Inquiry sentence generation device

Abstract

To provide an inquiry sentence generation device which generates an inquiry sentence related to administrative improvement of a facility, on the basis of moving image data in the facility.SOLUTION: An inquiry sentence generation device 10 has a storage unit 150, an acquisition part 160a, a first reception part 160b, an extraction part 160c, and a generation part 160d. The storage unit 150 stores multiple templates of an inquiry sentence to be input to a large scale language model. Each template has one or more blanks filled with a feature amount indicating a feature related to an attribute or a behavior of a user of a facility. The acquisition part 160a acquires moving image data in the facility. The first reception part 160b receives an instruction to select any one of the multiple templates. The extraction part 160c extracts from the moving image data, one or more feature amounts corresponding to the selected template. The generation part 160d fills the extracted feature amount in a corresponding blank in the selected template, for generating an output inquiry sentence.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a question generation device. [Background technology]

[0002] In recent years, generative AI such as large-scale language models (hereinafter sometimes referred to as "LLM") have been attracting attention. LLM generates and outputs answers to input questions. The questions input to LLM are also called prompts. Depending on the quality of the questions input to LLM, useful answers may not be obtained. In order to create good questions that can obtain useful answers from LLM, various information about the subject of the question must be included in the questions in detail, and it takes a lot of effort to analyze the survey target. In addition, since there is an upper limit on the number of characters (tokens) that can be input as a question, the creator of the question is also required to have the ability to summarize the sentences accurately. In other words, the quality of the question depends on the skill of the creator who creates the question. In recent years, various technologies have been proposed that enable appropriate questions to be created at low cost without requiring special skills, and one example of such a technology is the technology disclosed in Patent Document 1. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2017-27233 A Summary of the Invention [Problem to be solved by the invention]

[0004] In order to improve the management of retail stores and other stores, customer attributes or behaviors obtained from video data captured inside the store may be analyzed, and products to be displayed may be reviewed, the store layout may be improved, or sales promotion methods may be improved. In this case, LLM may be used as a means of analysis and a means of investigating improvement measures. However, there has been no technology to generate questions to be input into LLM from video data captured inside the store. [Means for solving the problem]

[0005] A mobile terminal according to a preferred embodiment of the present disclosure includes a first storage unit, an acquisition unit, a first reception unit, an extraction unit, and a generation unit. The first storage unit stores a plurality of templates, each of which has one or more blanks for completing a question sentence by filling the blanks with features representing attributes or behavioral features of users of the facility, and each template has a different type of feature. The acquisition unit acquires video data representing a video captured by a camera inside the facility. The first reception unit accepts an instruction to select one of the plurality of templates stored in the first storage unit. The extraction unit extracts one or more features corresponding to the template selected in response to the instruction accepted by the first reception unit by analyzing the video data acquired by the acquisition unit. The generation unit generates an output question sentence by filling the corresponding blanks in the selected template with the one or more features extracted by the extraction unit. Effect of the Invention

[0006] According to the present disclosure, questions to be input to the LLM are generated based on a template selected by a user considering a facility management improvement measure and various features extracted from video data within the facility, thereby reducing the user's workload in creating questions. Furthermore, according to the present disclosure, questions with a certain level of quality are generated regardless of the user's skill. [Brief description of the drawings]

[0007] [Figure 1] 1 is a diagram illustrating an example of a configuration of a question generation device 10 according to an embodiment of the present disclosure. [Diagram 2] FIG. 13 is a diagram illustrating an example of a template. [Diagram 3] FIG. 13 is a diagram showing an example of a question generated according to a template. [Figure 4] 11 is a flowchart showing the flow of a generation method executed by the processing device 160 of the question generation device 10 in accordance with the program PR1. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] (A. Embodiment) FIG. 1 is a diagram illustrating a configuration example of a question generation device 10 according to an embodiment of the present disclosure. The question generation device 10 is a device that generates a question to be input to an LLM to receive a proposal for improving the management of a facility, based on video data obtained by capturing images of the inside of the facility. An existing LLM is used as appropriate. The question generation device 10 in this embodiment is a personal computer, but may also be a mobile terminal such as a tablet terminal or a smartphone. As shown in FIG. 1, a camera 20 is connected to the question generation device 10 via a wired or wireless connection.

