Information processing system, information processing method, and information processing program

Abstract

The present invention is provided with: an acquisition unit that acquires voice data of a conversation from a mobile terminal received at each of a plurality of base stations, and position information of a base station that has received the voice data among the plurality of base stations, or position information of the mobile terminal that has received the voice data; a conversion unit that converts the voice data into character information; an information acquisition unit that obtains administrative division information from the position information; a storage unit that classifies and stores the character information on the basis of the administrative division information; and a neural network generation unit that generates a neural network for learning using the classified character information.

Description

Information processing system, information processing method, and information processing program

[0001] The present invention relates to an information processing system, an information processing method, and an information processing program, and more particularly to an information processing system, an information processing method, and an information processing program that can create generative AI having regional characteristics.

[0002] Conventionally, when collecting opinions from each region, nationwide surveys, telephone surveys, etc. Similarly, for example, word-of-mouth information, reputation, etc., can be obtained based on surveys, word-of-mouth reviews posted on review sites, etc. (See, for example, Patent Document 1).

[0003] Patent No. 5386806 specification

[0004] However, conventional methods have had problems such as being time-consuming and costly, making it difficult to collect information in a timely manner, and not being able to gather detailed opinions due to the halo effect, etc. Furthermore, when collecting opinions from social media, there are problems such as it being difficult to gather the opinions of the silent majority.

[0005] The present invention has been made in consideration of the above points, and aims to provide an information processing system, an information processing method, and an information processing program that are capable of picking up opinions that are in line with reality without spending time and money.

[0006] A first aspect of the present invention is an information processing system capable of communicating with a plurality of base stations, comprising an acquisition unit that acquires voice data of a conversation from a mobile terminal received at each of the plurality of base stations, and location information of a base station among the plurality of base stations that received the voice data, or location information of the mobile terminal that received the voice data; a conversion unit that converts the voice data into character information; an information acquisition unit that acquires administrative district information from the location information; a storage unit that classifies and stores the character information based on the administrative district information; and a neural network generation unit that generates a learning neural network using the classified character information.

[0007] In the first aspect of the present invention, the system may further include an artificial personality generation unit capable of generating an artificial personality for each region based on administrative division information by using a neural network formed by additional learning or transfer learning based on a pre-trained neural network of a large-scale language model trained using a neural network.

[0008] In the first aspect of the present invention, the system may further include a questionnaire unit that asks each of the artificial personalities generated for each of a plurality of regions questions a number of times proportional to the population of the region, obtains answers to the questions from the artificial personalities, and stores the obtained answers in a questionnaire result storage unit for each region.

[0009] In a first aspect of the present invention, the survey method may further include an extraction unit that performs natural language processing and then statistical processing on the responses for each of the multiple regions stored in the survey result storage unit, and extracts trends and / or peculiarities in the responses between a specified region among the multiple regions and other regions excluding the specified region among the multiple regions.

[0010] In the first aspect of the present invention, the system may further include an option providing unit and a selection result storage unit, and may repeatedly provide predetermined options to at least one of the artificial personalities generated for each administrative district a number of times proportional to the population of the administrative district, obtain a number of selection results for the predetermined options provided to the artificial personalities proportional to the population, and store all the obtained selection results for each administrative district in the selection result storage unit.

[0011] A second aspect of the present invention is an information processing modeling method, which is summarized as comprising the steps of: in an information processing system capable of communicating with a plurality of base stations, acquiring voice data of a conversation from a mobile terminal received at each of the plurality of base stations, and location information of a base station among the plurality of base stations that received the voice data, or location information of the mobile terminal that received the voice data; a conversion step of converting the voice data into character information; an information acquisition step of obtaining administrative district information from the location information; a storage step of classifying and storing the character information based on the administrative district information; and a neural network generation step of generating a training neural network using the classified character information.

