Retrieval assistance device, retrieval assistance method, and program

Abstract

To facilitate finding a statistical number suitable for an article.SOLUTION: A retrieval assistance device comprises a reception unit, an identification unit, and an output unit. The reception unit receives first input pertaining to an article as a retrieval object. The identification unit identifies a statistical number candidate by which the object to be retrieved is classified, on the basis of the first input and the classification information of each statistical number included in a statistical number group of hierarchical structure. The output unit outputs first information for requesting second input pertaining to the retrieval object which is additionally inputted and the statistical number candidate when a plurality of statistical number candidates are identifiable by identification means.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present disclosure relates to a search support device and the like. [Background technology]

[0002] When importing or exporting goods, it is necessary to assign a statistical number to the goods. For example, in the case of Japan, the customs office publishes the effective tariff schedule, explanations, classification guidelines, and case studies (for example, Non-Patent Document 1). The effective tariff schedule lists statistical numbers in correspondence with the names of the goods that the statistical numbers represent.

[0003] For example, Patent Document 1 describes a method of inputting invoice information for a specific product and presenting the suitability of a statistical number that may correspond to each of a number of models. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2022-54527 [Non-patent literature]

[0005] [Non-Patent Document 1] Customs import and export procedures, [online], [searched June 21, 2023], Internet<https: / / www.customs.go.jp / tsukan / index.htm> Summary of the Invention [Problem to be solved by the invention]

[0006] There are thousands of different statistical numbers. It is necessary to find the statistical number that is appropriate for an item from the large number of statistical numbers. For example, if you do not understand the characteristics and use of an item, you may not know what statistical number to assign to it.

[0007] An example of an objective of the present disclosure is to provide a search support device or the like that makes it easy to find a statistical number suitable for an item. [Means for solving the problem]

[0008] According to one aspect of the present disclosure, a search support device includes: A receiving means for receiving a first input related to an item to be searched for; A specifying means for specifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in a hierarchical structure of the statistical number group; When the identification means can identify a plurality of statistical number candidates, the device is provided with first information requesting a second input relating to the search target to be additionally input, and an output means for outputting the statistical number candidates.

[0009] A search support method according to one aspect of the present disclosure includes: Accepting a first input regarding an item to be searched; Identifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in the hierarchical structure of the statistical number group; When a plurality of statistical number candidates can be identified, first information requesting a second input related to the search target to be additionally input and the statistical number candidates are output. Execute the process.

[0010] In one aspect of the present disclosure, a program is provided that includes: Accepting a first input regarding an item to be searched; Identifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in the hierarchical structure of the statistical number group; When a plurality of statistical number candidates can be identified, first information requesting a second input related to the search target to be additionally input and the statistical number candidates are output. Execute the process.

[0011] Each program may be stored in a non-transitory computer-readable recording medium. Effect of the Invention

[0012] According to the present disclosure, it is possible to facilitate finding a statistical number suitable for an item. [Brief description of the drawings]

[0013] [Figure 1] This is an explanatory diagram of the HS code. [Diagram 2] FIG. 1 is an explanatory diagram showing an example of a hierarchical structure of HS codes. [Diagram 3] 1 is a block diagram showing an example configuration of a search support device according to the present disclosure. [Figure 4] 13 is a flowchart showing an operation example of the search support device. [Diagram 5] FIG. 2 is an explanatory diagram showing an example of a connection between the search support device and another device. [Figure 6] 1 is a block diagram showing a configuration example of a search support device and a terminal device according to the present disclosure. [Figure 7A] FIG. 11 is an explanatory diagram showing an example of a screen capable of accepting a first input. [Figure 7B] FIG. 13 is an explanatory diagram showing an example of a screen on which a first input and an HS code of a higher hierarchy are input. [Figure 8] FIG. 1 is an explanatory diagram showing an outline example of a process for identifying HS code candidates. [Figure 9] FIG. 1 is an explanatory diagram showing an example of identifying HS code candidates for a class hierarchy using a large-scale language model. [Figure 10] FIG. 1 is an explanatory diagram showing an example of identifying HS code candidates for a term hierarchy using a large-scale language model. [Figure 11] FIG. 11 is an explanatory diagram showing another example of identifying HS code candidates for a term hierarchy using a large-scale language model. [Figure 12] FIG. 13 is an explanatory diagram showing an example in which HS code candidates and questions from an AI are displayed on the same screen. [Figure 13]FIG. 11 is an explanatory diagram showing Example 1 in which detailed information about the item "6111" is displayed. [Figure 14] FIG. 11 is an explanatory diagram showing example 2 in which detailed information about item “6111” is displayed. [Figure 15] FIG. 11 is an explanatory diagram showing an example in which the HS code of a term is associated with a question ID. [Figure 16] 13 is an explanatory diagram showing an example in which second information indicating the reason why a second input requested by a question is necessary is displayed. FIG. [Figure 17] FIG. 13 is an explanatory diagram showing an example of how HS code candidates are displayed when a second input, which is an answer to a question, is entered. [Figure 18] FIG. 11 is an explanatory diagram showing an example of a screen displaying an answer input field for a question; [Figure 19] 13 is an explanatory diagram showing an example in which an answer input field for a question and information about the topic in which the question arises are displayed. FIG. [Figure 20] 13 is a flowchart showing an example of an operation of the search support device. [Figure 21] FIG. 2 is an explanatory diagram illustrating an example of a hardware configuration of a computer. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, with reference to the drawings, embodiments of a search support device, a search support method, a program, and a non-transitory recording medium for recording a program according to the present disclosure will be described in detail. The disclosed technology is not limited to the embodiments.