[0009] Camera 20 is installed in a facility that is the target of business improvement (hereinafter, target facility). Camera 20 captures video showing the state of the inside of the target facility. In this embodiment, the target facility is, for example, a retail store such as a convenience store. In this embodiment, the video captured by camera 20 shows customer behavior in the store, such as customers moving around the store, customers stopping in front of display shelves, or customers picking up displayed products. Camera 20 transmits video data showing the captured video to question sentence generation device 10.

[0010] 1, the question generation device 10 includes an input device 110, a display device 120, a communication device 130, a storage device 150, a processing device 160, and a bus 190. The input device 110, the display device 120, the communication device 130, and the storage device 150 are each connected to the processing device 160 by the bus 190 for communicating information. The bus 190 may be configured using a single bus, or may be configured using different buses between the devices.

[0011] The input device 110 includes a pointing device such as a mouse, and a keyboard having a plurality of operation keys. The input device 110 may include a touch panel in addition to the pointing device and the keyboard. The input device 110 accepts user operations on the question generation device 10. A user of the question generation device 10 is, for example, a person involved in improving the management of the target facility, such as an operations manager of the target facility. The display device 120 is, for example, a liquid crystal display. The display device 120 displays various information on the display surface under the control of the processing device 160. When the input device 110 includes a touch panel, the touch panel may be laminated on the display surface of the display device 120.

[0012] The communication device 130 is hardware (transmission / reception device) for communicating with other devices via a communication line in a wired or wireless manner. The communication device 130 is also called, for example, a network device, a network controller, a network card, or a communication module. Examples of other devices that communicate with the communication device 130 via a communication line include the camera 20 and a server device that provides an LLM utilization service. The communication device 130 receives video data transmitted from the camera 20. The communication device 130 supplies the received video data to the processing device 160. The communication device 130 also receives answer data representing an answer to the question (a proposal for improving the management of the facility) by transmitting question data representing a question generated by the processing device 160 to the server device via the communication line. The communication device 130 supplies the received answer data to the processing device 160.

[0013] The storage device 150 is a recording medium readable by the processing device 160. The storage device 150 includes, for example, a nonvolatile memory and a volatile memory. The nonvolatile memory includes, for example, a ROM (Read Only Memory), an EPROM (Erasable Programmable Read Only Memory), or an EEPROM (Electrically Erasable Programmable Read Only Memory). The volatile memory is, for example, a RAM (Random Access Memory). The nonvolatile memory of the storage device 150 stores a program PR1 that causes the processing device 160 to execute a method for generating a question sentence specific to the present disclosure. The volatile memory of the storage device 150 is used as a work area when executing the program PR1.

[0014] The storage device 150 also stores a plurality of templates D1 of questions to be input to the LLM in order to receive proposals for improving the management of the facility. The template D1 is a prototype of the questions. The template D1 has one or a plurality of blank spaces in which questions are completed by being filled with features representing the attributes or behavioral characteristics of users of the target facility (in this embodiment, customers who visited the target facility). Specific examples of the attributes of the facility users include the age and sex of the users. Specific examples of the behavioral characteristics of the facility users include the day of the week and time of the visit, the number of visitors representing whether or not the visitor came with other users, the area where the user stayed, the stay rate in the area, and the movement route within the store. The stay rate in an area is the ratio of the time the user stayed in the area to the time the user stayed in the store. The type of features corresponding to each template is different for each template. The storage device 150 is an example of a first storage unit that stores a plurality of templates of questions to be input to the LLM in order to receive proposals for improving the management of the facility. Note that a template having the same types of features as template D1 but with a different description order of the features or a different sentence expression itself may be stored in storage device 150 as a template of a variation pattern of template D1. Furthermore, the templates stored in storage device 150 may be added, modified, or deleted by a user's operation on input device 110.

