temporary save area, and may be configured by a memory such as a RAM (Random Access Memory). The communication unit 504 is an interface for transmitting and receiving data via a network. The input / output interface unit 506 is an interface for connecting to an external output device 510, an input device 512, a storage device 514, and the like to transmit / receive data. The CPU 500, the main storage unit 502, the communication unit 504, and the input / output interface unit 506 are connected to each other by the system bus 508. The storage device 514 may be configured by, for example, a hard disk device composed of a non-volatile memory such as a ROM (Read Only Memory), a magnetic disk, or a semiconductor memory.

The main storage unit 502 can be used as a work area for executing an operation in the data processing module 400 or the like. The CPU 500 can function as a control unit that controls arithmetic processing in the main storage unit 502. The storage device 514 can be used as a storage unit 420 and can store a trained learning model.

The communication unit 504 is a communication interface based on standards such as Ethernet (registered trademark) and Wi-Fi (registered trademark), and is a module for communicating with other devices. The learning model stored in the storage device 514 may be configured to receive from another device via the communication unit 504. For example, a frequently used learning model can be stored in the storage device 514, and infrequently used learning cell information can be configured to be read from another device.

The output device 510 may include a display such as a liquid crystal display device. The output device 510 can be used as a display device for presenting the processing result to the user. The input device 512 is a keyboard, a mouse, a touch panel, or the like, and can be used for a user to input a predetermined instruction to the information processing device 1000.

The functions of each part of the information processing apparatus 1000 according to the present embodiment can be realized in terms of hardware by mounting circuit components that are hardware components such as LSI (Large Scale Integration) in which a program is incorporated. Alternatively, it can be realized by software by storing the program providing the function in the storage device 514, loading the program in the main storage unit 502, and executing the program in the CPU 500.

Further, the configuration of the information processing device 1000 shown in FIG. 1 does not necessarily have to be configured as one independent device. For example, a part of the input unit 100, the processing unit 200, and the output unit 300, for example, the processing unit 200 may be arranged on the cloud, and an information processing system may be constructed by these.

Alternatively, the information processing apparatus 1000 can be configured as a data-driven data flow machine, for example, as shown in FIG.

For example, the information processing apparatus 1000 may include a plurality of data processing processors 600 and an input / output interface unit 620. The plurality of data processing processors 600 are connected in series. The first-stage data processing processor 600 of the series connection of the data processing processor 600 is connected to the input / output interface unit 620 and receives data from the input / output interface unit 620. Further, the data processing processor 600 at the final stage of the series connection body of the data processing processor 600 is connected to the input / output interface unit 620 and outputs data to the input / output interface unit 620. The data processing processor 600 has a function of outputting a value associated with a pattern at the information distance closest to the input. The input / output interface unit 620 is an interface for connecting to an external output device 630, input device 640, storage device 650, or the like to transmit / receive data.

Note that FIG. 9 shows one route from the input / output interface unit 620 to the input / output interface unit 620 via the plurality of data processing processors 600, but a plurality of routes may be provided in parallel. Further, a branching route or a merging route may be provided for a plurality of routes.

As shown in FIG. 10, for example, each of the plurality of data processing processors 600 includes an input processing unit 602, a plurality of (for example, m) inner stackers 604 ₁ to 604 _m , a comparator 606, a selector 608, and an output. It may be configured to include a processing unit 610. Each of the plurality of data processing processors 600 has a function of receiving data, performing predetermined processing on input data, and outputting the processed data.

The inner stacker 604 ₁ to 604 _m , the comparator 606 and the selector 608 may be configured by a logic gate circuit or the like. The patterns (PAT1 to PATm) and values (Value1 to Valuem) of the training data can be stored in registers. The patterns (PAT1 to PATm) and values (Value1 to Valuem) of the learning data can be stored in the storage device 650 in advance. In this case, when the information processing device 1000 is started, it can be read from the storage device 650 through the input / output interface unit 620 and set in each data processing processor 600.

The data output from the input / output interface unit 620 is input to the input processing unit 602 of the first stage data processing processor 600. The data input to the input processing unit 602 is the above-mentioned situation information data and conceptual information data. The input processing unit 602 outputs input data in parallel to each of the plurality of inner stackers 604 corresponding to the plurality of models constituting the learning model. For example, when the training model includes m models, the input data is input in parallel to the m inner stackers 604 ₁ to 604 _m .

