Information processing apparatus, program, and information processing system

Abstract

To provide an information processing device, a program, and an information processing system capable of improving accuracy of an answer to an inquiry content.SOLUTION: An information processing apparatus according to an embodiment includes an acquiring unit that acquires a question sentence including contents of a question input by a user, a generating unit that inputs a prompt generated based on the question sentence to a generation AI and causes the generation AI to generate a response sentence to the question sentence, a classifying unit that classifies each of the question sentence and the response sentence based on contents of the question sentence and the response sentence, an output unit that outputs the response sentence, and an output control unit that controls output of the output unit in accordance with a classification result of the classifying unit.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to an information processing device, a program, and an information processing system. [Background technology]

[0002] In recent years, natural language processing systems that utilize generative AI (Artificial Intelligence) such as large language models (LLMs) that can generate natural-sounding sentences have emerged.

[0003] The large-scale language model described above is designed to generate and output answers to user inquiries in a specified format. However, there is a possibility that the model may generate answers that are inconsistent with the query. For example, a quantized model that reduces the bit width of the data it handles can achieve high-speed text generation, but its inference accuracy is inferior to that of conventional models, and it may generate answers that are inconsistent with the query. Summary of the Invention [Problem to be solved by the invention]

[0004] The problem to be solved by the present invention is to provide an information processing device, a program, and an information processing system that can improve the accuracy of responses to inquiries. [Means for solving the problem]

[0005] The information processing device of the embodiment comprises an acquisition means for acquiring a query sentence including the query content entered by a user, a generation means for inputting a prompt generated based on the query sentence into a generation AI and causing the generation AI to generate an answer sentence to the query content, a classification means for classifying the query sentence and the answer sentence based on the contents of the query sentence and the answer sentence, an output means for outputting the answer sentence, and an output control means for controlling the output of the output means according to the classification result of the classification means. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a schematic diagram showing an example of a schematic configuration of an information processing system according to the first embodiment. [Figure 2] FIG. 2 is a block diagram illustrating an example of a hardware configuration of the interface device according to the first embodiment. [Figure 3] FIG. 3 is a block diagram showing an example of the hardware configuration of the text generator according to the first embodiment. [Figure 4] FIG. 4 is a block diagram illustrating an example of a functional configuration of the interface device according to the first embodiment. [Figure 5] FIG. 5 is a block diagram illustrating an example of a functional configuration of the text generator according to the first embodiment. [Figure 6] FIG. 6 is a sequence diagram showing an example of control processing performed by the interface device and the text generator according to the first embodiment. [Figure 7] FIG. 7 is a block diagram showing an example of a hardware configuration of an interface device according to the second embodiment. [Figure 8] FIG. 8 is a block diagram illustrating an example of a functional configuration of an interface device according to the second embodiment. [Figure 9] FIG. 9 is a block diagram illustrating an example of a functional configuration of a text generator according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0007] Hereinafter, embodiments of an information processing device, a program, and an information processing system will be described with reference to the drawings. Note that the present invention is not limited to the following embodiments.

[0008] (First embodiment) FIG. 1 is a schematic diagram showing an example of a schematic configuration of an information processing system 1 according to a first embodiment. As shown in FIG. 1, the information processing system 1 includes an interface device 10 and a text generator 20. The information processing system 1 is introduced, for example, by a store such as a retail shop that provides services on the web, and provides functions for responding to various inquiries from customers (hereinafter also referred to as users) who use the information processing system 1. The interface device 10 and the text generator 20 are connected to each other via wire or wirelessly so that they can communicate with each other.

[0009] The interface device 10 is, for example, a terminal device such as a PC used by a user of the information processing system 1. The interface device 10 exchanges various information with a text generation device 20, which will be described later. Note that the interface device 10 may also be a mobile terminal such as a smartphone, a tablet terminal, or the like.

[0010] Specifically, the interface device 10 acquires a query sentence including various inquiry contents input by a user. The interface device 10 transmits the acquired query sentence to the text generator 20. Then, when the interface device 10 receives an answer sentence from the text generator 20, it outputs the received answer sentence to the user. Here, the answer sentence is a text including a response to the matter (inquiry content, etc.) described in the query sentence.

[0011] The text generator 20 is an example of an information processing device. The text generator 20 generates an answer sentence based on a query sentence received from the interface device 10. The text generator 20 also acquires a classification result obtained by classifying the query sentence and the answer sentence based on a predetermined classification index. The text generator 20 then outputs the answer sentence to the interface device 10 based on the classification result.

[0012] In this embodiment, the text generator 20 is described as being implemented by a single device, but it may be implemented by multiple devices. Also, the interface device 10 and the text generator 20 may be integrated into one device.