[0015] FIG. 2 is a diagram showing an example of the template D1. The template D1 shown in FIG. 2 is a prototype of a question for receiving a proposal to improve the purchase rate of products displayed in a specific area in a target facility. The template D1 shown in FIG. 2 has (blank B1) to (blank B8). (Blank B1) corresponds to the age of the user. (Blank B2) corresponds to the gender of the user. (Blank B3) corresponds to the day of the week. (Blank B4) corresponds to the time period. (Blank B5) corresponds to the number of customers. (Blank B6) and (Blank B8) correspond to the area. (Blank B7) corresponds to the stay rate. Although the details will be described later, each of (Blank B1) to (Blank B8) in the template D1 is filled with a feature extracted from video data to complete the question. Although detailed illustration is omitted in FIG. 2, the template D1 includes data that associates the blank with the type of feature corresponding to the blank.

[0016] The storage device 150 also stores in advance store data D2 representing at least one of the store's location, sales floor area, and business conditions, such as business hours. Although details will be described later, the store data D2 is used to determine the appropriateness of an answer (i.e., a proposal for business improvement) obtained by inputting a question (hereinafter, output question) generated by filling in the blanks of the above-mentioned template with features into the LLM from a practical viewpoint such as the store's location. Note that the store data D2 is not limited to the data representing the store's location, sales floor area, and business conditions. Specifically, the store data D2 may include data representing items that can be changed at the discretion of the store's operations manager, such as at least one of the types and the number of each type of handled merchandise, the shelf allocation of handled merchandise, and the arrangement of shelves in the sales floor. The storage device 150 is an example of a second storage unit that stores store data. Furthermore, although details will be described later, the storage device 150 functions as a third storage unit that stores question data representing output questions that have been given good evaluations by users for proposals obtained by inputting the questions into the LLM, among the output questions.

[0017] The processing device 160 includes one or more central processing units (CPUs). The one or more CPUs are an example of one or more processors. Each of the processor and the CPU is an example of a computer. When the processing device 160 is instructed to execute the program PR1 by an operation on the input device 110, the processing device 160 reads the program PR1 from the non-volatile memory to the volatile memory. Then, the processing device 160 executes the program PR1 read to the volatile memory. The processing device 160 operating according to the program PR1 functions as the acquisition unit 160a, the first reception unit 160b, the extraction unit 160c, the generation unit 160d, the determination unit 160e, the discarding unit 160f, and the second reception unit 160g shown in FIG. 1. 1 are software modules realized by operating the processing device 160 in accordance with software. The functions of the acquiring unit 160a, the first accepting unit 160b, the extracting unit 160c, the generating unit 160d, the determining unit 160e, the discarding unit 160f, and the second accepting unit 160g are as follows.

[0018] Acquiring unit 160a communicates with camera 20 using communication device 130 to acquire video data representing a video captured by camera 20 inside the facility.

[0019] When an operation for designating one of the multiple templates D1 stored in the storage device 150 is performed on the input device 110, the first reception unit 160b receives the instruction. More specifically, the first reception unit 160b prompts the user to select one of the multiple templates D1 by displaying on the display device 120 an image of a UI screen in which identification information (e.g., character strings representing the names of the templates) uniquely identifying each of the multiple templates D1 stored in the storage device 150 is arranged in a list format. The user who has viewed this UI screen designates one of the multiple templates D1 stored in the storage device 150 by performing an operation such as clicking on the identification information of the template D1 to be selected on the input device 110. The first reception unit 160b reads out the template designated to be selected by the above operation from the storage device 150. The first reception unit 160b may read from the storage device 150 a template that matches the word (search keyword) input by the user through an operation on the input device 110 or the free-form text.

[0020] The extraction unit 160c extracts one or more feature amounts corresponding to the template read from the storage device 150 by the first reception unit 160b by analyzing the video data acquired by the acquisition unit 160a. It should be noted that existing techniques may be appropriately used to extract various feature amounts from video data. For example, when the template D1 shown in FIG. 2 is read from the storage device 150 by the first reception unit 160b, the extraction unit 160c extracts feature amounts corresponding to each of (blank B1) to (blank B8), that is, the age of the user appearing in the video represented by the video data, the sex of the user, the day of the week when the video was taken, the time period when the video was taken, the number of customers who visited the store, the area appearing in the video, and the stay rate of the user appearing in the video, from the video data.