[0015] In the following embodiments, the statistical number may be referred to as the HS (Harmonized System) code. The HS code and product name used in the following explanation can be obtained from the tariff schedule published by the customs (source: https: / / www.customs.go.jp / yusyutu / 2023_04_01 / index.htm). In addition, the explanation of the HS code used in the following explanation will be explained using the explanation of the class "61" published by the customs (source: https: / / www.customs.go.jp / tariff / kaisetu / data / 61r.pdf). Here, the HS code will be briefly explained using Figure 1. Figure 1 is an explanatory diagram of the HS code. The HS code is a code number established based on the international treaty on the Harmonized Commodity Description and Coding System. The HS code classifies trade items into 21 "sections" and represents them as six-digit numbers or more digits. The globally standardized number is up to six digits, and each country can add any number of digits beyond the sixth digit. Of the six digits, the first two digits are called the chapter, the first four digits including the chapter are called the heading, and the first six digits including the heading are called the sub-heading. FIG. 1 shows an example of expressing an HS code with nine digits. For example, the first nine digits including the heading are called subdivisions. In this embodiment, an example for searching a nine-digit HS code is used for explanation.

[0016] Here, HS codes have a tree structure. In other words, HS codes can be expressed in a hierarchical structure. An example of a hierarchical HS code group will be described below.

[0017] FIG. 2 is an explanatory diagram showing an example of a hierarchical HS code group. In FIG. 2, the HS code classes "61" and "62" are described by taking some of the hierarchical HS codes as examples. When the class is "61", the terms are "6101" to "6117", etc. When the term is "6103", the subdivision is "6103.10", etc. When the subdivision is "6103.10", the subdivision level is "6103.10.000", etc. For example, the level one level below the class level is the term level, and the level two levels below is the subdivision level. When the term level is the subdivision level, the level one level below is the subdivision level, and the level two levels below is the subdivision level. The class level is also called the two-digit level, the term level is also called the four-digit level, the subdivision level is also called the six-digit level, and the detailed level is also called the nine-digit level.

[0018] Furthermore, there are divisions at a higher level than the class hierarchy. For example, classes "61" and "62" are included in division "11".

[0019] As mentioned above, the Japanese Customs Office publishes the tariff schedule, classification notes, explanations, classification rules, and case studies. For example, the classification notes provide explanations of the classes. The explanations provide explanations of the items. Furthermore, the explanations provide the names of products that correspond to the numbers, which are the hierarchy below each item.

[0020] (Embodiment 1) First, a first embodiment will be described. In the first embodiment, an example of a basic function of a search support device will be described. FIG. 3 is a first block diagram showing a configuration example of a search support device 10 according to the present disclosure. The search support device 10 includes a receiving unit 101, a specifying unit 102, and an output unit 103.

[0021] The reception unit 101 receives a first input related to an item to be searched for. For example, the reception unit 101 may receive a keyword related to the item to be searched for. For example, the reception unit 101 may receive a keyword input in an input field displayed on a user's terminal device as the first input.

[0022] The identification unit 102 identifies HS code candidates into which the search target is classified, based on the first input and the classification information of each HS code included in the hierarchical HS code group. The classification information is, for example, a class note, a term name, and a term explanation. For example, when narrowing down the HS codes to HS codes in a class hierarchy of the HS code group, the classification information of the HS code may be a class note and a term name. Also, when narrowing down the HS codes to HS codes in a term hierarchy of the HS code group, the classification information of the HS code may be, for example, a class note and a term explanation.

[0023] Specifically, the identification unit 102 may identify HS code candidates by using, for example, a large-scale language model. As the large-scale language model, a Generative Pretrained Transformer (GPT) or the like may be used. More specifically, for example, the identification unit 102 may provide the first input and classification information of each HS code to the large-scale language model to identify HS code candidates.

[0024] When the identifying unit 102 is able to identify multiple HS code candidates, the output unit 103 outputs first information requesting a second input regarding the search target to be additionally input, and the HS code candidates. For example, the first information is information requesting additional input of information necessary for determining whether the search target is classified as an HS code candidate. More specifically, for example, the first information may represent a question. That is, for example, the first information may be information representing a question for narrowing down the HS code candidates into which the search target is classified, from the multiple HS code candidates.

[0025] The method by which the output unit 103 outputs is not particularly limited. For example, the output unit 103 displays the first information and the HS code candidate on the user's terminal device. The output unit 103 may cause the user's terminal device to display the first information and the HS code candidate. Then, for example, the user's terminal device may display the first information and the HS code candidate under the control of the output unit 103.

[0026] (flowchart) 4 is a flowchart showing an example of an operation of the search support device 10. The receiving unit 101 receives a first input related to an item to be searched (step S101). The identifying unit 102 identifies HS code candidates based on the first input and the classification information of each HS code (step S102). Then, the output unit 103 outputs the HS code candidates and first information related to the HS code candidates and requesting additional second input (step S103). Then, the search support device 10 ends the series of processes.

[0027] For example, when a customs broker assigns an HS code to an item, there may be cases where the information about the item is insufficient. For example, a customs broker needs to confirm more detailed information about the item from the manufacturer or customer, but does not know what specific information about the item to confirm, which is expected to increase the amount of communication between the customs broker and the manufacturer or customer. This creates a lot of work for users such as customs brokers.