[0013] Next, a description will be given of the hardware configuration of the interface device 10. Fig. 2 is a block diagram showing an example of the hardware configuration of the interface device 10 according to the first embodiment.

[0014] As shown in FIG. 2, the interface device 10 includes a CPU (Central Processing Unit) 101, which is an example of a processor, a ROM (Read Only Memory) 102, a RAM (Random Access Memory) 103, a memory unit 104, a display unit 105, an operation unit 106, an imaging unit 107, a speaker 108, a microphone 109, and a communication unit 110.

[0015] The CPU 101 is an example of a processor, and performs overall control of each unit of the interface device 10. The ROM 102 stores various programs. The RAM 103 is a workspace where programs and various data are developed.

[0016] Note that, as the processor of the control unit 100, another processor may be provided instead of or in addition to the CPU 101. As the other processor, various processors such as a GPU (Graphics Processing Unit), an NPU (Neural network Processing Unit), a DSP (Digital Signal Processor), or a dedicated arithmetic circuit realized by an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array) can be used as appropriate.

[0017] The memory unit 104 is a non-volatile memory such as a hard disk drive (HDD) or flash memory that retains stored information even when the power is turned off. The memory unit 104 includes a control program 121.

[0018] The control program 121 stores a control program for controlling the interface device 10. The CPU 101, ROM 102, RAM 103, and memory unit 104 are connected to one another via a bus 111. The CPU 101, ROM 102, and RAM 103 constitute a control unit 100 having a computer configuration. That is, the control unit 100 executes a control process of the interface device 10, which will be described later, by the CPU 101 operating in accordance with the control program 121 stored in the ROM 102 or memory unit 104 and loaded into the RAM 103.

[0019] The control unit 100 is connected to a display unit 105 , an operation unit 106 , an imaging unit 107 , a speaker 108 , a microphone 109 , and a communication unit 110 via a bus 111 .

[0020] The display unit 105 is a display device such as an LCD (Liquid Crystal Display), etc. Under the control of the CPU 101, the display unit 105 displays various types of information.

[0021] The operation unit 106 receives various inputs from the user. The operation unit 106 is, for example, a touch panel laminated on the display surface of the display unit 105. The operation unit 106 may also be an input device such as a keyboard or a pointing device.

[0022] The imaging unit 107 is an imaging device having an image sensor such as a CCD (Charge Coupled Device) or a CMOS (Complementary Metal Oxide Semiconductor), etc. The imaging unit 107 captures an image in response to an instruction from a user operating the interface device 10, for example.

[0023] The speaker 108 is an example of an audio output device, and outputs audio data input from the CPU 101 as audio.

[0024] The microphone 109 is an example of a voice input device. The microphone 109 acquires voice uttered by the user as voice data and outputs the acquired voice data to the CPU 101.

[0025] The communication unit 110 is a communication interface such as a LAN I / F (Interface), and is connected to a network Na. The communication unit 110 transmits and receives various information to and from, for example, the text generator 20, via the network Na. Under the control of the control unit 100, the communication unit 110 can also connect to a network such as the Internet or other information processing devices.

[0026] Next, a description will be given of the hardware configuration of the text generator 20. Fig. 3 is a block diagram showing an example of the hardware configuration of the text generator 20 according to the first embodiment.

[0027] As shown in FIG. 3, the text generator 20 includes a CPU 201, which is an example of a processor, a ROM 202, a RAM 203, a memory unit 204, a communication unit 205, and the like.

[0028] The CPU 201 comprehensively controls each unit of the text generator 20. The ROM 202 stores various programs. The RAM 203 is a workspace where programs and various data are expanded.

[0029] The memory unit 204 is a non-volatile memory such as a HDD or flash memory that retains stored information even when the power is turned off. The memory unit 204 includes a control program 221, a text generation model 222, and a text classification model 223.

[0030] Control program 221 stores a control program for controlling text generator 20. CPU 201, ROM 202, RAM 203, and memory unit 204 are connected to one another via bus 206. CPU 201, ROM 202, and RAM 203 constitute control unit 200, which is a computer configuration. That is, control unit 200 executes control processing of text generator 20, which will be described later, by CPU 201 operating in accordance with control program 221 stored in ROM 202 or memory unit 204 and loaded into RAM 203.

[0031] The text generation model 222 is a learning model for generating text. The text generation model 222 can be realized, for example, by using a decoder model among Transformer models. The text generation model 222 is an example of a generative AI that generates text. The text generation model 222 receives input of text (also called a prompt) generated based on a query from a user, and generates a response sentence according to the query sentence. Note that the text generation model 222 may use a known machine learning model or deep learning model such as GPT (Generative Pre-trained Transformer) or LlaMa.