[0012] A third aspect of the present invention is an information processing program executed in an information processing system capable of communicating with a plurality of base stations, which causes a computer to realize an acquisition function for acquiring voice data of a conversation from a mobile terminal received at each of the plurality of base stations, location information of a base station among the plurality of base stations that received the voice data, or location information of the mobile terminal that received the voice data, a conversion function for converting the voice data into character information, an information acquisition function for obtaining administrative district information from the location information, a storage function for classifying and storing the character information based on the administrative district information, and a neural network generation function for generating a training neural network using the classified character information.

[0013] According to the present invention, it is possible to provide an information processing system, an information processing method, and an information processing program that are capable of collecting opinions that are in line with reality without spending time and money.

[0014] FIG. 1 is a diagram illustrating an outline of an information processing system according to an embodiment of the present invention. FIG. 2 is a block diagram illustrating an example of the configuration of a data collection unit of the information processing system according to the embodiment. FIG. 3 is a block diagram illustrating an example of the configuration of an information processing unit of the information processing system according to the embodiment. FIG. 4 is a flowchart illustrating an information processing method according to the embodiment. FIG. 5 is a flowchart illustrating an example of an operation showing a usage form of a learning neural network generated by the information processing method according to the embodiment.

[0015] Next, an embodiment of the present invention will be described with reference to the drawings. In the description of the drawings relating to the embodiment, the same or similar parts are designated by the same or similar reference numerals. Of course, there are also parts whose relationships differ between the drawings.

[0016] Furthermore, the embodiments are merely examples of devices and methods for embodying the technical idea of ​​the present invention, and the technical idea of ​​the present invention does not limit the configuration of each component to those described below. The technical idea of ​​the present invention can be modified in various ways within the technical scope defined by the claims.

[0017] An overview of an information processing system 10 according to this embodiment will be described with reference to Fig. 1. As shown in Fig. 1, an example of the information processing system 10 according to this embodiment includes data collection units 100a and 100b connected to a plurality of base stations 11a and 11b, respectively, and an information processing unit 200 connected to the data collection units 100a and 100b via a network 300.

[0018] The mobile terminal 12a communicates with the base station 11a that has jurisdiction over the mobile terminal 12a, and the mobile terminal 12b communicates with the base station 11b that has jurisdiction over the mobile terminal 12b. In Fig. 1, two base stations 11a and 11b are shown as examples of base stations connected to the information processing system 10 according to this embodiment, and the mobile terminals 12a and 12b are shown as examples of mobile terminals that communicate with the base stations 11a and 11b, respectively, but this is not limited to these, and the base stations connected to the information processing system 10 according to this embodiment may be two or more base stations, and the mobile terminals that communicate with each of the base stations may be one or more mobile terminals.

[0019] 1 shows only data collection units 100a and 100b as the data collection units that are components constituting the information processing system 10 according to this embodiment, but one data collection unit is connected to each base station, and one data collection unit is connected to each of two or more base stations connected to the information processing system 10 according to this embodiment, so the information processing system 10 according to this embodiment is composed of two or more data collection units 100a, 100b, ... and an information processing unit 200. The information processing system 10 according to this embodiment may be a computer such as a server, desktop, laptop, tablet, or smartphone, for example.

[0020] FIG. 2 is a block diagram showing an example of the configuration of the data collection unit 100a of the information processing system 10 according to this embodiment. FIG. 3 is a block diagram showing an example of the configuration of the information processing unit 200 of the information processing system 10 according to this embodiment. The data collection unit 100a is composed of an acquisition unit 101, a conversion unit 102, an information acquisition unit 103, a first storage unit 104, and a first communication unit 105. The information processing unit 200 is composed of a second communication unit 203, a second storage unit 204, a neural network generation unit 201, and an artificial personality generation unit 202. Note that a plurality of data collection units 100a, 100b, etc. are connected to the information processing unit 200 via a network 300, but the data collection units 100b, etc. are not shown in FIG. 2. The data collection units 100b, etc. are composed of the same components as the data collection unit 100a shown in FIG. 2.

[0021] In the information processing system according to this embodiment, the base station 11a receives analog-format received data transmitted from a mobile terminal 12a in a call state, and then converts the received analog-format received data into digital format and uses the converted data.