[0021] The generating unit 160d generates an output question by filling one or more feature amounts extracted from the video data by the extracting unit 160c into corresponding blanks in the template read by the first receiving unit 160b from the storage device 150. For example, assume that the template D1 shown in FIG. 2 is read from the storage device 150 by the first receiving unit 160b, and the extracting unit 160c extracts "30s" as a feature amount corresponding to (blank B1), "male" as a feature amount corresponding to (blank B2), "Monday" as a feature amount corresponding to (blank B3), "12 o'clock" as a feature amount corresponding to (blank B4), "alone" as a feature amount corresponding to (blank B5), "Area A" as a feature amount corresponding to (blank B6) and (blank B8), and "35%" as a feature amount corresponding to (blank B8). In this case, the generating unit 160d generates question sentence data representing the output question sentence P1 shown in Fig. 3 by filling in (blank space B1) to (blank space B8) in the template D1 shown in Fig. 2 with the feature amount extracted by the extracting unit 160c. Then, the generating unit 160d transmits the generated question sentence data to the LLM using the communication device 130, thereby acquiring the answer data.

[0022] When some of all the features to be filled in each of (blank B1) to (blank B8) are not extracted by the extraction unit 160c, the generation unit 160d may notify the user that there is a blank from which a corresponding feature has not been extracted, and may further execute a predetermined process among the following modes A, B, and C. In mode A, the generation unit 160d generates question sentence data by filling the corresponding blank with a predetermined word. When mode A is adopted, it is necessary to predefine a default word to be used when a feature has not been extracted for each of (blank B1) to (blank B8). If there is a blank from which a corresponding feature has not been extracted, the question sentence cannot be completed, and even if a low-quality question sentence including a blank is input to the LLM, there is a possibility that only a low-accuracy answer will be obtained. According to mode A, even if there is a blank from which a corresponding feature has not been extracted, the question sentence can be completed, and a deterioration in the quality of the question sentence can be prevented. In mode B, the generation unit 160d generates a question sentence by deleting a part related to the corresponding blank from the template. According to the mode B, as in the mode A, even if there is a blank for which a corresponding feature has not been extracted, the question can be completed, and the quality of the question can be prevented from deteriorating. In the mode C, the generating unit 160d stops the process of generating the question itself and prompts the user to specify another template. When there is a blank for which a corresponding feature has not been extracted, which process among the mode A, the mode B, and the mode C is to be adopted may be set in advance, or the user may be prompted to specify the process in response to a notification that there is a blank for which a corresponding feature has not been extracted. When the user specifies the mode A, the user may set a default word corresponding to the corresponding blank by operating the input device 110.

[0023] The determination unit 160e reads out store data from the storage device 150. The determination unit 160e determines the appropriateness of an answer to an output question generated by the generation unit 160d, based on the location, sales floor area, or business status represented by the store data read out from the storage device 150. The determination unit 160e determines, for example, an answer proposing a change in location (i.e., relocation of the store), an expansion or reduction in size of the store, or a change in business hours to be an inappropriate answer. This is because a change in location, an expansion or reduction in size of the store, or a change in business hours is often not realistically adoptable as a measure for improving management.

[0024] The discarding unit 160f discards answer data representing answers that have been determined to be inappropriate by the determining unit 160e.

[0025] The second receiving unit 160g causes the answer determined to be appropriate by the determining unit 160e to be displayed on the display device 120. Then, the second receiving unit 160g receives an input of evaluation data representing a user's evaluation of the quality of the answer displayed on the display device 120. The second receiving unit 160g writes question sentence data representing the output question sentence that led to the answer for which evaluation data indicating a good evaluation was received, in association with the evaluation data, into the storage device 150.