[0028] In the first embodiment, when the search support device 10 receives a first input related to a search target item, the search support device 10 identifies an HS code candidate based on the first input and the classification information of each HS code. Then, when a plurality of HS code candidates are identified, the search support device 10 outputs first information related to the HS code candidate and requesting additional second input, and the HS code candidate. By checking the information requesting additional input, which is information related to the HS code candidate, the user can easily input additional information related to the search target for classifying the search target. This makes it easier to find an HS code suitable for the item. Therefore, the user can know what information to check about the item, which can save the user's effort.

[0029] (Embodiment 2) Next, the second embodiment will be described in detail with reference to the drawings. In the second embodiment, an example of displaying first information requesting additional input of information required to determine whether or not a document is classified as an HS code candidate, and second information indicating the reason why the second input requested by the question that is the first information is necessary, will be described in detail. Also, in the second embodiment, an example of causing a large-scale language model to determine whether or not to classify a document into an HS code will be described. Below, to the extent that the description of the second embodiment is not unclear, any description that overlaps with the above description will be omitted.

[0030] 5 is an explanatory diagram showing an example of a connection between the search support device 10 and other devices. The search support device 10 is a system that supports searches for HS codes. For example, the search support device 10 accepts user operations and presents information to the user via a terminal device 11. For example, the search support device 10 is connected to the terminal device 11 via a communication network NT.

[0031] For example, the terminal device 11 may be capable of displaying information from the search support device 10, or may have an application program pre-installed therein for transmitting information to the search support device 10. For example, the terminal device 11 may access the site of the search support device 10, which is a website, via the communications network NT.

[0032] The type of the terminal device 11 is not particularly limited and may be a PC (Personal Computer), a smartphone, a tablet device, etc. The number of terminal devices 11 may be prepared for each user, and is not particularly limited.

[0033] The search support device 10 is also connected to a large-scale language model server 12 via a communication network NT. The large-scale language model server 12 is a server that executes a process of inputting information to the large-scale language model and outputting an answer. The search support device 10 transmits a prompt in which a question or the like is described to the large-scale language model server 12. The large-scale language model server 12 inputs the question described in the prompt to the large-scale language model and obtains an answer from the large-scale language model. Then, the large-scale language model server 12 transmits information on the answer to the search support device 10. The search support device 10 only needs to receive information on the answer from the large-scale language model server 12. This allows the search support device 10 to provide a question to the large-scale language model and obtain an answer to the question from the large-scale language model.

[0034] 5, an example in which the large-scale language model server 12 uses a large-scale language model has been described, but the search support device 10 may have a large-scale language model. In this case, the search support device 10 inputs a prompt in which a question is described to the large-scale language model, and obtains an answer to the question from the large-scale language model.

[0035] Fig. 6 is a block diagram showing an example of the configuration of a search support device 10 and a terminal device 11 according to the present disclosure. For example, the search support device 10 includes a receiving unit 101, a specifying unit 102, and an output unit 103, similar to the example shown in Fig. 3. Furthermore, the search support device 10 may include a master DB 1001.

[0036] The master DB 1001 includes, for example, HS codes that have been assigned to goods in the past by users such as customs brokers. Furthermore, for example, the master DB 1001 may store master data that associates HS codes with information related to the import and export of goods.

[0037] The terminal device 11 also includes, for example, an input receiving unit 111 and a display unit 112.

[0038] Next, each functional unit will be described in more detail.

[0039] For example, the receiving unit 101 receives a first input related to an item to be searched for. The receiving unit 101 may further receive a specification of an HS code of a higher hierarchy.

[0040] 7A is an explanatory diagram showing an example of a screen capable of accepting a first input. The output unit 103 outputs a screen including an input field for the first input to the user terminal. Then, for example, the display unit 112 of the user terminal displays a screen including an input field for the first input according to the control of the output unit 103. For example, the information accepted as the first input is not limited to a keyword, but as an example, the first input is also referred to as a keyword input. As an example, the input field for the first input is referred to as a keyword input field i01.

[0041] The screen also includes an input field in which a higher-level HS code candidate can be input. In Fig. 7A, the HS code filter input field i02 is an input field in which a higher-level HS code candidate can be input. For example, a higher-level HS code candidate is an HS code candidate for the class hierarchy. The screen also includes an execute button. The specified higher-level HS code candidate is also called an HS code filter.

[0042] 7B is an explanatory diagram showing an example of a screen on which a first input and an HS code of a higher hierarchy are input. When the execute button is pressed, the input acceptance unit 111 of the user terminal accepts the information entered in the keyword input field i01 as the first input, and accepts the information entered in the input field i02 as an HS code filter. Then, the input acceptance unit 111 transmits the accepted first input and HS code filter to the search support device 10. Then, the acceptance unit 101 of the search support device 10 accepts the first input and the HS code filter by receiving them from the terminal device 11.

[0043] Note that detailed information transmission and reception between the input receiving unit 111 of the user terminal and the receiving unit 101 of the search support device 10 will be omitted in the following description. Also, for example, when the output unit 103 displays on the user terminal, the display unit 112 displays a screen on the user terminal, and detailed description between the output unit 103 and the display unit 112 will also be omitted.

[0044] In FIG. 7B, the receiving unit 101 receives "children's pants" as a keyword input, and receives the HS code "61" of the category as an HS code filter.

[0045] Next, the identification unit 102 identifies HS code candidates into which the search target is classified, based on the first input and the classification information of each HS code included in the hierarchical HS code group. The classification information is as explained in embodiment 1 of the present invention. Here, an example will be described in which the identification unit 102 identifies HS code candidates using a large-scale language model.

[0046] Fig. 8 is an explanatory diagram showing an outline of an example of the process of identifying HS code candidates. In Fig. 8, the identification unit 102 narrows down the categories based on information about the goods, and then narrows down the items. In Fig. 8, the information about the goods is the first input. For example, a category is designated by an HS code filter. That is, the HS code filter becomes an HS code candidate for the category hierarchy.