[0022] The acquisition unit 101 of the data collection unit 100a acquires the voice data of the conversation and location information including information on the location of the base station 11a from the transmission data, excluding personal information related to the owner of the mobile terminal 12a, among the received data converted into digital format by the base station 11a. Note that in the information processing system 10 according to the present embodiment, the acquisition unit 101 acquires the location information of the base station 11a that received the voice data, but this is not limited thereto, and the acquisition unit 101 may also acquire location information including information on the location of the mobile terminal 12a that received the voice data at the time of transmission.

[0023] The conversion unit 102 converts the voice data into text information by voice recognition.

[0024] The information acquisition unit 103 acquires administrative division information from the location information. The administrative division information is, for example, information on the location of the base station 11a or the location of the mobile terminal 12a at the time of transmission, such as which city, ward, town, village, etc. in the country is included in.

[0025] The first memory unit 104 stores the character information together with location information belonging to the character information, i.e., administrative district information obtained from either the location information of the base station that received the voice data converted into character information or the location information of the mobile terminal that received the voice data.

[0026] The acquisition of voice data and location information by the acquisition unit 101, the conversion of the voice data into character information by the conversion unit 102, and the acquisition of administrative district information from the location information by the information acquisition unit 103 are carried out for a certain period of time, and the character information and administrative district information obtained during that period are stored and accumulated in the first memory unit 104.

[0027] The first communication unit 105 transmits the character information and administrative division information stored in the first storage unit 104 to the information processing unit 200 via the network 300 .

[0028] The second communication unit 203 of the information processing unit 200 receives the character information and the administrative district information from the data collection unit 100a via the network 300. The second communication unit 203 also receives the character information and the administrative district information from the data collection units 100b... (not shown in Fig. 2) via the network 300.

[0029] The second storage unit 204 classifies and stores the character information received by the second communication unit 203 for each administrative district listed in the administrative district information. The second communication unit 203 also receives character information and administrative district information from the data collection unit 100a and data collection units 100b, etc. not shown in FIG. 2 via the network 300. Therefore, the second storage unit 204 receives character information and administrative district information from the data collection unit 100a and data collection units 100b, etc. not shown in FIG. 2, and classifies and stores all of the received character information for each administrative district listed in the administrative district information received together with the character information.

[0030] The neural network generation unit 201 generates a learning neural network by having the neural network generation AI learn the character information classified for each administrative district as training data.

[0031] The artificial personality generation unit 202 generates an artificial personality for each administrative district based on the administrative district information by additionally learning or transfer learning the training neural network generated by the neural network generation unit 201 on a general-purpose large-scale language model. Existing large-scale language models are able to understand basic language grammar through pre-training, understand responses to questions and the like based on this basic language understanding, and generate answers. By additionally learning or transfer learning the training neural network generated by the neural network generation unit 201 on the existing large-scale language model, it becomes possible to understand grammar specific to that administrative district in addition to basic language grammar, and an artificial personality specific to that administrative district is formed, for example, capable of generating answers that reflect the dialect of that administrative district. The additional learning or transfer learning described above requires a small amount of calculation for learning, and therefore can be completed in a short time and with few calculation resources.

[0032] The information processing system 10 according to the present embodiment may further include a questionnaire unit and a questionnaire result storage unit. The questionnaire unit may repeatedly ask predetermined questions to at least one of the artificial personalities generated for each administrative district a number of times proportional to the population of the administrative district, obtain answers to the predetermined questions given to the artificial personalities a number of times proportional to the population, and store all the obtained answers for each administrative district in the questionnaire result storage unit.

[0033] Generally, when a question is posed to an artificial personality generated using a large-scale language model, for example, if the same question is posed repeatedly a predetermined number of times, different answers of the same sentence will be obtained the predetermined number of times. For example, if a single question for which the answer is somewhat uniquely determined is posed to an artificial personality 10 times, 10 answers that are nearly the same in content will be obtained. As an example, if the question "What mammal has horns and weighs more than one ton?" is posed to an artificial personality 10 times, there are 10 possible answers, but the content of all answers is considered to be "rhinoceros." On the other hand, if a single question for which the answer is not uniquely determined is posed to an artificial personality 10 times, answers that differ greatly from one another may be obtained depending on the number of times the question is asked. As an example, if the question "What do you think our future will be with the advent of AI?" is posed to an artificial personality 10 times, all 10 answers may be significantly different from one another.

