Information processing system

Abstract

To provide a highly usable information processing system that optimally performs analysis of vehicle driving data using natural language. [Solution] An information processing system according to one aspect of the present disclosure comprises an acquisition unit, a conversion unit, a response unit, and an output unit. The acquisition unit acquires vehicle driving data. The conversion unit converts the driving data into text data having time-series information and generates intermediate data. The response unit acquires instruction information and generates a response sentence for the instruction information in natural language based on the intermediate data using a trained large-scale language model. The output unit outputs the generated response sentence.

Description

Technical Field

[0004]

[0001] The present invention relates to an information processing system.

Background Art

[0002] By analyzing the driving data of a vehicle, which is time-series data, it is possible to perform prediction, evaluation, or anomaly detection of vehicle driving. Here, if the analysis result is output in natural language, the user can grasp the analysis result even if they are not proficient in the technology. Patent Document 1 discloses a computer-implemented method for outputting multivariate predictive time-series data.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The computer-implemented method described in Patent Document 1 uses a trained model of a generative adversarial network (GAN: Generative Adversarial Networks). This GAN architecture can learn raw waveform features and generate (predict) multivariate multi-step time-series data, and can receive raw waveform data and generate raw waveform data for subsequent time steps. However, the computer-implemented method described in Patent Document 1 does not describe outputting multivariate predictive time-series data in natural language. Therefore, the user has to read the meaning of the generated raw waveform by themselves.

[0005] In view of the above problems, the present disclosure provides a highly usable information processing system that preferably performs analysis on vehicle driving data in natural language.

Means for Solving the Problems

[0006] An information processing system according to one aspect of this disclosure comprises an acquisition unit, a conversion unit, a response unit, and an output unit. The acquisition unit acquires vehicle driving data. The conversion unit converts the driving data into text data having time-series information and generates intermediate data. The response unit acquires instruction information and generates a response sentence for the instruction information in natural language based on the intermediate data using a trained large-scale language model. The output unit outputs the generated response sentence.

[0007] The above information processing system may further include a drive system determination unit for determining the drive system of a vehicle and a vehicle type determination unit for determining the vehicle type. In this information processing system, the response unit may include multiple trained large-scale language models corresponding to each of the multiple drive systems, and may input the information of the determined vehicle type into the large-scale language model corresponding to the determined drive system to generate a response sentence.

[0008] In the above information processing system, the vehicle is an automobile, the driving data includes at least one of CAN information and navigation route information, and the conversion unit may use a time-series base model as an encoder and convert the driving data into text data using an embedded representation.

[0009] In the above information processing system, the acquisition unit may acquire multiple driving data, the conversion unit may generate multiple intermediate data corresponding to each of the multiple driving data, and the response unit may generate a response statement for comparative evaluation of the multiple driving data based on the multiple intermediate data.

[0010] An information processing system according to one aspect of this disclosure comprises an acquisition unit, a representation unit, a conversion unit, and an output unit. The acquisition unit acquires instruction information related to the driving of a vehicle. The representation unit uses a trained large-scale language model to generate intermediate data having text data indicating time-series information based on instruction information consisting of natural language. The conversion unit converts the intermediate data into vehicle driving data. The output unit outputs the driving data. [Effects of the Invention]

[0011] According to this disclosure, in view of the above-mentioned issues, this disclosure can provide a highly usable information processing system that suitably performs analysis of vehicle driving data using natural language. [Brief explanation of the drawing]

[0012] [Figure 1] This is a diagram showing the configuration of the information processing system according to Embodiment 1. [Figure 2] This is a flowchart of the information processing method according to Embodiment 1. [Figure 3] This is a block diagram of the information processing system according to Embodiment 2. [Figure 4] This is a block diagram of the information processing system according to Embodiment 3. [Figure 5] This is a block diagram of the information processing system according to Embodiment 4. [Figure 6] This is a block diagram illustrating the hardware configuration of a computer. [Modes for carrying out the invention]

[0013] The present invention will be described below through embodiments of the invention, but the invention claimed is not limited to the following embodiments. Furthermore, not all of the configurations described in the embodiments are necessarily essential as means of solving the problem. For clarity of explanation, the following descriptions and drawings have been omitted and simplified as appropriate. In each drawing, the same elements are denoted by the same reference numerals, and redundant explanations have been omitted where necessary.