[0032] The text generation model 222 is configured to generate and output a response sentence based on a prompt containing conditions such as an inquiry, in response to the input of the prompt.

[0033] The text generation model 222 may be fine-tuned to be specialized for the intended use of the information processing system 1. Fine-tuning may change the content of responses to input prompts or may change the wording of output sentences. For example, the text generation model 222 used in this embodiment may have learned specific wording, such as the tone of voice or endings of characters of a store that uses the information processing system 1.

[0034] In addition, the text generation model 222 may use a known natural language processing technique such as RAG (Retrieval-Augmented Generation) to accept input of a prompt generated based on a query statement and supplementary information obtained from an external storage device, etc., and generate an answer statement based on the query statement and supplementary information.

[0035] The text classification model 223 is a learning model for classifying text. The text classification model 223 can be realized using, for example, an encoder model among Transformer models. Note that the text classification model 223 may use a well-known machine learning model or deep learning model such as BERT (Bidirectional Encoder Representations from Transformers).

[0036] In this embodiment, when text such as a query or response is input, the text classification model 223 classifies the text based on its content. The classification method and number of classifications are not particularly important, and the text classification model 223 classifies the text based on a predetermined classification index. For example, when a query or response is input, the text classification model 223 outputs a classification result expressed in vector format calculated using a known natural language processing technique or the like.

[0037] The classification method of the text classification model 223 may be fine-tuned depending on the application of the information processing system 1. For example, the text classification model 223 may be fine-tuned to classify the genre indicated by the text of a query or answer. The text classification model 223 may also be fine-tuned to classify the emotion implied in the text of a query or answer.

[0038] The control unit 200 is connected to a communication unit 205 via a bus 206. The communication unit 205 is a communication interface such as a LAN I / F, and is connected to a network Na. The communication unit 205 transmits and receives various information to and from, for example, the interface device 10 via the network Na. Under the control of the control unit 200, the communication unit 205 can also be connected to a network such as the Internet or other information processing devices.

[0039] Next, a description will be given of the functional configuration of the interface device 10. Fig. 4 is a block diagram showing an example of the functional configuration of the interface device 10 according to the first embodiment.

[0040] As shown in FIG. 4, the control unit 100 of the interface device 10 includes an information input unit 1001, an information transmitting / receiving unit 1002, and a display control unit 1003 as functional components.

[0041] Specifically, the control unit 100 (CPU 101) of the interface device 10 realizes the above-mentioned functional configuration by executing the control program 121 stored in the memory unit 104. Note that in this embodiment, the above-mentioned functional configuration is a software configuration realized by cooperation between the processor of the interface device 10 and the program, but this is not limiting, and a hardware configuration in which part or all of the functional configuration is realized by a dedicated circuit or the like may also be used. Furthermore, the functional configuration of the interface device 10 is not limited to this.

[0042] The information input unit 1001 acquires information input by the user to the interface device 10. In other words, the information input unit 1001 accepts information input by the user. For example, the information input unit 1001 acquires a query sentence input by the user's operation via the operation unit 106.

[0043] The information transmitting and receiving unit 1002 is an example of a transmitting means. The information transmitting and receiving unit 1002 transmits and receives various information to and from the text generator 20 via the communication unit 110. For example, the information transmitting and receiving unit 1002 transmits a query sentence acquired by the information input unit 1001 to the text generator 20. The information transmitting and receiving unit 1002 also receives an answer sentence, which is an example of text, from the text generator 20.

[0044] The display control unit 1003 is an example of a display control means. The display control unit 1003 displays various information on the display unit 105. For example, the display control unit 1003 displays a screen for assisting in inputting a query on the display unit 105. The display control unit 1003 also displays a response sentence acquired by the information transmitting / receiving unit 1002, i.e., a response sentence to the query, on the display unit 105.

[0045] Next, a description will be given of the functional configuration of the text generator 20. Fig. 5 is a block diagram showing an example of the functional configuration of the text generator 20 according to the first embodiment.

[0046] As shown in FIG. 5, the control unit 200 of the text generator 20 has, as its functional configuration, an input receiving unit 2001, a first text classification unit 2002, a text generation unit 2003, a second text classification unit 2004, a comparison unit 2005, an output control unit 2006, and an output unit 2007.

[0047] Specifically, the control unit 200 (CPU 201) of the text generator 20 realizes the above-mentioned functional configuration by executing the control program 221 stored in the memory unit 204. Note that in this embodiment, the above-mentioned functional configuration is a software configuration realized by cooperation between the processor of the text generator 20 and the program, but this is not limiting, and a hardware configuration in which part or all of the functional configuration is realized by a dedicated circuit or the like may also be used. Furthermore, the functional configuration of the text generator 20 is not limited to this.