[0034] When the questionnaire unit repeatedly asks predetermined questions to the artificial personalities a number of times proportional to the population of the administrative division, different answers are obtained from the artificial personalities a number of times proportional to the population. The questionnaire unit asks predetermined questions to at least one of the artificial personalities generated for each administrative division a number of times proportional to the population of the administrative division, and different answers are obtained from each artificial personality a number of times proportional to the population of the administrative division to which the artificial personality belongs. The artificial personalities generated by the information processing system 10 according to this embodiment have tendencies specific to the administrative division to which the artificial personality belongs. Therefore, although the answers obtained by the questionnaire unit are answers to questions asked to the artificial personality, they can be considered to be equivalent to answers obtained by asking questions to actual residents in a number proportional to the population of the administrative division to which the artificial personality belongs.

[0035] The information processing system 10 according to this embodiment may further include an extraction unit. The extraction unit performs natural language processing and then statistical processing on the responses for each of the multiple administrative divisions stored in the survey result storage unit. This makes it possible to extract words, grammar, etc. that are used specifically in a specific administrative division. In this way, it is possible to extract trends and / or peculiarities characteristic of a specific administrative division from the responses in that administrative division among the multiple administrative divisions. When the extraction unit performs statistical processing, it is possible to extract characteristic and unique responses by adjusting the parameters used in the statistical processing.

[0036] The information processing system 10 according to the present embodiment may further include an option providing unit. The option providing unit may repeatedly provide predetermined options to at least one of the artificial personalities generated for each administrative district a number of times proportional to the population of the administrative district, obtain a number of selection results for the predetermined options provided to the artificial personalities proportional to the population, and store all the obtained selection results for each administrative district in the selection result storage unit.

[0037] The information processing method according to this embodiment will be described with reference to the flowchart of FIG.

[0038] In step S401, the acquisition unit acquires voice data of a conversation from a mobile terminal received at each of the multiple base stations, and location information of the base station among the multiple base stations that received the voice data, or location information of the mobile terminal that received the voice data (acquisition step).

[0039] In step S402, the conversion unit converts the voice data into character information (conversion step).

[0040] In step S403, the information acquisition unit acquires administrative division information from the location information (information acquisition step).

[0041] In step S404, the classification unit classifies the character information based on the administrative division information (classification step).

[0042] In step S405, the neural network generation unit generates a neural network for training using the classified character information.

[0043] An example of an operation showing a usage form of the training neural network generated in the procedure shown in Fig. 4 will be described with reference to the flowchart of Fig. 5. The example shown in Fig. 5 shows a procedure in which the information processing system 10 according to this embodiment further includes a questionnaire unit and a questionnaire result storage unit, and after an artificial personality is generated using the training neural network, a questionnaire survey is carried out by the questionnaire unit.

[0044] In step S501, the artificial personality generation unit generates an artificial personality for each administrative district based on the administrative district information by additionally learning the learning neural network generated by the neural network generation unit to a general-purpose large-scale language model (artificial personality generation step).

[0045] In step S502, the questionnaire unit repeatedly asks predetermined questions to at least one of the artificial personalities generated for each administrative district a number of times proportional to the population of each administrative district (question providing step).

[0046] In step S503, the questionnaire unit obtains answers to the predetermined questions given to the artificial personality the number of times proportional to the population, and stores all the obtained answers in the questionnaire result storage unit for each administrative district (answer acquisition step).