[0014] <Embodiment 1> Referring to Figure 1, the information processing system 10 according to Embodiment 1 will be described. Figure 1 is a configuration diagram of the information processing system 10 according to Embodiment 1. The information processing system 10 acquires vehicle driving data and generates response sentences in natural language in response to instructions. Here, a vehicle is a moving object that has a drive unit and moves. A vehicle is, for example, an automobile, a train, or an unmanned transport device. Driving data is time-series data related to the vehicle's movement. Driving data includes data related to location information, travel path information, vehicle speed, acceleration / deceleration operations, or fuel level information. The information processing system 10 comprises an acquisition unit 101, a conversion unit 102, a response unit 103, and an output unit 104.

[0015] The acquisition unit 101 acquires vehicle driving data. Driving data is data related to the vehicle's operation. Driving data includes, for example, control information, monitoring information, or vehicle movement path information for each part of the vehicle. Driving data is usually multivariate time series data and continuous numerical data, but is not limited to this. Driving data may also be series data aggregated by road link. Driving data may also be univariate data.

[0016] The conversion unit 102 generates intermediate data by converting the acquired driving data into text data. This allows time-series data, which is difficult for language models to process, to be converted into text data, which is easier for language models to process. Methods for converting to text data include, for example, text conversion, quantization, or embedding of numerical data.

[0017] The response unit 103 acquires instruction information. The instruction information includes instructions for generating the response text. The instructions may be requests from the user who entered them. Examples of instructions include summaries of driving data, evaluations of the guidance route, or suggestions for solutions to driving risks. The instruction information may be acquired by selecting from predetermined options, or by acquiring it from the entered text.

[0018] The response unit 103 receives, as instruction content, the voice, gestures, or operations of the user, etc. via an input interface such as a connected microphone, camera, touch panel, or button. The response unit 103 performs processes such as image analysis, speech recognition, or word extraction on the received instruction content to obtain instruction information. According to this, the information processing system 10 can answer the content that the user wants to know in the response sentence. Also, the instruction information may be set in advance.

[0019] Also, the response unit 103 uses a pre-trained large language model to generate a response sentence for the instruction information in natural language based on intermediate data. The large language model is a generative AI (Artificial Intelligence) model constructed using a large amount of text data and deep learning. The large language model is a natural language processing model and is not suitable for processing time-series data. The large language model is characterized by a huge amount of computational power, data volume, and number of parameters. The large language model is also called an LLM (Large Language Model).

[0020] The output unit 104 outputs the response sentence generated by the response unit 103. The output unit 104 is connected to, for example, a monitor or a printer, etc., and visually displays the response sentence. The output unit 104 may be connected to a speaker, etc., and output the response sentence as voice. Also, the output unit 104 may be connected to an external device and output the response sentence to the external device as data.

[0021] According to this, the information processing system 10 can output the processing result for the time-series data in natural language. Therefore, the information processing system 10 can suitably output a response sentence for the instruction information based on the running data.

[0022] Figure 2 is a flowchart of the information processing method according to Embodiment 1. The flow of the information processing method for the information processing system 10 includes steps S11 to S15.

[0023] In step S11, the information processing system 10 acquires vehicle driving data. This driving data includes, for example, control information for each part of the vehicle or movement path information. In step S12, the information processing system 10 converts the driving data acquired in step S11 from continuous numerical data to text data and generates intermediate data. This allows the information processing system 10 to process the driving data using a language model.

[0024] In step S13, the information processing system 10 acquires instruction information. Here, the instruction information is information that includes instructions for generating a response. The instructions may be requests from the user who entered the information. Examples of instructions include summarizing driving data, evaluating the guidance route, or proposing solutions to driving risks. The instruction information may be acquired by selecting from predetermined options, or by text input or speech transcription.

[0025] In step S14, the information processing system 10 uses a pre-trained large-scale language model to generate a response sentence in natural language based on intermediate data for the instruction information. The response sentence may be in the same language as the input instruction information or in a different language. In step S15, the information processing system 10 outputs the response sentence generated in step S14. The information processing system 10 may output the response sentence by displaying it visually or audibly. The information processing system 10 is not limited to these methods and may also output the response sentence by communicating it externally.

[0026] As explained above, the information processing system 10 outputs a response in natural language based on the vehicle's driving data. This allows the information processing system 10 to perform analysis on the vehicle's driving data in natural language, thereby improving usability. Therefore, users of the information processing system 10 can easily understand the results of the driving data analysis.