[0047] First, if the HS code filter is not received, the identification unit 102 narrows down the classes. Here, an example will be described in which the identification unit 102 uses an API (Application Programming Interface) to input a prompt to the large-scale language model. The identification unit 102 calls APIs for the number of classes. Then, the identification unit 102 inputs class notes for each class, items included in the class, and the first input, and causes the large-scale language model via the API to determine whether the product represented by the first input is classified into the HS code of the class. For example, the identification unit 102 obtains a list of classes into which the search target product may be classified from the large-scale language model. The process of narrowing down the classes will be described with reference to FIG. 9.

[0048] FIG. 9 is an explanatory diagram showing an example of identifying HS code candidates for a class hierarchy using a large-scale language model. Since GPT is used as an example of a large-scale language model, the API is represented as GPT API in each figure. As described above, for example, the identification unit 102 calls APIs for the number of classes. For example, an instruction to have the large-scale language model determine the HS code that determines the tariff rate when importing and exporting goods is described in the class screening prompt. Here, the instruction may be described in the form of a question, for example. In other words, having the large-scale language model determine the HS code for the class that determines the tariff rate when importing and exporting goods may be to have the large-scale language model answer the question described in the class screening prompt. Note that the instruction described in each prompt may also be described in the form of a question.

[0049] For example, the class screening prompt may be written with instructions to the large-scale language model such as "Please determine whether the item falls under the HS code." For example, the class screening prompt may be written to allow input of a first input as information about the item, a class note as an explanation of the HS code, and the name of an item included in a lower hierarchy of the HS code of the class. The class screening prompt may be written with instructions to output to the large-scale language model whether the item is likely to be classified into the HS code of the class.

[0050] Taking class "1" as an example, the identification unit 102 inputs the class notes, the item name, and the information of the search target item, which is the first input, for class "1" into the class screening prompt, and provides the class screening prompt to the API for class "1" to have the large-scale language model determine whether the search target item is likely to be classified into class "1". For example, the identification unit 102 obtains a determination result of whether the search target item is likely to be classified into class "1" from the large-scale language model. Note that the "AI output" shown in FIG. 14, which will be described later, is an example of a specific determination result output from the large-scale language model. More specifically, for example, the identification unit 102 transmits the class screening prompt to the large-scale language model server 12 via the API, and the large-scale language model server 12 inputs the class screening prompt into the large-scale language model. Then, the large-scale language model server 12 obtains a determination result of whether the search target item is likely to be classified into class "1" from the large-scale language model, and transmits the determination result for class "1" to the search support device 10. As a result, the identifying unit 102 can receive the determination result for the class "1" and thereby obtain the determination result for the class "1" from the large-scale language model.

[0051] The specification unit 102 may perform the same process for each of the classes 2 to 97. For example, the specification unit 102 may generate a list of classes that are determined to have a high probability of being classified, from among the determination results of each class.

[0052] In the example of FIG. 9, the identification unit 102 executes the process of making the large-scale language model judge in parallel for the number of classes. This can shorten the processing time. The identification unit 102 may also execute the judgment process in series, for example. For example, when the identification unit 102 executes the judgment process in series, the previous judgment result may be given as an input for the next judgment process. This can improve the judgment accuracy. The identification unit 102 may also combine parallel and serial, and the method of executing the judgment process is not particularly limited. Note that the example of FIG. 9 shows a large-scale language model with a small amount of data that can be read at one time, so the identification unit 102 executes the judgment process in parallel for each class. However, if the large-scale language model has a large amount of data that can be read at one time, the judgment process may be executed at one time.

[0053] The output unit 103 may output a "list of possible classes" shown in FIG. 8, which is an intermediate result, to the terminal device 11 of the user.

[0054] Returning to the description of FIG. 8, when the HS code filter is accepted, in other words, when the HS code of the class is specified, the specification unit 102 does not need to perform a process of narrowing down the class. When "61" is specified as shown in FIG. 7B, the specification unit 102 does not perform a process of narrowing down the class. That is, when an HS code candidate of the class is specified by the HS code filter in FIG. 8, a large-scale language model for narrowing down the class is not used. For example, in the case of using a large-scale language model, the more the number of times of use is increased, the higher the fee. In such a case, by specifying an HS code candidate of the class, it is possible to reduce the cost of using the large-scale language model. In addition, for example, when the search support device 10 is realized by a local server or the like and the local server has a large-scale language model, it is expected that resources such as a processor and memory are limited. In this case, it is expected that the search support device 10 cannot parallelize each process, or even if each process can be parallelized, the number of parallel processes is small. For this reason, when the search support device 10 has a large-scale language model as described above, by specifying an HS code candidate of the class by the HS code filter, it is possible to shorten the processing time.

[0055] Furthermore, when multiple HS codes for a class are specified, the identifying unit 102 may perform a process of narrowing down the search for each of the multiple specified classes.

[0056] Next, after narrowing down the classes, or when an HS code candidate for a class is specified, the identifying unit 102 calls APIs for the number of classes of HS code candidates. Then, for the class of HS code candidates, the identifying unit 102 causes the large-scale language model via the API to determine whether the item to be searched is classified into a term in a lower hierarchy of the class of HS code candidates. Also, the identifying unit 102 causes the large-scale language model via the API to identify information required to identify whether the item to be searched is classified into a term in a lower hierarchy of the class of HS code candidates.