[0026] Moreover, the processing device 160 operating according to the program PR1 executes a generation method specific to the present disclosure. Fig. 4 is a flowchart showing the flow of processing in this generation method. As shown in Fig. 4, this generation method includes an acquisition process SA110, a first reception process SA120, an extraction process SA130, a generation process SA140, a determination process SA150, a discard process SA160, and a second reception process SA170. The processing contents of each process of the acquisition process SA110, the first reception process SA120, the extraction process SA130, the generation process SA140, the determination process SA150, the discard process SA160, and the second reception process SA170 are as follows.

[0027] In the acquisition process SA110, the processing device 160 functions as an acquisition unit 160a. In the acquisition process SA110, the processing device 160 communicates with the camera 20 using the communication device 130 to acquire video data representing a video captured by the camera 20 inside the facility.

[0028] In the first reception process SA120, the processing device 160 functions as a first reception unit 160b. In the first reception process SA120, the processing device 160 causes the display device 120 to display a UI screen that prompts the user to select one template from among a plurality of templates stored in the storage device 150. Then, in the first reception process SA120, the processing device 160 reads out from the storage device 150 the template that has been instructed to be selected by an operation on the UI screen.

[0029] In the extraction process SA130, the processing device 160 functions as an extraction unit 160c. In the extraction process SA130, the processing device 160 extracts one or more feature amounts corresponding to the template read from the storage device 150 in the first reception process SA120 by analyzing the video data acquired in the acquisition process SA110.

[0030] In the generation process SA140, the processing device 160 functions as a generating unit 160d. In the generation process SA140, the processing device 160 generates an output question by filling in corresponding blanks in a template read from the storage device 150 in the first reception process SA120 with one or more feature amounts extracted from the video data in the extraction process SA130. Then, the processing device 160 transmits question data representing the generated output question to the LLM using the communication device 130, thereby acquiring answer data.

[0031] In the determination process SA150, the processing device 160 functions as a determination unit 160e. In the determination process SA150, the processing device 160 reads out the store data D2 from the storage device 150. In the determination process SA150, the processing device 160 determines whether the answer to the output question generated in the generation process SA140 is appropriate based on the location, sales floor area, or business status represented by the store data D2 read out from the storage device 150. If the answer is determined to be inappropriate, the determination result of the determination process SA150 is "No". If the determination result of the determination process SA150 is "No", the processing device 160 executes the discarding process SA160. If the answer to the output question generated in the generation process SA140 is determined to be appropriate, the determination result of the determination process SA150 is "Yes". If the determination result of the determination process SA150 is "Yes", the processing device 160 executes the second reception process SA170.

[0032] In the discard process SA160, the processing device 160 functions as a discarding unit 160f. In the discard process SA160, the processing device 160 discards answer data representing answers determined to be inappropriate in the determination process SA150. Upon completing the execution of the discard process SA160, the processing device 160 executes the first reception process SA120 again, and prompts the user to reselect a template.

[0033] In the second reception process SA170, the processing device 160 functions as a second reception unit 160g. In the second reception process SA170, the processing device 160 causes the display device 120 to display the answer determined to be appropriate in the determination process SA150. Then, the processing device 160 receives an input of evaluation data indicating a user's evaluation of the quality of the answer displayed on the display device 120. The processing device 160 writes question sentence data indicating an output question sentence that has led to the answer for which evaluation data indicating a good evaluation has been received, in association with the evaluation data, into the storage device 150.

[0034] As described above, according to this embodiment, questions are automatically generated based on a template selected by the user and various feature amounts extracted from video data obtained by capturing images of the interior of the target facility. Therefore, regardless of the skill of the user who analyzes data and creates questions for business improvement, the quality of the questions is kept constant and the user's workload is reduced.

[0035] Furthermore, according to this embodiment, if an answer to a question generated based on a template selected by the user and various features extracted from video data obtained by capturing footage of the inside of the target facility is inappropriate considering the location, sales floor area, or business status of the target facility, the answer is discarded without being presented to the user, thereby avoiding the proposal of management improvement measures that are inappropriate considering the location, sales floor area, or business status of the target facility.