[0027] The information processing system 10 may also have a processor and a storage device, although these are not shown in the diagram. The storage device of the information processing system 10 may include, for example, a non-volatile memory such as flash memory or an SSD (Solid State Drive). In this case, the storage device stores a computer program (hereinafter also simply referred to as a program) for executing the above-described method. The processor loads the computer program from the storage device into a buffer memory such as DRAM (Dynamic Random Access Memory) and executes the program.

[0028] Each component of the information processing system 10 may be implemented with dedicated hardware. Furthermore, some or all of each component may be implemented by general-purpose or dedicated circuits, processors, etc., or combinations thereof. These may be implemented by a single chip or by multiple chips connected via a bus. Some or all of each component of each device may be implemented by a combination of the aforementioned circuits, etc., and programs. Processors include CPUs (Central Processing Units), GPUs (Graphics Processing Units), FPGAs (Field-Programmable Gate Arrays), etc. Also, at least a portion of the processing performed by the information processing system 10 may be provided as SaaS (Software as a Service). The descriptions of the configurations described herein may also apply to other systems described below in this disclosure.

[0029] Furthermore, while the information processing system 10 is typically implemented by hardware mounted on a mobile device, some hardware may be installed elsewhere rather than on the mobile device. In this case, the device installed elsewhere (e.g., a server) and the device installed on the mobile device are connected via a communication network.

[0030] <Embodiment 2> Figure 3 is a block diagram of the information processing system 11 according to Embodiment 2. The information processing system 11 determines the drive system and vehicle type of an automobile and generates a response statement corresponding to the driving data, drive system, and vehicle type. The information processing system 11 according to Embodiment 2 has some of the same configuration as the information processing system 10 in Figure 1. Therefore, a description of the configuration of the information processing system 11, which performs the same processing as the information processing system 10, will be omitted. The information processing system 11 includes an acquisition unit 111, a conversion unit 112, a drive system determination unit 113, a vehicle type determination unit 114, a response unit 115, and an output unit 116.

[0031] The acquisition unit 111 acquires driving data from the vehicle, including at least one of CAN (Controller Area Network) information and navigation route information. For example, the acquisition unit 111 acquires and stores driving data sequentially from the vehicle while it is driving. Alternatively, the acquisition unit 111 may acquire driving data stored by the vehicle at regular intervals or after driving. The acquisition unit 111 may acquire some or all of the driving data within the in-vehicle system or in-vehicle navigation system, or it may acquire some of the driving data by inferring from data stored within the in-vehicle system or in-vehicle navigation system. The acquisition unit 111 may acquire some or all of the driving data via a network, or it may acquire some of the driving data by inferring from stored data in the cloud.

[0032] CAN information includes, for example, location information, vehicle status information, operation information, or abnormality information. Here, location information is information such as latitude, longitude, and time. Vehicle status information is information such as vehicle speed, acceleration, and fuel or battery level and consumption. Operation information is information such as accelerator or brake operation amount, shift lever position, and steering wheel operation amount. Abnormality information is, for example, diagnostic information and abnormality detection notifications from driver abnormality detection means. Diagnostic information is information of the diagnostic results from the vehicle system's self-diagnostic function.

[0033] Navigation route information is travel route information set in the in-car navigation system or navigation app. Navigation route information includes, for example, route identification information, static road information, dynamic road information, or landmark information. Route identification information is information that identifies the guidance route and includes the route's link ID (Identifier), latitude, longitude, and waypoint list. Static road information is static information associated with the road and includes information such as road gradient, speed limit, road type, or presence or absence of traffic lights. Dynamic road information is dynamic information associated with the road and includes information such as congestion information, predicted speed, road surface information, or weather information. Landmark information is information that indicates candidate landmarks and includes information such as the type and location of nearby POIs (Points of Interest), such as refueling facilities or charging facilities.

[0034] The conversion unit 112 generates intermediate data by converting the driving data into text data using an embedded representation. Here, embedding is a method of converting input data into processing data that expresses meaning and relationships. In this case, it is converted into text data for processing by a language model. Specifically, the conversion unit 112 extracts information such as waveform features and relationships between multiple variables from the driving data and converts it into text data using an embedded representation so that the language model can handle it. As a result, the information processing system 11 can appropriately handle driving data even if it is large-scale or multivariate data.