[0057] 10 is an explanatory diagram showing an example of identifying HS code candidates for the term hierarchy using a large-scale language model. Taking the case where "61" is identified as the HS code candidate for the class hierarchy, an example of identifying HS code candidates for the term hierarchy, which is a hierarchy lower than the class "61", is explained.

[0058] For example, the identifying unit 102 calls APIs for the number of items in the lower hierarchy of the narrowed-down class. For example, the HS codes of the items in the lower hierarchy of the class "61" range from "6101" to "6117." That is, 17 APIs are called.

[0059] In addition, the term screening prompt may include an instruction to have the large-scale language model determine the HS code that determines the tariff rate for importing and exporting goods. As described above, the instruction may be written in the form of a question. In other words, having the large-scale language model determine the HS code of a term that determines the tariff rate for importing and exporting goods may mean having the large-scale language model answer the question written in the term screening prompt.

[0060] For example, the class screening prompt may include instructions to the large-scale language model such as "Please determine whether the item falls under the HS code of the class." For example, the term screening prompt may include instructions to input a first input as information about the item, and a class note and an explanation of the item as an explanation of the HS code. The term screening prompt may include instructions to output to the large-scale language model whether the item is likely to be classified into the item.

[0061] For example, the term screening prompt may include instructions to output the reason why a term is not likely to be classified as a term. For example, the term screening prompt may include instructions to output the reason why a term is likely to be classified as a term.

[0062] Also, for example, the term screening prompt may include an instruction to output information requesting additional input of information required for identification when there is insufficient information to determine whether it is classified as a term. Here, the information requesting additional input is the first information described in the first embodiment. The information requesting additional input may be output in the form of a question. Note that the information requesting additional input corresponds to, for example, a question from an AI shown in FIG. 12 described later.

[0063] Then, for each of the HS codes of the terms, the identification unit 102 inputs the class note "61", the explanation of the term, and the information of the item, which is the first input, into the term screening prompt, and provides the term screening prompt to the API to cause the large-scale language model to determine whether the item to be searched for is likely to be classified into the term. Also, the identification unit 102 causes the large-scale language model to determine information required to determine whether the item to be searched for is classified into the HS code of the term. More specifically, for example, the identification unit 102 transmits the term screening prompt to the large-scale language model server 12 via the API, and the large-scale language model server 12 inputs the term screening prompt into the large-scale language model. Then, the large-scale language model server 12 acquires a determination result from the large-scale language model, including information on the answer as to whether the item to be searched for is likely to be classified into the HS code of the term, information required to determine whether the item is classified into the HS code of the term, and the like. The large-scale language model server 12 transmits the determination result to the search support device 10. As a result, the identification unit 102 can acquire a determination result from the large-scale language model by receiving each piece of information.

[0064] Here, the HS code candidates are those that are determined to have a high probability of classifying the search target, and those that are determined to have insufficient information to determine whether the search target can be classified.

[0065] There may be cases where the information required for identification is duplicated. Therefore, the identification unit 102 inputs the judgment results for each term into a result aggregation prompt, and aggregates the judgment results into a large-scale language model. The result aggregation prompt may describe an instruction to aggregate the judgment results for candidate HS codes. For example, the result aggregation prompt may describe an instruction to output a list of candidate terms and information required for identification. The identification unit 102 may store information that associates terms with information required for term identification.

[0066] FIG. 11 is an explanatory diagram showing another example of identifying HS code candidates for a term hierarchy using a large-scale language model. When the identification unit 102 causes the large-scale language model to determine whether the search target item is classified into a term, the identification unit 102 may further provide the master DB 1001 to the large-scale language model. In FIG. 11, when the large-scale language model determines whether the item is classified into the term "6101", for example, the identification unit 102 inputs master data indicating that the item is classified into the term "6101" in the master DB 1001, a class note for the class "61", an explanation for the term "6101", and information on the item into a term screening prompt. Then, the identification unit 102 inputs the term screening prompt into the large-scale language model via an API. By the identification unit 102 using the master DB 1001, it may be possible to easily classify even items that are not described in the explanation of the term. For example, it is possible to classify even new item names that are not described in the Harmonized Tariff Schedule.

[0067] Next, when multiple HS code candidates can be identified, the output unit 103 outputs first information requesting a second input regarding the search target to be additionally input, and the HS code candidates. As described above, the first information may be information representing a question necessary to identify whether the item to be searched is classified into the HS code candidate. The first information may be a question. The output unit 103 displays the HS code candidate and the first information on the terminal device 11. The output unit 103 may display the HS code candidate and the first information on the same screen of the user terminal.

[0068] Fig. 12 is an explanatory diagram showing an example in which a candidate HS code and a question from AI (Artificial Intelligence) are displayed on the same screen. In Fig. 12, the screen displays the candidate HS code and a question from the AI. The question from the AI ​​is the question represented by the above-mentioned first information.

[0069] For example, if the item being searched for is a pair of pants for children and a suit, if it is a boy's suit, it will be classified under item "6103," and if it is a girl's suit, it will be classified under item "6104." For this reason, the AI ​​may ask questions such as, "Is it for boys or girls?"

[0070] Also, for example, if the item being searched for is clothing for infants and toddlers whose height is 86 cm or less, it will be classified under item "6111." For this reason, the AI ​​may ask questions such as, "Is your height 86 cm or more?"

[0071] Furthermore, the output unit 103 may output information related to the HS code candidate for the HS code candidate. Examples of information related to the HS code candidate, which is a term, include information on a higher level of the term, information on a lower level of the term, link information on the explanation of the term, a determination result obtained from a large-scale language model, and the master DB 1001. For example, the information on a higher level of the term or information on a lower level of the term may be the HS code itself or may be the product name corresponding to the HS code, and is not particularly limited. For example, the link information indicates a link destination to PDF data of the explanation of the term.