[0036] Furthermore, according to this embodiment, among the proposals obtained by inputting the output question sentence to the LLM, the output question sentence that has led to a proposal that has been given a good evaluation is associated with the evaluation data representing the evaluation and stored in the storage device 150, so that it becomes possible to judge the quality of each template through the evaluation of the proposal. The evaluation result may also be displayed when the first reception process SA120 is executed. By displaying the evaluation result, it is possible to assist the user in making a decision when selecting a template.

[0037] (B: Transformation) The present disclosure is not limited to the above-described exemplary embodiments. Specific modified aspects are as follows. Two or more aspects selected from the following examples may be combined. (B-1: Variation 1) In the above embodiment, the facility is a store such as a retail store, but it may be a public facility such as a library or a community center.

[0038] (B-2: Variation 2) In the above embodiment, the acquiring unit 160a, the first accepting unit 160b, the extracting unit 160c, the generating unit 160d, the judging unit 160e, the discarding unit 160f, and the second accepting unit 160g are software modules. However, any one, any two, any three, any four, any five, any six, or all of the acquiring unit 160a, the first accepting unit 160b, the extracting unit 160c, the generating unit 160d, the judging unit 160e, the discarding unit 160f, and the second accepting unit 160g may be hardware modules. Specific examples of hardware modules include a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), and an FPGA (Field Programmable Gate Array). The same effects as those of the above embodiment can be achieved even if any one, any two, any three, any four, any five, any six, or all of the acquisition unit 160a, the first reception unit 160b, the extraction unit 160c, the generation unit 160d, the judgment unit 160e, the discarding unit 160f, and the second reception unit 160g are hardware modules.

[0039] (B-3: Variation 3) Determining whether the answer is appropriate based on the store's location, sales floor area, and business status is not necessarily required, and the determining unit 160e and the discarding unit 160f may be omitted. In addition, a user's evaluation of the appropriateness of the answer obtained by inputting the output question sentence to the LLM is not necessarily required, and the second receiving unit 160g may be omitted. In an aspect in which the second receiving unit 160g is omitted, the storage device 150 does not need to function as a third storage unit.

[0040] (B-4: Variation 4) In the above embodiment, the program PR1 is stored in the storage device 150 of the question generation device 10, but the program PR1 may be manufactured or sold separately. When selling the program PR1, the program PR1 may be provided to a purchaser by writing the program PR1 to a computer-readable recording medium such as a flash ROM and distributing it, or by downloading it via a telecommunications line.

[0041] (C:Other) (1) In the above-described embodiment, ROM, RAM, etc. are given as examples of storage device 150, but storage device 150 may also be a flexible disk, a magneto-optical disk (e.g., a compact disk, a digital versatile disk, a Blu-ray (registered trademark) disc), a smart card, a flash memory device (e.g., a card, a stick, a key drive), a CD-ROM (Compact Disc-ROM), a register, a removable disk, a hard disk, a floppy (registered trademark) disk, a magnetic strip, a database, a server, or any other suitable storage medium.

[0042] (2) In the above-described embodiments, the described information, signals, etc. may be represented using any of a variety of different technologies. For example, data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.

[0043] (3) In the above-described embodiment, the input / output information, etc. may be stored in a specific location (e.g., memory) or may be managed using a management table. The input / output information, etc. may be overwritten, updated, or added. The output information, etc. may be deleted. The input information, etc. may be transmitted to another device.

[0044] (4) In the above-described embodiments, the determination may be made based on a value represented by 1 bit (0 or 1), a Boolean (true or false) value, or a numerical comparison (e.g., comparison with a predetermined value).

[0045] (5) The order of the process steps, sequences, flow charts, etc. illustrated in the above-described embodiments may be changed without causing any inconsistency. For example, the methods described in this disclosure present elements of various steps using an example order, and are not limited to the particular order presented.

[0046] (6) Each function illustrated in FIG. 1 is realized by any combination of at least one of hardware and software. Furthermore, the method of realizing each functional block is not particularly limited. That is, each functional block may be realized by using one device that is physically or logically combined, or may be realized by using two or more devices that are physically or logically separated and directly or indirectly connected (for example, by using wires, wirelessly, etc.). The functional block may be realized by combining the one device or the multiple devices with software.