[0035] Here, the conversion unit 112 uses a time-series-based model as an encoder. The time-series-based model is an AI model that has been trained on a large amount of time-series data. While time-series-based models are usually used for predicting time-series data, here they are used for embedding the time-series data. The time-series-based model is, for example, MOMENT, but it is not limited to this; other time-series models that can appropriately embed waveform features may also be used.

[0036] The drive system determination unit 113 determines the drive system of the vehicle. The drive system determination unit 113 obtains drive system information from the vehicle's in-vehicle network and determines the drive system based on the drive system information. The drive system information includes, for example, gasoline engine, PHEV (Plug-in Hybrid Electric Vehicle), BEV (Battery Hybrid Electric Vehicle), FCEV (Fuel Cell Electric Vehicle), etc., but is not limited to these. Note that the drive system is also called the powertrain.

[0037] The vehicle type identification unit 114 identifies the vehicle type of the automobile. The vehicle type identification unit 114 obtains vehicle type identification information from the automobile's in-vehicle network and identifies the vehicle type based on the vehicle type identification information. The vehicle type identification information is the vehicle identification number, vehicle identification number, or vehicle model number. The identified vehicle type is not limited to the so-called vehicle type, but may also be the vehicle model number or the vehicle name.

[0038] The response unit 115 acquires instruction information and generates a response sentence for the instruction information based on intermediate data, drive system, and vehicle type. The instruction information is information that includes the instruction content related to the generation of the response sentence. The instruction content may be, for example, a summary of driving data, causal inference, or anomaly detection. Specifically, the instruction content may be voice input such as "Summarize the driving history" or "Tell me a route with a low risk of running out of power." The instruction content may also be a user response to the output of the information processing system 11.

[0039] The response unit 115 may use predetermined instruction information. The response unit 115 may also automatically select instruction information from a plurality of predetermined instruction candidates based on the execution timing, etc. The execution timing may be, for example, when setting the navigation system, after arriving at the destination, or before shutting down the in-vehicle system. With this, the information processing system 11 can automatically perform an explanation of the navigation route or an evaluation of the driving history at a predetermined timing.

[0040] The response unit 115 includes multiple pre-trained large-scale language models corresponding to each driving method. Here, the large-scale language model is, for example, Qwen, but it is not limited to this and can be any language model. Based on the driving method determined by the driving method determination unit 113, the response unit 115 reads out the corresponding large-scale language model. This allows the information processing system 11 to respond to changes in the characteristics of the driving data due to differences in driving methods. Each large-scale language model may have undergone supervised learning on intermediate data and response sentences in advance.

[0041] The response unit 115 inputs the vehicle type into a large-scale language model. This allows the information processing system 11 to generate response sentences that correspond to the different characteristics and performance of each vehicle type. Therefore, the accuracy of the response sentences can be improved. Furthermore, by not training the large-scale language model for each vehicle type individually, but instead inputting the vehicle type into the large-scale language model, the training cost can be reduced while maintaining the accuracy of the response sentences.

[0042] Furthermore, the response unit 115 may convert the received instructions into a different language to obtain instruction information, or it may generate a response in a language different from the instructions. This allows for the generation of a response in a language that is easily processed by a large-scale language model, thereby improving response accuracy.

[0043] The output unit 116 outputs the response text generated by the response unit 115. When the output unit 116 outputs the response text in written or spoken form, it may output it in a language different from the instruction content and the generated response text. Specifically, if the response unit 115 receives a question in Japanese as the instruction content, it may output it in English as well as Japanese. Furthermore, the output unit 116 may output in multiple languages. This allows the response content to be conveyed to passengers with different backgrounds.

[0044] As explained above, the information processing system 11 can respond to changes in the characteristics of driving data due to differences in drive systems and vehicle types, and can improve the accuracy of the analysis results for vehicle driving data. As a result, the information processing system 11 can output analysis results of driving data that are more in line with the user's intentions.

[0045] Furthermore, the information processing system 11 does not necessarily have to include a vehicle type identification unit 114. In this case, the response unit 115 generates a response sentence corresponding to the instruction information from the intermediate data and the drive system. Also, the information processing system 11 does not necessarily have to include a drive system identification unit 113 and a vehicle type identification unit 114 separately. In this case, the vehicle type identification unit 114 also functions as a drive system identification unit 113 by identifying the vehicle's drive system from the model number, etc.

[0046] Furthermore, the information processing system 11 may prompt the user to input drive system and vehicle type information and generate a response. In this case, the information processing system may provide a driving evaluation service and the driving data may be used as a driving history uploaded by the user to the server. In this case, the information processing system 11 can be used for driving skill evaluation and feedback.