[0048] The input receiving unit 2001 is an example of an acquiring unit. The input receiving unit 2001 acquires a query statement including the content of the query entered by the user. Specifically, the input receiving unit 2001 acquires the query statement transmitted from the interface device 10 via the communication unit 205. The input receiving unit 2001 also temporarily stores (holds) the acquired query statement in the RAM 203.

[0049] The first text classification unit 2002 is an example of a classification means. The first text classification unit 2002 inputs a query sentence to the text classification model 223 and obtains the classification result output by the text classification model 223.

[0050] Specifically, the first text classification unit 2002 inputs the query sentence acquired by the input receiving unit 2001 to the text classification model 223. The first text classification unit 2002 also acquires the first classification result output by the text classification model 223. Here, the first classification result is an example of a classification result. The first text classification unit 2002 also temporarily stores (holds) the acquired first classification result in the RAM 203.

[0051] The text generator 2003 is an example of a generating means. The text generator 2003 generates a prompt based on a query sentence, and inputs the prompt into the text generation model 222 to generate a response sentence to the query content.

[0052] Specifically, the text generator 2003 generates a prompt including the query sentence acquired by the input receiver 2001. The text generator 2003 inputs the generated prompt to the text generation model 222. The text generator 2003 then acquires the answer sentence output by the text generation model 222. The text generator 2003 also temporarily stores (holds) the acquired answer sentence in the RAM 203.

[0053] Here, the prompt includes an instruction specifying the format and details of the answer to be generated. For example, the prompt preferably includes specific instructions such as "output an answer that is in line with the gist of the inquiry." Furthermore, the prompt may be configured to switch from multiple templates depending on the language of the inquiry, etc.

[0054] The second text classification unit 2004 is an example of a classification means. The second text classification unit 2004 inputs the answer sentence to the text classification model 223 and obtains the classification result output by the text classification model 223.

[0055] Specifically, the second text classification unit 2004 inputs the answer sentence generated by the text generation unit 2003 to the text classification model 223. In addition, the second text classification unit 2004 acquires the second classification result output by the text classification model 223. Here, the second classification result is an example of the classification result.

[0056] The comparison unit 2005 compares the first classification result obtained by the first text classification unit 2002 with the second classification result obtained by the second text classification unit 2004 .

[0057] Specifically, the comparison unit 2005 compares the first classification result with the second classification result and calculates the similarity between the classifications indicated by both classification results. The comparison unit 2005 then acquires the calculated similarity as a determination result. Here, as a method for calculating the similarity, for example, the cosine similarity between the first classification result and the second classification result is calculated. The comparison unit 2005 then outputs the acquired determination result to the output control unit 2006, which will be described later.

[0058] The comparison unit 2005 may output to the output control unit 2006 a determination result as to whether or not the classifications of the first classification result (inquiry sentence) and the second classification result (answer sentence) match based on the calculated similarity.

[0059] The output control unit 2006 is an example of an output control means. The output control unit 2006 controls the output of the answer sentence in accordance with the determination result obtained from the comparison unit 2005.

[0060] Specifically, if the similarity of the judgment result obtained from the comparison unit 2005 is equal to or greater than a predetermined threshold, the output control unit 2006 obtains the answer sentence generated by the text generation unit 2003 and causes the output unit 2007 to output the answer sentence.

[0061] In addition, if the similarity of the judgment result obtained from the comparison unit 2005 is less than a predetermined threshold, the output control unit 2006 discards the answer sentence generated by the text generation unit 2003 and causes the text generation unit 2003 to regenerate a new answer sentence.

[0062] For example, the output control unit 2006 regenerates an answer sentence by inputting the query sentence stored in the RAM 203 to the text generation unit 2003. When regenerating the answer sentence, the output control unit 2006 may instruct the text generation unit 2003 to change the conditions to be included in the prompt. Furthermore, the output control unit 2006 may specify the classification of the query sentence as a condition to be included in the prompt.

[0063] As described above, when the classifications of the first classification result (query sentence) and the second classification result (answer sentence) do not match or the similarity is less than the threshold, the output control unit 2006 causes the text generation unit 2003 to regenerate a new answer sentence. Furthermore, when the classifications of the first classification result and the second classification result match or the similarity is equal to or greater than the threshold, the output control unit 2006 causes the output unit 2007 to output the answer sentence. This makes it possible to prevent an answer sentence that is inappropriate to a user's inquiry from being output to the user, thereby improving the accuracy of the answer to the inquiry.

[0064] The output unit 2007 is an example of an output means. The output unit 2007 outputs various information to the interface device 10. For example, when the output unit 2007 receives an instruction to transmit an answer sentence from the output control unit 2006, the output unit 2007 transmits the answer sentence to the interface device 10 via the communication unit 205.