[0047] (7) The programs exemplified in the above-described embodiments should be broadly interpreted to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, functions, etc., regardless of whether software is called software, firmware, middleware, microcode, hardware description language, or by other names.

[0048] Additionally, software, instructions, information, etc. may be transmitted or received over a transmission medium. For example, if the software is transmitted from a website, server, or other remote source using wired technologies (such as coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL)), and / or wireless technologies (such as infrared, microwave, etc.), then these wired and / or wireless technologies are included within the definition of transmission media.

[0049] (8) In the above embodiments, the terms "system" and "network" are used interchangeably.

[0050] (9) The information, parameters, etc. described in this disclosure may be expressed using absolute values, may be expressed using relative values ​​from a predetermined value, or may be expressed using other corresponding information.

[0051] (10) In the above-described embodiments, the mobile terminal includes a mobile station (MS). A mobile station may be referred to by those skilled in the art as a subscriber station, a mobile unit, a subscriber unit, a wireless unit, a remote unit, a mobile device, a wireless device, a wireless communication device, a remote device, a mobile subscriber station, an access terminal, a mobile terminal, a wireless terminal, a remote terminal, a handset, a user agent, a mobile client, a client, or some other suitable term. In this disclosure, the terms "mobile station", "user terminal", "user equipment (UE)", "terminal", etc. may be used interchangeably.

[0052] (11) In the above-mentioned embodiments, the terms "connected" and "coupled" or any variation thereof refer to any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are "connected" or "coupled" to each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, "connected" may be read as "accessed". As used in this disclosure, two elements may be considered to be "connected" or "coupled" to each other using at least one of one or more wires, cables, and printed electrical connections, as well as electromagnetic energy having wavelengths in the radio frequency range, microwave range, and light (both visible and invisible) range, as some non-limiting and non-exhaustive examples.

[0053] (12) In the above embodiments, the phrase "based on" does not mean "based only on," unless otherwise specified. In other words, the phrase "based on" means both "based only on" and "based at least on."

[0054] (13) As used in this disclosure, the terms "determining" and "determining" may encompass a wide variety of actions. "Determining" and "determining" may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, search, inquiry (e.g., searching in a table, database, or other data structure), ascertaining, and the like. Also, "determining" and "determining" may include receiving (e.g., receiving information), transmitting (e.g., sending information), input, output, accessing (e.g., accessing data in a memory), and the like. In addition, "judgment" and "decision" can include considering resolving, selecting, choosing, establishing, comparing, etc., to be a "judgment" or "decision." In other words, "judgment" and "decision" can include considering some action to be a "judgment" or "decision." In addition, "judgment (decision)" can be interpreted as "assuming," "expecting," "considering," etc.

[0055] (14) In the above embodiments, when "include," "including," and variations thereof are used, these terms are intended to be inclusive, similar to the term "comprising." Furthermore, the term "or" used in this disclosure is not intended to be an exclusive or.

[0056] (15) In this disclosure, where articles have been added by translation, such as a, an, and the in English, the disclosure may include that the nouns following these articles are in the plural.

[0057] (16) In this disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "combined" may also be interpreted in the same way as "different."

[0058] (17) Each aspect / embodiment described in this disclosure may be used alone, in combination, or switched according to execution. In addition, notification of specific information (e.g., notification that "X is the case") is not limited to being done explicitly, but may be done implicitly (e.g., not notifying the specific information).

[0059] (D: Aspects understood from the above-described embodiments or modifications) Although the present disclosure has been described in detail above, it is clear to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented as modified and altered forms without departing from the spirit and scope of the present disclosure as defined by the claims. Therefore, the description of the present disclosure is intended to be illustrative and does not have any limiting meaning on the present disclosure. The following aspects can be understood from at least one of the above-mentioned embodiments or modifications.