[0072] Specifically, for example, when the input receiving unit 111 receives a click operation on an HS code candidate by the user, the receiving unit 101 receives an instruction to output detailed information. Then, the output unit 103 outputs, for the HS code candidate clicked, information on upper and lower hierarchical levels of the term, link information on the explanation of the term, the determination result obtained from the large-scale language model, the master DB 1001, etc.

[0073] 13 is an explanatory diagram showing an example 1 in which detailed information related to the item "6111" is displayed. For example, when the item "6111" is clicked, the output unit 103 displays a screen including information related to the item "6111" superimposed on the screen shown in FIG.

[0074] In Fig. 13, the screen displays class "61", which is a higher level of item "6111", the name of the class "61", and the name of the division "11". The screen displays the selected item "6111" and the name of the item "6111".

[0075] 13, the screen displays link information of the explanation of the section “61.11” published on the website of the customs. For example, when the input receiving unit 111 detects a click operation on a link indicated by the link information, the output unit 103 may display the linked website on the terminal device 11.

[0076] In Fig. 13, the screen allows selection between the master and the AI ​​output, and the master is selected. Therefore, the screen displays the master DB 1001. The screen may also display the master DB 1001 for items previously classified into the operated item "6111".

[0077] Fig. 14 is an explanatory diagram showing Example 2 in which detailed information on the term "6111" is displayed. In Fig. 14, the master and the output of the AI ​​can be selected on the screen, and the output of the AI ​​is selected. Therefore, the determination result of the term output from the large-scale language model in the identification unit 102 is displayed on the screen as the output of the AI.

[0078] Returning to the description of Fig. 12, for example, the output unit 103 may output second information indicating the reason why the second input requested by the question that is the first information is necessary. The second information may include the HS code candidate and the product name of the HS code candidate related to the first information.

[0079] For example, each question may be associated with the HS code of the term in which the question occurred. For example, the question DB may store question identification information for identifying the question, the question, and the HS code of the term in association with each other. The question identification information is not particularly limited as long as it can uniquely identify the question, such as a question ID (IDentifier).

[0080] Fig. 15 is an explanatory diagram showing an example in which a question ID is associated with an HS code of a term. In Fig. 15, the question ID of a question, which is the first information, is associated with the HS code of the term in which the question occurred.

[0081] FIG. 16 is an explanatory diagram showing an example in which second information indicating the reason why the second input requested by the question is necessary is displayed. For example, when the input receiving unit 111 receives a click operation on the question "Is it for boys or girls?" from the AI, the receiving unit 101 receives an instruction to output the second information. Then, the output unit 103 displays the HS code of the item and the name of the item as the second information indicating the reason why the second input requested by the clicked question is necessary. More specifically, for example, the output unit 103 may identify the HS code of the item associated with the question ID that identifies the clicked question in the question DB, and display the identified HS code and the name of the item on the screen.

[0082] In FIG. 16, the items "6103", "6104", "6107", and "6108" and their respective product names are displayed on the screen.

[0083] The second information may be an explanation of the term, or may be a judgment result of the term as shown in FIG.

[0084] Returning to the explanation of FIG. 12, for example, the user may make a second input related to the item to be searched for in the keyword input field i01 in the form of answering a question from the AI. For example, "boys'" and "86 cm or less" may be input next to "children's pants". When the input receiving unit 111 receives a press of the execute button, the receiving unit 101 receives the second input. Then, the identifying unit 102 identifies a new HS code candidate based on the first input, the additional second input, and the classification information of each HS code included in the hierarchical HS code group. The HS code candidates are as described above. Then, when the identifying unit 102 can identify multiple HS code candidates, the output unit 103 may output the first information related to the HS code candidate and requesting a second input related to the search target to be additionally input, and the HS code candidate.

[0085] Fig. 17 is an explanatory diagram showing an example in which HS code candidates are displayed when a second input, which is an answer to a question, is entered. In Fig. 17, the newly added "for men" is the second input. The number of HS code candidates in the screen shown in Fig. 17 is reduced compared to the screen shown in Fig. 12.

[0086] Here, an example of a usage flow by a user will be briefly described. For example, as shown in FIG. 7B, the user may input information on the product related to the search target and an HS code filter for the class, which may be incomplete at first. Then, when the execute button is pressed, the search support device 10 calls an API for the term under the class input in the HS code filter, and causes the large-scale language model to determine whether the search target is classified as a term. The search support device 10 displays the HS code candidate and the first information requesting additional input of information required to determine whether the search target is classified as the HS code candidate on the screen. Then, when there are many questions from the AI, the user inputs an answer to the question from the AI ​​in the keyword input field i01 and presses the execute button. Also, each time the execute button is pressed, the search support device 10 executes a process of causing the large-scale language model to determine whether the search target is classified as a term based on the input entered in the keyword input field i01. For example, when the number of questions from the AI ​​decreases, the user may click the HS code candidate and determine the final HS code while referring to the master DB 1001 and the output of the AI, as shown in FIG. 13 and FIG. 14.

[0087] Here, if the second input is an input to the keyword input field i01, the identification unit 102 executes a process each time the execute button is pressed to have the large-scale language model determine whether the search target is classified as a term for each of the HS codes of the terms subordinate to the class specified in the HS code filter entered in the input field i02.