[0065] Next, the control processing of the interface device 10 and the text generator 20 will be described with reference to FIG.

[0066] Fig. 6 is a sequence diagram showing an example of control processing performed by the interface device 10 and the text generator 20 according to the first embodiment. The sequence diagram shown in Fig. 6 shows a processing example in which a response sentence to a query sentence inputted by the interface device 10 is generated and provided by the text generator 20.

[0067] First, the information input unit 1001 of the interface device 10 acquires a query sentence input by a user's operation via the operation unit 106 (step S101). Next, the information transmitting / receiving unit 1002 of the interface device 10 transmits the input query sentence to the text generator 20 (step S102).

[0068] The input receiving unit 2001 of the text generator 20 acquires the query sentence sent from the interface device 10 (step S103) and stores the acquired query sentence in the RAM 203 (step S104). Next, the first text classification unit 2002 of the text generator 20 inputs the query sentence to the text classification model 223 (step S105) and acquires the first classification result output by the text classification model 223 (step S106). Then, the first text classification unit 2002 stores the acquired first classification result in the RAM 203 (step S107).

[0069] Furthermore, the text generator 2003 of the text generator 20 generates a prompt including the query sentence based on the query sentence (step S108).

[0070] Next, the text generator 2003 of the text generator 20 inputs the prompt to the text generation model 222 (step S109) to obtain the answer sentence output by the text generation model 222 (step S110). Then, the text generator 2003 stores the obtained answer sentence in the RAM 203 (step S111).

[0071] Next, the second text classification unit 2004 of the text generator 20 inputs the answer sentence generated by the text generation unit 2003 to the text classification model 223 (step S112). Next, the second text classification unit 2004 acquires the second classification result output by the text classification model 223 (step S113).

[0072] Next, the comparison unit 2005 of the text generator 20 compares the first classification result acquired by the first text classification unit 2002 with the second classification result acquired by the second text classification unit 2004, and acquires the similarity between the two classification results as a determination result (step S114).The comparison unit 2005 then outputs the acquired determination result to the output control unit 2006.

[0073] If the similarity of the determination result is less than a predetermined threshold, the output control unit 2006 of the text generator 20 discards the answer sentence generated by the text generator 2003 and causes the text generator 2003 to regenerate a new answer sentence (step S115).

[0074] If the similarity of the determination result is equal to or greater than a predetermined threshold, the output control unit 2006 acquires the answer sentence generated by the text generation unit 2003 and causes the output unit 2007 to output the answer sentence (step S116). Next, the output unit 2007 of the text generation device 20 transmits the answer sentence to the interface device 10 (step S117).

[0075] The information transmitting / receiving unit 1002 of the interface device 10 receives the answer sentence from the text generator 20 (step S118). Next, the display control unit 1003 of the interface device 10 displays the answer sentence on the display unit 105 (step S119).

[0076] As described above, the text generation device 20 (an example of an information processing device) of this embodiment comprises an acquisition means for acquiring a query sentence including the query content entered by a user, a generation means for inputting a prompt generated based on the query sentence into a generation AI and causing the generation AI to generate an answer sentence to the query content, a classification means for classifying the query sentence and the answer sentence based on their contents, an output means for outputting the answer sentence, and an output control means for controlling the output of the output means according to the classification result of the classification means.

[0077] As a result, the text generator 20 determines whether the classification of the query sentence indicating the content of the query entered by the user and the classification of the answer sentence corresponding to the query sentence match or are similar, and controls the regeneration of the answer sentence or its transmission to the interface device 10 based on the determination result. If the classifications of the query sentence and the answer sentence do not match or are not similar, the text generator 20 regenerates a new answer sentence, and if the classifications of the query sentence and the answer sentence match or are similar, it outputs the answer sentence. Therefore, if an answer sentence that is inappropriate to the content of the user's query is generated, it is possible to prevent the answer sentence from being output to the user, thereby improving the accuracy of the answer to the content of the query.

[0078] The above-described embodiments can be modified as needed by changing some of the configurations or functions of the above-described devices. Therefore, below, several modifications of the above-described embodiments will be described as other embodiments. Below, differences from the above-described embodiments will be mainly described, and components that are common to the contents already described will be given the same reference numerals, and detailed descriptions will be omitted as appropriate. The modifications described below may be implemented individually or in combination as needed.

[0079] (Second embodiment) Next, a second embodiment will be described. For example, when a query is short, in other words, when the query contains only a small amount of information, the answer sentences generated by the text generator 20 may all be similar even when they are regenerated.