[0047] <Embodiment 3> Figure 4 is a block diagram of the information processing system 12 according to Embodiment 3. The information processing system 12 has multiple acquisition units to acquire multiple driving data and generates a response statement based on a comparative evaluation between the multiple driving data. The information processing system 12 according to Embodiment 3 has some of the same configuration as the information processing system 10 in Figure 1. Therefore, the description of the configuration of the information processing system 12, which performs the same processing as the information processing system 10, will be omitted. The information processing system 12 includes a first acquisition unit 1211, a second acquisition unit 1212, a first conversion unit 1221, a second conversion unit 1222, a response unit 123, and an output unit 124.

[0048] The first acquisition unit 1211 acquires first driving data. The second acquisition unit 1212 acquires second driving data. The first and second driving data may be acquired at different times and may be acquired from different vehicles. The first and second driving data may be navigation route information for different routes.

[0049] The first conversion unit 1221 generates first intermediate data by converting the first driving data into text data using an embedded representation. The second conversion unit 1222 generates second intermediate data by converting the second driving data into text data using an embedded representation. The first conversion unit 1221 and the second conversion unit 1222 may use the same time-series-based model or different time-series-based models.

[0050] The response unit 123 acquires instruction information and generates a response statement for the instruction information based on the first intermediate data and the second intermediate data. Here, the instruction information includes instructions regarding a comparison between the first travel data and the second travel data. The instructions may include, for example, a summary of the differences between multiple routes, or a comparative evaluation of the second travel data with respect to the first travel data. The output unit 124 outputs the response statement generated by the response unit 123.

[0051] As explained above, the information processing system 12 can output the results of a comparative evaluation of multiple driving data sets. This allows users of the information processing system 12 to easily understand the comparison results of multiple driving data sets. Although the information processing system 12 described with reference to Figure 4 acquires two driving data sets, it is not limited to this, and may also acquire three or more driving data sets and output a response statement based on a comparative evaluation.

[0052] Furthermore, the information processing system 12 may store a portion of multiple driving data and output a response regarding the comparison result with newly input driving data. This allows the information processing system 12 to, for example, acquire ideal driving data first and then provide a response comparing it to recent driving history. Additionally, if a pre-stored route is impassable due to an accident or construction, the information processing system 12 can summarize the differences between the pre-stored route and an alternative route and provide a response.

[0053] <Embodiment 4> Figure 5 is a block diagram of the information processing system 13 according to Embodiment 4. The information processing system 13 generates driving data based on instruction information. That is, the information processing system 13 executes the processing of the information processing system 10 described with reference to Figure 1 in reverse order. The information processing system 13 includes an acquisition unit 131, a representation unit 132, a conversion unit 133, and an output unit 134.

[0054] The acquisition unit 131 acquires instruction information related to the vehicle's operation. The instruction information is in natural language text format. The instruction information includes instructions related to the driving data desired by the user. The instruction information can be acquired, for example, by transcribing instructions given via voice input. Alternatively, the instruction information may be acquired by the user selecting from multiple options. The method of acquiring instruction information is not limited to these. The instruction content may, for example, be a suggestion of route information that meets the requirements, but is not limited to this.

[0055] The representation unit 132 uses a pre-trained large-scale language model to generate intermediate data, including text data with time-series information, based on instruction information consisting of natural language. The intermediate data includes, for example, representations of features necessary for driving data related to driving. Features necessary for driving data include, for example, waveform features and information such as relationships between multiple variables.

[0056] The conversion unit 133 converts the intermediate data generated by the representation unit 132 into vehicle driving data. Here, the conversion unit 133 uses a time series base model as a time series decoder and decodes the intermediate data into multivariate time series information based on the representation of features necessary for the driving data. Note that the data decoded from the intermediate data is not limited to multivariate time series information, but may also be univariate time series information or series information.

[0057] The output unit 134 outputs the driving data converted by the conversion unit 133. For example, the output unit 134 outputs time-series data, which is navigation route information, to the in-vehicle navigation system as driving data. The output unit 134 may also output time-series data as control signals to a vehicle equipped with an autonomous driving function.

[0058] As explained above, the information processing system 13 can output time-series data in accordance with instructions given by the user in natural language. Therefore, the information processing system 13 can perform analysis of vehicle driving data using natural language. As a result, the user of the information processing system 13 can obtain desired navigation route information or control the vehicle by simply giving instructions in natural language.