Each of the inner product devices 604 ₁ to 604 _m performs an inner product calculation of the pattern of the input data and the pattern of the learning model. For example, it is assumed that the pattern of the input data is composed of the data as shown in FIG. 11A, and the pattern of the learning model corresponding to the inner stacker ₆₀₄₁ is composed of the data as shown in FIG. 11B. Further, it is assumed that the value as shown in FIG. 11C is associated with the pattern of the learning model of FIG. 11B. In this case, the inner product device 604 ₁
IN ₁ x PAT1 ₁ + IN ₂ x PAT1 ₂ + ... + IN ₉ x PAT1 ₉
The sum of the multiplications for each element is calculated as in. The calculation results of the inner stackers 604 ₁ to 604 _m are input to the comparator 606.

In this way, by performing the inner product calculation of the input data for the patterns of a plurality of learning models, it is possible to complete all the inner product calculations in one cycle, and the number of cycles of one processing is compared with the processing by the program. Can be significantly reduced. If the amount of resources of the processor is insufficient, the inner product calculation process may be performed in a plurality of times while exchanging the patterns of the learning data supplied to each inner product unit 604.

The comparator 606 compares the output values of the inner product units 604 ₁ to 604 _m , and outputs the learning model number (1 to m) indicating the largest inner product value with respect to the input data to the selector 608.

The selector 608 selects the value associated with the learning model corresponding to the number output from the comparator 606 from the values associated with the patterns of the plurality of learning models, and outputs the value to the output processing unit 610. do. The output processing unit 610 outputs the selected value to the next-stage data processing processor 600.

When shifting to the processing of the next-stage data processing processor 600 after passing the processing in the data processing processor 600, data may be transmitted from the input processing unit 602 to the output processing unit 610. Further, when performing processing such as a state transition, the processing result by the data processing processor 600 may be returned from the output processing unit 610 to the input processing unit 602.

As described above, according to the present embodiment, it is possible to realize an information processing device and an information processing method capable of information processing with an algorithm closer to human thinking.

[Second Embodiment]
The information processing apparatus according to the second embodiment of the present invention will be described with reference to FIG. The same components as those of the information processing apparatus according to the first embodiment are designated by the same reference numerals, and the description thereof will be omitted or simplified. FIG. 12 is a schematic diagram showing a schematic configuration of an information processing apparatus according to the present embodiment.

As shown in FIG. 12, the information processing apparatus 1000 according to the present embodiment has an input unit 100, a processing unit 200, and an output unit 300.

The input unit 100 has a function of generating first data representing a situation grasped from the information based on the received information. The processing unit 200 has a function of generating second data representing a situation-related concept based on the first data received from the input unit 100. The output unit 300 has a function of generating a word expressing the concept based on the second data received from the processing unit 200. Further, the output unit 300 is configured to output the generated words as new information to the input unit 100 when the generated words do not match the solution corresponding to the information.

With this configuration, it is possible to realize an information processing device that can process information with an algorithm that is closer to human thinking.

[Modification Embodiment]
The present invention is not limited to the above embodiment and can be modified in various ways.

For example, an example in which a partial configuration of any of the embodiments is added to another embodiment or an example in which a partial configuration of another embodiment is replaced with another embodiment is also an embodiment of the present invention.

Further, there are also processing methods in which a program for operating the configuration of the embodiment is recorded on a recording medium so as to realize the function of the above-described embodiment, the program recorded on the recording medium is read out as a code, and the program is executed by a computer. Included in the category of embodiments. That is, a computer-readable recording medium is also included in the scope of each embodiment. Further, not only the recording medium on which the above-mentioned program is recorded but also the program itself is included in each embodiment.

As the recording medium, for example, a floppy (registered trademark) disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a magnetic tape, a non-volatile memory card, or a ROM can be used. Further, not only the program recorded on the recording medium that executes the process alone, but also the program that operates on the OS and executes the process in cooperation with other software and the function of the expansion board is also an embodiment. Is included in the category of.

The above embodiments are merely examples of embodiment of the present invention, and the technical scope of the present invention should not be construed in a limited manner by these. That is, the present invention can be implemented in various forms without departing from the technical idea or its main features.

A part or all of the above embodiment may be described as in the following appendix, but is not limited to the following.

(Appendix 1)
Based on the received information, an input unit that generates first data representing the situation grasped from the information, and an input unit.
A processing unit that generates a second data representing a concept related to the situation based on the first data, and a processing unit.
It has an output unit that generates and outputs a word expressing the concept based on the second data.
The information processing unit is configured to output the generated words as new information to the input unit when the generated words do not match the solution corresponding to the information. Device.

(Appendix 2)
The information processing apparatus according to Appendix 1, wherein the output unit is configured to further output an instruction regarding an action according to the concept represented by the second data.

(Appendix 3)
The processing unit has a first processing unit that generates a third data conceptualizing the situation represented by the first data, and the second data by converting the third data into another concept. The information processing apparatus according to Appendix 1 or 2, characterized in that it has a second processing unit to be generated.