[0080] In such cases, it may be possible to generate a variety of replies by including additional information in the query. For example, by including information related to the user (hereinafter also referred to as related information) in the query, it is possible to generate a variety of replies and to generate replies that the user desires.

[0081] Therefore, in this embodiment, a configuration will be described in which a reply sentence can be generated based on a query sentence input by a user and related information.

[0082] In the second embodiment, the interface device 10 acquires sensor data relating to the user or the environment around the interface device 10 using various sensors 113 (see FIG. 7 ), and transmits the sensor data to the text generator 20. Then, upon receiving the sensor data from the interface device 10, the text generator 20 generates a prompt based on the query sentence and the state indicated by the sensor data.

[0083] First, a description will be given of the hardware configuration of the interface device 10. Fig. 7 is a block diagram showing an example of the hardware configuration of the interface device 10 according to the second embodiment.

[0084] As shown in FIG. 7, the interface device 10 includes the above-mentioned CPU 101, ROM 102, RAM 103, memory unit 104, display unit 105, operation unit 106, imaging unit 107, speaker 108, microphone 109 and communication unit 110, as well as a device interface 112 and a sensor 113.

[0085] The device interface 112 acquires sensor data from the sensor 113. The sensor data is an example of related information. If the sensor 113 outputs an analog value, the device interface 112 includes a signal processing circuit and an A / D converter. If the sensor 113 has a communication function and transmits a measurement value to the interface device 10 as digital data, the device interface 112 includes a communication interface capable of wired or wireless communication with the sensor 113. The sensor data acquired by the device interface 112 is transmitted to the control unit 100.

[0086] The sensor 113 is a sensor device that senses the surrounding environment. Here, the surrounding environment in this embodiment refers to the environment around the user or the interface device 10 to which the answer sentence is provided. Note that the surroundings of the user may include not only the vicinity of the user but also the range of the user and the area where the user is located. For example, the surrounding environment refers to the temperature, humidity, wind speed, weather, etc. around the user or the interface device 10. Note that the surrounding environment may also include other elements.

[0087] In this embodiment, the sensor 113 is provided in, for example, the interface device 10, and measures data related to the environment around the installation location of the sensor 113. The sensor 113 is, for example, a temperature sensor, a humidity sensor, a barometric pressure sensor, an illuminance sensor, a human presence sensor, an ultrasonic sensor, or other various sensors. Note that the sensor 113 is not limited to the above-mentioned sensors. The various sensors 113 transmit measurement results as sensor data to the device interface 112. The sensor data is, for example, numerical data indicating the measurement results of the sensor 113, such as temperature and humidity. Note that the sensor data output from the various sensors 113 may be analog data or digital data. Hereinafter, the sensor data related to the surrounding environment measured by the various sensors 113 will also be referred to as "environmental information."

[0088] Next, the functional configuration of the interface device 10 will be described. Fig. 8 is a block diagram showing an example of the functional configuration of the interface device 10 according to the second embodiment. The interface device 10 functions as a sensor data acquisition unit 1004 in addition to an information input unit 1001, an information transmission / reception unit 1002, and a display control unit 1003, as a result of the CPU 101 operating in accordance with a control program stored in the ROM 102 or memory unit 104. Note that each of the above functional configurations may be realized by a hardware configuration such as a dedicated circuit mounted on the interface device 10.

[0089] The sensor data acquiring unit 1004 acquires sensor data from various sensors 113. Specifically, the sensor data acquiring unit 1004 acquires sensor data from the sensors 113 via the device interface 112. Note that the sensor data acquiring unit 1004 may acquire sensor data from multiple sensors 113.

[0090] Furthermore, the information transmitting / receiving unit 1002 of the second embodiment transmits the query sentence acquired by the information input unit 1001 and the sensor data acquired by the sensor data acquiring unit 1004 to the text generator 20.

[0091] Next, the functional configuration of the text generator 20 will be described. Fig. 9 is a block diagram showing an example of the functional configuration of the text generator 20 according to the second embodiment. The text generator 20 functions as an input receiving unit 2001, a first text classification unit 2002, a text generator 2003, a second text classification unit 2004, a comparison unit 2005, an output control unit 2006, an output unit 2007, and also as a sensor data verbalization unit 2008, as a result of the CPU 201 operating in accordance with a control program stored in the ROM 202 or the memory unit 204. Note that each of the above functional configurations may be realized by a hardware configuration such as a dedicated circuit mounted on the text generator 20.

[0092] The sensor data verbalization unit 2008 converts the state indicated by the sensor data obtained from the sensor that senses the user's surrounding environment into label text that qualitatively expresses the state.