[0047] Each unit of the information processing system 10 may be realized as a function of a computer's arithmetic processing device or the like. That is, the acquisition unit 101, conversion unit 102, information acquisition unit 103, first storage unit 104, and first communication unit 105 of the data collection unit 100a of the information processing system 10 may be realized as an acquisition function, a conversion function, an information acquisition function, a first storage function, and a first communication function, respectively, by a computer's arithmetic processing device or the like. Furthermore, the second communication unit 203, second storage unit 204, neural network generation unit 201, and artificial personality generation unit 202 of the information processing unit 200 of the information processing system 10 may be realized as a second communication function, a second storage function, a neural network generation function, and an artificial personality generation function, respectively, by a computer's arithmetic processing device or the like. An information processing program can cause a computer to realize each of the above-mentioned functions. The information processing program may be recorded on a computer-readable non-transitory storage medium, such as a memory, a solid-state drive, a hard disk drive, or an optical disc. The storage medium may also be rephrased as a non-transitory computer-readable medium that stores the information processing program. The information processing program may also be transmitted online. As described above, each unit of the information processing system 10 may be realized by a processing unit of a computer or the like. The processing unit or the like is configured, for example, by an integrated circuit or the like. Therefore, each unit of the information processing system 10 may be realized as a circuit constituting the processing unit or the like. That is, the acquisition unit 101, the conversion unit 102, the information acquisition unit 103, the first storage unit 104, and the first communication unit 105 of the data collection unit 100a of the information processing system 10 may be realized as an acquisition circuit, a conversion circuit, an information acquisition circuit, a first storage circuit, and a first communication circuit constituting the processing unit or the like of a computer. Furthermore, the second communication unit 203, the second storage unit 204, the neural network generation unit 201, and the artificial personality generation unit 202 of the information processing unit 200 of the information processing system 10 may be realized as a second communication circuit, a second storage circuit, a neural network generation circuit, and an artificial personality generation circuit constituting the processing unit or the like of a computer.

[0048] The information processing system 10 can combine one or any combination of the above-described multiple units. In this disclosure, the term "information" is used, but the term "information" can be replaced with "data," and the term "data" can be replaced with "information."

[0049] The information processing program of the embodiment can be implemented using, for example, scripting languages ​​such as ActionScript, JavaScript (registered trademark), Python, and Ruby, or compiler languages ​​such as C, C++, C#, Objective-C, Swift, and Java (registered trademark).

[0050] In one aspect of the information processing method, an information processing system capable of communicating with a plurality of base stations includes an acquisition step of acquiring voice data of a conversation from a mobile terminal received at each of the plurality of base stations, and location information of a base station among the plurality of base stations that received the voice data, or location information of the mobile terminal that received the voice data, a conversion step of converting the voice data into text information, an information acquisition step of obtaining administrative district information from the location information, a storage step of classifying and storing the text information based on the administrative district information, and a neural network generation step of generating a training neural network using the classified text information. This allows the information processing method to achieve additional learning and effects similar to those of the information processing system of the above-mentioned aspect.

[0051] An information processing program according to one embodiment is an information processing modeling program executed in an information processing system capable of communicating with a plurality of base stations, and causes a computer to realize an acquisition function for acquiring voice data of a conversation from a mobile terminal received at each of the plurality of base stations, location information of a base station among the plurality of base stations that has received the voice data, or location information of the mobile terminal that has received the voice data, a conversion function for converting the voice data into text information, an information acquisition function for acquiring administrative district information from the location information, a storage function for classifying and storing the text information based on the administrative district information, and a neural network generation function for generating a training neural network using the classified text information. As a result, the information processing program can achieve the same effects as the information processing system according to the above-described embodiment.

[0052] The information processing system of the present invention makes it possible to conduct questionnaire surveys and opinion surveys that eliminate obstacles that tend to occur in interpersonal communication, such as the halo effect. When natural language processing and statistical processing are performed on the survey results for the artificial personality generated by the information processing system of the present invention, unique opinions can be extracted by adjusting parameters. Since the system is based on telephone conversations, conversation data that reflects everyday mental states can be used as learning data. In addition, incremental learning is used to generate the artificial personality, and a learning method with a low computational load is used, making it possible to learn in a short period of time with limited computational resources.