(Appendix 4)
The first processing unit includes a first learning model and a first identification unit.
The first data is data that maps the relationship between a plurality of elements representing the situation and their element values.
The first learning model has a pattern that maps the relationship between a plurality of elements representing a specific situation and their element values, and a plurality of elements associated with the pattern and representing a concept corresponding to the specific situation. Contains multiple models, each containing a value that maps the relationship between and their element values,
The first identification unit transfers the value associated with the pattern having the highest goodness of fit to the first data among the plurality of models of the first learning model. The information processing apparatus according to Appendix 3, characterized in that it is selected as data.

(Appendix 5)
The second processing unit includes a second learning model and a second identification unit.
The third data is data that maps the relationship between a plurality of elements representing the concept and their element values.
The second learning model represents a pattern that maps the relationship between a plurality of elements representing a specific concept and their element values, and another concept associated with the pattern and assumed from the specific concept. Contains multiple models, each containing a value that maps the relationship between multiple elements and their element values.
The second identification unit transfers the value associated with the pattern having the highest goodness of fit to the third data among the plurality of models of the second learning model. The information processing apparatus according to Appendix 3 or 4, characterized in that it is selected as data.

(Appendix 6)
The second processing unit corresponds to the at least a part of the elements when the model corresponding to at least a part of the concept represented by the third data is not included in the second learning model. The information processing apparatus according to Appendix 5, wherein the model is added as one of the models constituting the second learning model.

(Appendix 7)
The processing unit is composed of a plurality of data processing processors connected in series.
5. The item according to any one of Supplementary note 1 to 6, wherein each of the plurality of data processing processors is configured to output a value associated with a pattern at the information distance closest to the input. Information processing device.

(Appendix 8)
Based on the received information, the first step of generating the first data representing the situation grasped from the information, and
Based on the first data, a second step of generating second data representing the concept related to the situation, and
It has a third step of generating and outputting a word expressing the concept based on the second data.
An information processing method characterized by repeating the generated words as new information from the first step when the words generated in the third step do not match the solution corresponding to the information.

(Appendix 9)
The information processing method according to Appendix 8, wherein in the third step, an instruction regarding an action according to the concept represented by the second data is further output.

(Appendix 10)
The second step is
A step of generating a third data that conceptualizes the situation represented by the first data, and
The information processing method according to Supplementary note 8 or 9, further comprising a step of converting the third data into another concept to generate the second data.

(Appendix 11)
The first data is data that maps the relationship between a plurality of elements representing the situation and their element values.
The first learning model includes a pattern that maps the relationship between a plurality of elements representing a specific situation and their element values, and a plurality of elements associated with the pattern and representing a concept corresponding to the specific situation. Includes multiple models, each containing a value that maps the relationship to those element values,
In the step of generating the third data, the value associated with the pattern having the highest goodness of fit to the first data among the plurality of models of the first learning model is referred to. The information processing method according to Appendix 10, wherein the data is selected as the third data.

(Appendix 12)
The third data is data that maps the relationship between a plurality of elements representing the concept and their element values.
The second learning model is a pattern that maps the relationship between a plurality of elements representing a specific concept and their element values, and a plurality of patterns that are associated with the pattern and represent another concept that is assumed from the specific concept. Contains multiple models, each containing a value that maps the relationship between the elements of and their element values.
In the step of generating the second data, the value associated with the pattern having the highest goodness of fit to the third data among the plurality of models of the second learning model is referred to. The information processing method according to Appendix 10 or 11, characterized in that it is selected as the second data.

(Appendix 13)
When the model corresponding to at least a part of the elements of the concept represented by the third data is not included in the second learning model, the model corresponding to the at least a part of the elements is referred to as the second learning model. The information processing method according to Appendix 12, further comprising a step to be added as one of the models constituting the above.

(Appendix 14)
Computer,
A means for generating first data representing the situation grasped from the information based on the received information.
A means of generating second data representing a concept related to the situation based on the first data.
Based on the second data, a word expressing the concept is generated, and when the generated word does not match the solution corresponding to the information, the generated word is used as new information to generate the first data. Means to output to means,
A program that functions as.

(Appendix 15)
A computer-readable recording medium in which the program described in Appendix 14 is described.

This application claims priority based on Japanese application Japanese Patent Application No. 2020-132276 filed on August 4, 2020, and incorporates all of its disclosures here.