[0093] Specifically, when the sensor data verbalization unit 2008 receives sensor data from the interface device 10, it acquires label text corresponding to the sensor data. Here, the label text is text that is set in advance for each type of sensor data and each value of the sensor data. For example, the label text corresponding to the temperature (sensor data) value measured by a temperature sensor, "15°C," is set to "chilly."

[0094] Furthermore, the text generator 2003 of the second embodiment generates a prompt based on the label text acquired by the sensor data verbalizer 2008 and the query sentence.

[0095] The control processing performed by the interface device 10 and the text generator 20 of the second embodiment will be described below with reference to the sequence diagram of Fig. 6. The following describes the processing contents that differ from those of the first embodiment.

[0096] First, the information transmitting / receiving unit 1002 of the interface device 10 transmits the query sentence and the sensor data acquired by the sensor data acquiring unit 1004 to the text generator 20 in step S102.

[0097] Meanwhile, in step S103, the input receiving unit 2001 of the text generator 20 receives the query statement and sensor data from the interface device 10. The sensor data verbalization unit 2008 of the text generator 20 acquires label text corresponding to the sensor data. Then, in step S108, the text generation unit 2003 of the text generator 20 generates a prompt based on the query statement and the label text.

[0098] In step S103, the input receiving unit 2001 of the text generator 20 may measure the length of the query sentence when receiving the query sentence from the interface device 10. If the measured length of the query sentence is less than a predetermined length, the input receiving unit 2001 may generate a prompt based on the query sentence and the label text, and if the measured length of the query sentence is equal to or greater than the predetermined length, the input receiving unit 2001 may generate a prompt based only on the query sentence.

[0099] Furthermore, the output control unit 2006 of the text generator 20 may store the number of times that an answer sentence has been regenerated in the RAM 203, and when the number of times that an answer sentence has been regenerated reaches a predetermined number or more, may instruct the text generator 2003 to generate a prompt including related information. In this case, the text generator 2003 generates an answer sentence based on the prompt including related information.

[0100] Furthermore, in step S102, the information transmitting / receiving unit 1002 of the interface device 10 may transmit an image of the user (hereinafter also referred to as a user image) captured via the imaging unit 107 to the text generator 20 in addition to (or instead of) the sensor data. In this case, the input receiving unit 2001 of the text generator 20 extracts user attribute information (hereinafter also referred to as a user attribute) from the user image received from the interface device 10. Here, the user attribute is an example of related information. Then, the sensor data verbalization unit 2008 of the interface device 10 acquires label text corresponding to the user attribute extracted by the input receiving unit 2001.

[0101] As described above, the text generation device 20 of this embodiment further includes a sensor data verbalization unit 2008 that converts the sensor data acquired by the sensor data acquisition unit 1004 of the interface device 10 into label text, and the text generation unit 2003 generates a prompt based on the label text and the query sentence.

[0102] This allows the text generator 20 to input a prompt containing information related to the user into the text generation model 222, thereby generating a reply that reflects the information related to the user. Furthermore, by including information related to the user in a query, it is possible to expect the generation of a reply that the user desires. Therefore, the accuracy of the reply to the query can be improved.

[0103] (Variation 1) In the above-described embodiment, the output unit 2007 of the text generator 20 transmits the answer sentence to the interface device 10. However, the present invention is not limited to this, and the output unit 2007 may transmit the first classification result and the second classification result in addition to the answer sentence to the interface device 10. Furthermore, the transmission destination is not limited to the interface device 10, and may be an external server (not shown).

[0104] As a result, for example, it is possible to statistically analyze inquiry sentences from users who use the information processing system 1 and response sentences corresponding to the inquiry sentences by using the first classification result and the second classification result.

[0105] (Variation 2) In the above embodiment, when a query or response sentence is input, the text classification model 223 outputs a classification result expressed in vector format calculated using a known natural language processing technique, etc. However, this is not limited to this. For example, the text classification model 223 may output the number of occurrences of a specific phrase included in the query or response sentence as the classification result.

[0106] In this case, the first classification result is the number of times a specific phrase appears in the query sentence, and the second classification result is the number of times a specific phrase appears in the answer sentence. Therefore, if the number of times a specific phrase appears in the first classification result and the second classification result does not match or if the number of times a specific phrase appears in the first classification result and the second classification result is less than a predetermined threshold, the comparison unit 2005 of the text generator 20 controls the text generation unit 2003 to regenerate the answer sentence. Furthermore, if the number of times a specific phrase appears in the first classification result and the second classification result matches or if the number of times a specific phrase appears in the first classification result and the second classification result is equal to or greater than a predetermined threshold, the comparison unit 2005 causes the output unit 2007 to output the answer sentence.