[0088] Furthermore, the output unit 103 may display a first area capable of accepting a first input and a second area different from the first area capable of accepting a second input. Then, the accepting unit 101 may accept information entered in the second area as a second input that is additional information related to the search target.

[0089] Fig. 18 is an explanatory diagram showing an example of a screen including an answer input field for a question. In Fig. 18, a first area capable of accepting a first input is a keyword input field i01 in which "baby pants" has been input.

[0090] 18, when the input receiving unit 111 receives a click operation on the question, which is the first information, the output unit 103 displays an answer input field i03 for inputting an answer to the question. This answer input field i03 is a second area capable of receiving a second input.

[0091] Similarly, for other questions, when the input receiving unit 111 receives a click operation on a question, the output unit 103 may display an answer input field i03 for the question.

[0092] FIG. 19 is an explanatory diagram showing an example in which an answer input field for a question and information about the item from which the question arises are displayed. In FIG. 19, when the input receiving unit 111 receives a click operation on a question, the output unit 103 may display the answer input field i03 for the question, as well as the HS code of the item that is the reason for the question, the item's product name, and an explanation of the item. The answer "for men" is entered in the answer input field i03. The input in this answer input field i is the second This is the input.

[0093] Returning to the explanation of Fig. 18, the screen further includes an answer button. When the input accepting unit 111 accepts pressing of the answer button, it accepts the information (second input) entered in the answer input field i03. Then, the accepting unit 101 receives the second input entered in the answer input field i03 from the terminal device 11, thereby accepting the second input.

[0094] Then, the identification unit 102 identifies new HS code candidates based on the second input and the classification information of the HS code candidates related to the first information requesting the second input. That is, the identification unit 102 further narrows down the HS code candidates from the HS code candidates. Specifically, for example, the identification unit 102 can identify the correspondence between the second input entered in the answer input field i03 to the question and the first information representing the question by previously associating the question with the HS code of the term as shown in FIG. 15. The identification unit 102 can call an API for the term to which the question to which the answer has been received has occurred, and cause the large-scale language model to execute a process to determine whether the product to be searched is classified into the term. This can reduce the number of processes to be determined by the large-scale language model from the second time onwards. For example, when the processes to be determined are executed in series, the processing time can be shortened. In addition, when the usage fee for the large-scale language model increases as the number of processes increases, the fee can be reduced by reducing the number of processes to be determined.

[0095] Furthermore, the output unit 103 may output the first information in a chat format. For example, when the user replies with "Why are you asking that question?", the output unit 103 may output the reason, which is the second information. Furthermore, for example, the output unit 103 may output questions, which are the first information, in order. Specifically, for example, when the receiving unit 101 receives an answer to a question (second input), the output unit 103 may output the next question.

[0096] In the above, an example has been described in which HS code candidates are identified for the term hierarchy. For example, the identifying unit 102 may identify HS code candidates for the number hierarchy by using the master DB 1001.

[0097] Furthermore, when there are multiple HS code candidates, the output unit 103 may preferentially display the HS code candidate determined to be highly likely to classify the search target among, for example, the HS code candidate determined to be highly likely to classify the search target and the HS code candidate determined to be insufficient for information to determine whether the search target is classified. As an example of displaying with priority, for example, the output unit 103 may display, in order, the HS code candidate determined to be highly likely to classify the search target, and the HS code candidate determined to be insufficient for information to determine whether the search target is classified. As an example of displaying with priority, for example, the output unit 103 may display the HS code candidate determined to be highly likely to classify the search target, and not display the HS code candidate determined to be insufficient for information to determine whether the search target is classified.

[0098] Furthermore, when there are multiple HS code candidates, the output unit 103 may display the multiple HS code candidates in numerical order. In this manner, there is no particular limitation on the method of displaying the multiple HS code candidates.

[0099] 20 is a flowchart showing an example of an operation of the search support device 10. The receiving unit 101 receives a first input related to an item to be searched for (step S201). For example, the receiving unit 101 may receive an input to a keyword input field i01 as the first input. In step S201, the receiving unit 101 may further receive an input to a filter input field of an HS code filter.

[0100] Next, the identification unit 102 identifies HS code candidates using a large-scale language model based on the first input and the classification information of each HS code (step S202). In step S202, the identification unit 102 may identify first information requesting an additional second input using the large-scale language model. The first information may be information representing a question. The second input may be an answer to the question.

[0101] The output unit 103 determines whether multiple HS code candidates have been identified by the identification unit 102 (step S203). When a large-scale language model is used to narrow down the search down to the term level, the output unit 103 may determine in step S203 whether multiple HS code candidates have been identified for the term level. When a large-scale language model is used to narrow down the search down to the issue level, the output unit 103 may determine in step S203 whether an HS code candidate has been uniquely identified.

[0102] On the other hand, if multiple HS code candidates have been identified (step S203: Yes), the output unit 103 displays the HS code candidates and first information requesting additional second input on the terminal device 11 (step S204). For example, the first information represents a question used to identify whether the item is classified as an HS code candidate. The second input is a response to the question.

[0103] Next, the receiving unit 101 determines whether an operation on the first information has been received (step S205). In step S205, for example, the receiving unit 101 may determine whether a click operation on a question has been received as shown in FIG.

[0104] If an operation for the first information has been accepted (step S205: Yes), the output unit 103 displays second information indicating the reason why a second input for the first information is required on the terminal device 11 (step S206). If an operation for the first information has not been accepted (step S205: No), or after step S206, the accepting unit 101 accepts a second input (step S207). After step S207, the search support device 10 returns to step S202. This causes the process of narrowing down HS code candidates to be repeated using the new input.