100 ... Input unit 200 ... Processing unit 210 ... Recognition processing unit 220 ... Conceptual processing unit 300 ... Output unit 400 ... Data processing module 410 ... Data acquisition unit 420 ... Storage unit 430 ... Identification unit 440 ... Data output unit 500 ... CPU
502 ... Main storage unit 504 ... Communication unit 506 ... Input / output interface unit 508 ... System bus 510 ... Output device 512 ... Input device 514 ... Storage device 1000 ... Information processing device

Claims (15)

1. Based on the received information, an input unit that generates first data representing the situation grasped from the information, and an input unit.
   A processing unit that generates a second data representing a concept related to the situation based on the first data, and a processing unit.
   It has an output unit that generates and outputs a word expressing the concept based on the second data.
   The information processing unit is configured to output the generated words as new information to the input unit when the generated words do not match the solution corresponding to the information. Device.
2. The information processing apparatus according to claim 1, wherein the output unit is configured to further output an instruction regarding an action according to the concept represented by the second data.
3. The processing unit has a first processing unit that generates a third data conceptualizing the situation represented by the first data, and the second data by converting the third data into another concept. The information processing apparatus according to claim 1 or 2, further comprising a second processing unit to be generated.
4. The first processing unit includes a first learning model and a first identification unit.
   The first data is data that maps the relationship between a plurality of elements representing the situation and their element values.
   The first learning model has a pattern that maps the relationship between a plurality of elements representing a specific situation and their element values, and a plurality of elements associated with the pattern and representing a concept corresponding to the specific situation. Contains multiple models, each containing a value that maps the relationship between and their element values,
   The first identification unit transfers the value associated with the pattern having the highest goodness of fit to the first data among the plurality of models of the first learning model. The information processing apparatus according to claim 3, wherein the information processing apparatus is selected as data.
5. The second processing unit includes a second learning model and a second identification unit.
   The third data is data that maps the relationship between a plurality of elements representing the concept and their element values.
   The second learning model represents a pattern that maps the relationship between a plurality of elements representing a specific concept and their element values, and another concept associated with the pattern and assumed from the specific concept. Contains multiple models, each containing a value that maps the relationship between multiple elements and their element values.
   The second identification unit transfers the value associated with the pattern having the highest goodness of fit to the third data among the plurality of models of the second learning model. The information processing apparatus according to claim 3 or 4, wherein the information processing apparatus is selected as data.
6. The second processing unit corresponds to the at least a part of the elements when the model corresponding to at least a part of the concept represented by the third data is not included in the second learning model. The information processing apparatus according to claim 5, wherein the model is added as one of the models constituting the second learning model.
7. The processing unit is composed of a plurality of data processing processors connected in series.
   The invention according to any one of claims 1 to 6, wherein each of the plurality of data processing processors is configured to output a value associated with a pattern at the information distance closest to the input. Information processing device.
8. Based on the received information, the first step of generating the first data representing the situation grasped from the information, and
   Based on the first data, a second step of generating second data representing the concept related to the situation, and
   It has a third step of generating and outputting a word expressing the concept based on the second data.
   An information processing method characterized by repeating the generated words as new information from the first step when the words generated in the third step do not match the solution corresponding to the information.
9. The information processing method according to claim 8, wherein in the third step, an instruction regarding an action according to the concept represented by the second data is further output.
10. The second step is
    A step of generating a third data that conceptualizes the situation represented by the first data, and
    The information processing method according to claim 8 or 9, further comprising a step of converting the third data into another concept to generate the second data.
11. The first data is data that maps the relationship between a plurality of elements representing the situation and their element values.
    The first learning model includes a pattern that maps the relationship between a plurality of elements representing a specific situation and their element values, and a plurality of elements associated with the pattern and representing a concept corresponding to the specific situation. Includes multiple models, each containing a value that maps the relationship to those element values,
    In the step of generating the third data, the value associated with the pattern having the highest goodness of fit to the first data among the plurality of models of the first learning model is referred to. The information processing method according to claim 10, wherein the data is selected as the third data.
12. The third data is data that maps the relationship between a plurality of elements representing the concept and their element values.
    The second learning model is a pattern that maps the relationship between a plurality of elements representing a specific concept and their element values, and a plurality of patterns that are associated with the pattern and represent another concept that is assumed from the specific concept. Contains multiple models, each containing a value that maps the relationship between the elements of and their element values.
    In the step of generating the second data, the value associated with the pattern having the highest goodness of fit to the third data among the plurality of models of the second learning model is referred to. The information processing method according to claim 10 or 11, wherein the data is selected as the second data.
13. When the model corresponding to at least a part of the elements of the concept represented by the third data is not included in the second learning model, the model corresponding to the at least a part of the elements is referred to as the second learning model. 12. The information processing method according to claim 12, further comprising a step of adding as one of the models constituting the above.
14. Computer,
    A means for generating first data representing the situation grasped from the information based on the received information.
    A means of generating second data representing a concept related to the situation based on the first data.
    Based on the second data, a word expressing the concept is generated, and when the generated word does not match the solution corresponding to the information, the generated word is used as new information to generate the first data. Means to output to means,
    A program that functions as.
15. A computer-readable recording medium in which the program according to claim 14 is described.

Images