[0060] A question generation device according to a first aspect of the present disclosure includes a first storage unit, an acquisition unit, a first reception unit, an extraction unit, and a generation unit. The first storage unit stores a plurality of templates, each of which has one or more blanks for completing a question by filling the blanks with features representing attributes or behavioral features of users of the facility, and each template has a different type of feature. The acquisition unit acquires video data representing a video captured by a camera inside the facility. The first reception unit accepts an instruction to select one of the plurality of templates stored in the first storage unit. The extraction unit extracts one or more features corresponding to the template selected in response to the instruction accepted by the first reception unit by analyzing the video data acquired by the acquisition unit. The generation unit generates an output question by filling the blanks corresponding to the one or more features extracted by the extraction unit in the selected template.

[0061] According to the first aspect of the question generation device, questions are automatically generated based on a template selected by the user and various features extracted from video data obtained by capturing footage of the interior of the target facility. This keeps the quality of the questions constant regardless of the user's skills in analyzing data for business improvement and creating questions, thereby reducing the user's workload.

[0062] The facility in the second aspect (example of the first aspect) of the present disclosure is a store. The question generation device according to the example of the second aspect includes a second memory unit, a determination unit, and a discard unit. The second memory unit stores store data representing at least one of the location, sales floor area, and business status of the store. The determination unit determines the appropriateness of an answer obtained by inputting the output question sentence into the large-scale language model based on the location, sales floor area, or business status represented by the store data. The discard unit discards answers determined to be inappropriate by the determination unit.

[0063] According to the second aspect, if an answer to a question generated based on a template selected by a user and various features extracted from video data obtained by capturing images of the inside of a target facility is inappropriate considering the location, sales floor area, or business status of the store, the answer is discarded without being presented to the user, thereby avoiding the proposal of management improvement measures that are inappropriate considering the location, sales floor area, or business status of the store.

[0064] A question generation device according to a third aspect (an example of the first aspect) of the present disclosure includes a second reception unit and a third storage unit. The second reception unit receives an input of evaluation data regarding the quality of an answer obtained by inputting the output question into the large-scale language model. The third storage unit stores the output question that led to the answer received by the second reception unit, and evaluation data indicating a good evaluation.

[0065] According to the third aspect, among the proposals obtained by inputting the output question sentence into the large-scale language model, the output question sentence that led to a proposal that was given a good evaluation is associated with the evaluation data representing the evaluation and stored in the storage device 150, so that it becomes possible to judge the quality of each template through the evaluation of the proposals. [Explanation of symbols]

[0066] 10...question generation device, 110...input device, 120...display device, 130...communication device, 150...storage device, 160...processing device, 160a...acquisition unit, 160b...first reception unit, 160c...extraction unit, 160d...generation unit, 160e...determination unit, 160f...discarding unit, 160g...second reception unit, 190...bus, PR1...program.

Claims

1. a first storage unit that stores a plurality of templates, the templates being templates for questions to be input to a large-scale language model in order to receive proposals for improving the management of a facility, the templates having one or more blanks that are filled with features that represent attributes or behavioral features of users of the facility to complete the questions, the templates corresponding to different types of features; An acquisition unit that acquires video data representing a video captured by a camera inside the facility; a first reception unit that receives an instruction to select one template from among a plurality of templates stored in the first storage unit; an extracting unit that extracts one or more feature amounts corresponding to a template selected in response to the instruction accepted by the first accepting unit by analyzing the video data acquired by the acquiring unit; a generation unit that generates an output question sentence by filling in corresponding blanks in the selected template with one or more feature amounts extracted by the extraction unit; A question generation device comprising:

2. The facility is a store, a second storage unit that stores store data indicating at least one of the location, sales floor area, and business status of the store; a determination unit that determines whether an answer obtained by inputting the output question sentence into the large-scale language model is appropriate based on the location, sales floor area, or business status represented by the store data; a discarding unit that discards answers that are determined to be inappropriate by the determining unit; The question generation device according to claim 1 , further comprising:

3. a second reception unit that receives an input of evaluation data regarding the quality of an answer obtained by inputting the output question sentence into the large-scale language model; a third storage unit configured to store an output question sentence that has led to an answer that has been received by the second receiving unit and that indicates a favorable evaluation; The question generation device according to claim 1 , further comprising:

Images