[0053] As mentioned above, the present invention naturally includes various embodiments not described herein. Therefore, the technical scope of the present invention is defined only by the invention-specifying matters according to the scope of the claims that are appropriate from the above description.

[0054] The present invention aims to enable the collection of realistic opinions without the time and cost involved in conducting questionnaire surveys and opinion polls. Generally, conducting nationwide questionnaire surveys and opinion polls requires time and cost, making it difficult to conduct the survey in a timely manner, and making it difficult to collect detailed opinions due to factors such as the halo effect. By using the present invention, nationwide questionnaire surveys and opinion polls can be conducted in a short amount of time and at low cost, thereby contributing to the achievement of Goal 9 of the Sustainable Development Goals (SDGs), "Build resilient infrastructure, promote inclusive and sustainable industrialization, and promote innovation and resilience."

[0055] 10 Information processing system 11a, 11b Base station 100a, 100b Data collection unit 101 Acquisition unit 102 Conversion unit 103 Information acquisition unit 104 First memory unit 105 First communication unit 200 Information processing unit 201 Neural network generation unit 202 Artificial personality generation unit 203 Second communication unit 204 Second memory unit 300 Network

Claims

1. An information processing system capable of communicating with a plurality of base stations, comprising: an acquisition unit that acquires voice data of a conversation from a mobile terminal received at each of the plurality of base stations, and location information of the base station among the plurality of base stations that received the voice data, or location information of the mobile terminal that received the voice data; a conversion unit that converts the voice data into character information; an information acquisition unit that acquires administrative district information from the location information; a storage unit that classifies and stores the character information based on the administrative district information; and a neural network generation unit that uses the classified character information to generate a neural network for training.

2. The information processing system according to claim 1, further comprising an artificial personality generation unit capable of generating an artificial personality for each region based on the administrative division information, using a neural network formed by additional learning or transfer learning based on a pre-trained neural network of a large-scale language model trained using a neural network.

3. The information processing system according to claim 2, further comprising a questionnaire unit and a questionnaire result storage unit, wherein the system asks each of the artificial personalities generated for each of the plurality of regions a number of questions proportional to the population of the region, obtains answers to the questions from the artificial personalities, and stores the obtained answers in the questionnaire result storage unit for each of the regions.

4. The information processing system described in claim 3, further comprising an extraction unit that performs natural language processing and then statistical processing on the responses for each of the plurality of regions stored in the survey result storage unit, and extracts trends and / or peculiarities in the responses between a specified region among the plurality of regions and other regions among the plurality of regions excluding the specified region.

5. An information processing system according to claim 2, further comprising an option providing unit and a selection result storage unit, wherein predetermined options are repeatedly given to each of the artificial personalities generated for each of the plurality of regions a number of times proportional to the population of the region, a number of selection results are obtained for the predetermined options given to the artificial personalities a number of times proportional to the population, and all of the obtained selection results are stored in the selection result storage unit for each region.

6. An information processing method in an information processing system capable of communicating with a plurality of base stations, wherein a computer executes the following steps: an acquisition step of acquiring voice data of a conversation from a mobile terminal received at each of the plurality of base stations, and location information of the base station among the plurality of base stations that received the voice data, or location information of the mobile terminal that received the voice data; a conversion step of converting the voice data into character information; an information acquisition step of obtaining administrative division information from the location information; a storage step of classifying and storing the character information based on the administrative division information; and a neural network generation step of generating a training neural network using the classified character information.

7. An information processing modeling program executed in an information processing system capable of communicating with multiple base stations, which causes a computer to realize the following: an acquisition function that acquires voice data of a conversation from a mobile terminal received at each of the multiple base stations, and location information of the base station among the multiple base stations that received the voice data, or location information of the mobile terminal that received the voice data; a conversion function that converts the voice data into text information; an information acquisition function that acquires administrative division information from the location information; a storage function that classifies and stores the text information based on the administrative division information; and a neural network generation function that uses the classified text information to generate a neural network for training.

Images