[0059] It should be noted that the present invention is not limited to the embodiments described above, and can be modified as appropriate without departing from the spirit of the invention. For example, the present invention can be implemented in the form of a method or a program. Furthermore, the information processing system may include any additional discriminators other than the drive type discriminator and the vehicle type discriminator. Here, the additional discriminator may be, for example, a driving tendency discriminator that discriminates the driver's driving tendencies. With this, the information processing system can propose a driving method or navigation route suitable for the driving tendencies.

[0060] <Example hardware configuration> The following describes examples of how each functional configuration of the information processing system described in this disclosure can be realized through a combination of hardware and software.

[0061] Figure 6 is a block diagram illustrating the hardware configuration of a computer. The information processing system in this disclosure can realize the above-described functions using a computer 500 including the hardware configuration shown in the figure. The computer 500 may be a portable computer such as a smartphone or tablet terminal, or a stationary computer such as a PC. The computer 500 may be a dedicated computer designed to realize each device, or it may be a general-purpose computer. The computer 500 can realize the desired functions by installing a predetermined application.

[0062] Computer 500 includes a bus 502, a processor 504, memory 506, a storage device 508, an input / output interface (I / F) 510, and a network interface (I / F) 512. Bus 502 is a data transmission path for the processor 504, memory 506, storage device 508, input / output interface 510, and network interface 512 to send and receive data to and from each other. However, the method of connecting the processor 504 and other components to each other is not limited to bus connection.

[0063] Processor 504 is a processor such as a CPU, GPU, or FPGA. Memory 506 is main memory implemented using RAM (Random Access Memory), etc.

[0064] The storage device 508 is an auxiliary storage device implemented using a hard disk, SSD, memory card, or ROM (Read Only Memory). The storage device 508 stores a program for realizing a desired function. The processor 504 reads this program into memory 506 and executes it to realize each functional component of each device.

[0065] The input / output interface 510 is an interface for connecting the computer 500 to input / output devices. For example, input devices such as keyboards and output devices such as display devices are connected to the input / output interface 510. The network interface 512 is an interface for connecting the computer 500 to a network. [Explanation of symbols]

[0066] 10, 11, 12, 13 Information Processing Systems 101, 111 Acquisition Department 102, 112 Conversion section 103, 115, 123 Answer section Output sections 104, 116, 124 113 Drive system determination unit 114 Vehicle type identification section 1211 First Acquisition Department 1212 Second Acquisition Department 1221 First Conversion Unit 1222 Second Conversion Unit 131 Acquisition Department 132 Expression part 133 Conversion section 134 Output section 500 Computers Bus 502 504 Processors 506 memory 508 Storage Devices 510 Input / Output Interfaces 512 Network Interfaces

Claims

1. An acquisition unit that acquires vehicle driving data, A conversion unit that converts the aforementioned driving data into text data having time-series information and generates intermediate data, A response unit that acquires instruction information and generates a response sentence for the instruction information in natural language based on the intermediate data using a trained large-scale language model, An output unit that outputs the generated response text, An information processing system equipped with the following features.

2. A drive system determination unit for determining the drive system of the aforementioned vehicle, The vehicle further comprises a vehicle type identification device for determining the type of vehicle, The aforementioned response section is, The system comprises multiple trained large-scale language models, each corresponding to one of the multiple driving methods, The large-scale language model corresponding to the identified drive system is input with the identified vehicle information to generate the response sentence. The information processing system according to claim 1.

3. The aforementioned vehicle is an automobile, The aforementioned driving data includes at least one of CAN information and navigation route information. The conversion unit uses a time-series-based model as an encoder and converts the driving data into text data using an embedded representation. The information processing system according to claim 1 or 2.

4. The acquisition unit acquires a plurality of the aforementioned driving data, The conversion unit generates a plurality of intermediate data corresponding to each of the plurality of driving data, The response unit generates the response statement relating to the comparative evaluation of the multiple driving data based on the multiple intermediate data. The information processing system according to claim 1 or 2.

5. An acquisition unit that acquires instruction information related to the vehicle's operation, A representation unit that uses a pre-trained large-scale language model to generate intermediate data, including text data having time-series information, based on the instruction information consisting of natural language, A conversion unit that converts the aforementioned intermediate data into the vehicle's driving data, An output unit that outputs the aforementioned driving data, An information processing system equipped with the following features.

Images