[0107] (Variation 3) In the above embodiment, the information input unit 1001 of the interface device 10 acquires a query sentence input by a user's operation from the operation unit 106, but this is not limiting. For example, the information input unit 1001 may acquire voice data input from the microphone 109. In this case, the control unit 100 of the interface device 10 converts the voice data into text data (query sentence) and transmits it to the text generator 20.

[0108] (Variation 4) In the above embodiment, the display control unit 1003 of the interface device 10 displays the answer sentence on the display unit 105, but this is not limiting. For example, the control unit 100 of the interface device 10 may convert text data (answer sentence) into audio data and output it from the speaker 108.

[0109] (Variation 5) In each of the above-described embodiments, the answer sentence output by the text generation model 222 is provided to the interface device 10, but this is not limiting, and the answer sentence output by the text generation model 222 may be edited before being provided to the interface device 10. For example, if there is a need to incorporate sales promotion of a specific product or a specific manufacturer into the product proposal, the text generation unit 2003 may add promotional text regarding sales promotion of the specific product or a specific manufacturer to the answer sentence output from the text generation model 222 by RAG.

[0110] The programs executed by the information processing system 1 of the embodiment and the modified example may be stored on a computer connected to a network such as the Internet and provided by being downloaded via the network. Also, the programs executed by the information processing system 1 of the embodiment and the modified example may be provided or distributed via a network such as the Internet.

[0111] The programs executed by each device in the above-described embodiments are provided in a state where they are pre-installed in a ROM, a storage unit, etc. The programs executed by each device in the above-described embodiments may be provided by being recorded in an installable or executable file format on a computer-readable recording medium such as a CD-ROM, a flexible disk (FD), a CD-R, or a digital versatile disk (DVD).

[0112] These novel embodiments and their modifications may be embodied in various other forms, and various omissions, substitutions, changes, and combinations may be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the inventions, and are also included in the inventions described in the claims and their equivalents. [Explanation of symbols]

[0113] 1. Information Processing Systems 10 Interface Device 20 Text Generator 1001 Information input section 1002 Information transmission and reception unit 1003 Display control unit 1004 Sensor data acquisition unit 2001 Input Reception Section 2002 First Text Classification Unit 2003 Text Generation Department 2004 Second Text Classification Unit 2005 Comparison Section 2006 Output control section 2007 Output section 2008 Sensor Data Verbalization Department [Prior art documents] [Patent documents]

[0114] [Patent Document 1] Patent No. 7334230

Claims

1. An acquisition means for acquiring a query statement including the content of the query entered by the user; a generating means for inputting a prompt generated based on the query sentence into a generating AI and causing the generating AI to generate a response sentence to the query content; a classification means for classifying each of the inquiry sentences and the answer sentences based on the contents of the inquiry sentences and the answer sentences; an output means for outputting the answer sentence; an output control means for controlling the output of the output means in accordance with the classification result of the classification means; An information processing device comprising:

2. The output control means If the categories of the inquiry sentence and the answer sentence match or the similarity of the categories is equal to or greater than a threshold, causing the output means to output the answer sentence; If the classifications of the query sentence and the answer sentence do not match or the similarity is less than a threshold, causing the generation means to regenerate the answer sentence. The information processing device according to claim 1 .

3. the output control means changes an instruction sentence relating to output of the answer sentence to be written in the prompt when the classifications of the query sentence and the answer sentence do not match or the similarity is less than a threshold. The information processing device according to claim 2 .

4. the acquiring means acquires related information related to the user; the generating means generates a prompt based on the query statement and the relationship information. The information processing device according to claim 1 .

5. The computer of the information processing device, An acquisition means for acquiring a query statement including the content of the query entered by the user; a generating means for inputting a prompt generated based on the query sentence into a generating AI and causing the generating AI to generate a response sentence to the query content; a classification means for classifying each of the inquiry sentences and the answer sentences based on the contents of the inquiry sentences and the answer sentences; an output means for outputting the answer sentence; an output control means for controlling the output of the output means in accordance with the classification result of the classification means; A program to function as a

6. An information processing system including an information processing device and a terminal device capable of communicating with the information processing device, The terminal device a transmitting means for transmitting an inquiry statement indicating the content of the inquiry input by the user to the information processing device; a display control means for acquiring an answer sentence from the information processing device and displaying the answer sentence on a display unit, The information processing device includes: An acquisition means for acquiring a query statement including the content of the query entered by the user; a generating means for inputting a prompt generated based on the query sentence into a generating AI and causing the generating AI to generate a response sentence to the query content; a classification means for classifying each of the inquiry sentences and the answer sentences based on the contents of the inquiry sentences and the answer sentences; an output means for outputting the answer sentence; an output control means for controlling the output of the output means in accordance with the classification result of the classification means; An information processing system comprising:

Images