[0105] If multiple HS code candidates are not identified (step S203: No), the output unit 103 displays the HS code candidates on the terminal device 11 (step S208), and the search support device 10 ends the series of processes.

[0106] As described above, in the second embodiment, the search support device 10 further outputs second information indicating the reason why the second input requested by the first information is necessary. The second information includes at least one of the HS code candidate related to the first information, the product name of the HS code candidate, and the explanation of the HS code candidate section related to the first information. This allows the user to more accurately confirm what information is necessary to classify the product to be searched for into the presented HS code candidate.

[0107] Furthermore, the search support device 10 outputs the second information in response to an operation on the first information, thereby making it possible to display information indicating the reason why additional input is required, if required by the user.

[0108] Furthermore, the search support device 10 displays the HS code candidates and the first information on the same screen, which allows the user to check the HS code candidates and the question from the AI ​​at a glance.

[0109] The search support device 10 also displays a first area capable of accepting a first input and a second area different from the first area capable of accepting a second input. The search support device 10 then accepts information input to each area. This allows the search support device 10 to identify a correspondence between the first information, which is a question from the AI, and the second input, which is an answer to the question. The search support device 10 then identifies a new HS code candidate based on the second input and classification information of the HS code candidate related to the first information requesting the second input. In this way, the search support device 10 can narrow down the HS code candidates related to the first information, which is a question from the AI, and cause the large-scale language model to perform a process of determining again. Therefore, the number of processes to be determined by the large-scale language model can be reduced. For example, when the processes to be determined are performed in series, the processing time can be shortened. Furthermore, when the usage fee for the large-scale language model increases as the number of processes increases, the cost can be reduced by reducing the number of processes to be determined.

[0110] Furthermore, the search support device 10 further identifies HS code candidates into which the search target is classified based on information about HS codes previously assigned to goods by the user. By using such information, it may be possible to classify goods that are not described in the explanations published by the customs. For example, it may be possible to classify new goods names that are not described in the Harmonized Tariff Schedule published by the customs.

[0111] This concludes the description of each embodiment. The embodiments may be combined. In each embodiment, the search support device 10 may include each functional unit and part of the information.

[0112] The above is the end of the description of the application examples. In addition, each embodiment is not limited to the above-mentioned examples, and can be modified in various ways. In addition, the configuration of the search support device 10 in the embodiment is not particularly limited. For example, the functional units of the search support device 10 may be realized by one device. Or, for example, each functional unit or DB of the search support device 10 in the embodiment may be realized by a different device and configured as a system. For example, each functional unit of the search support device 10 may be configured by a plurality of servers and realized as a system. For example, it may be realized by a database server including each DB and a server having each functional unit. Each functional unit of the search support device 10 may be realized by a cloud server or the like.

[0113] In the embodiment, each DB may include a part of the above-mentioned information. Also, each piece of information may include information other than the above-mentioned information.

[0114] Furthermore, the process of generating information to be displayed on the terminal device 11 may be performed by the output unit 103 of the search support device 10. This process may also be performed by the display unit 112 of the terminal device 11. That is, the terminal device 11 generates information for a screen to be displayed on the terminal device 11 based on the data received from the search support device 10, and displays the screen. Furthermore, each screen is an example, and the positions, colors, sizes, and the like of input fields, buttons, and the like may be changed as appropriate.

[0115] (Example of computer hardware configuration) Next, an example of a hardware configuration in which each device, such as the search support device 10, the terminal device 11, and the large-scale language model server 12, described in the embodiment, is realized by a computer will be described. Fig. 21 is an explanatory diagram showing an example of a hardware configuration of a computer. For example, a part or all of each device can be realized by using any combination of a computer 80 and a program as shown in Fig. 21.

[0116] The computer 80 includes, for example, a processor 801, a read only memory (ROM) 802, a random access memory (RAM) 803, and a storage device 804. The computer 80 also includes a communication interface 805 and an input / output interface 806. The components are connected to each other, for example, via a bus 807. The number of each component is not particularly limited, and there may be one or more components.

[0117] The processor 801 controls the entire computer 80. As the processor 801, for example, a central processing unit (CPU), a digital signal processor (DSP), a graphics processing unit (GPU), a physics processing unit (PPU), a tensor processing unit (TPU), a quantum processor, or a combination of these can be used, and is not particularly limited.

[0118] The computer 80 also includes a ROM 802, a RAM 803, and a storage device 804. Examples of the storage device 804 include a semiconductor memory such as a flash memory, a hard disk drive (HDD), and a solid state drive (SSD). For example, the storage device 804 stores an operating system (OS) program, an application program, and a program related to the embodiment. Alternatively, the ROM 802 stores an application program and a program related to the embodiment. The RAM 803 is used as a work area for the processor 801.

[0119] The processor 801 also loads programs stored in the storage device 804, ROM 802, etc. Then, the processor 801 executes each process coded in the program. The processor 801 may also download various programs via the communication network NT. The processor 801 also functions as a part or the whole of the computer 80. Then, the processor 801 may execute processes or instructions in the illustrated flowchart based on the program.

[0120] The communication interface 805 is connected to a communication network NT, such as a LAN (Local Area Network) or a WAN (Wide Area Network), via a wireless or wired communication line. The communication network NT may be composed of a plurality of communication networks NT. As a result, the computer 80 is connected to an external device or an external computer 80 via the communication network NT. The communication interface 805 serves as an interface between the communication network NT and the inside of the computer 80. The communication interface 805 controls the input and output of data from the external device or the external computer 80.

[0121] The input / output interface 806 is connected to at least one of an input device, an output device, and an input / output device. The connection method may be wireless or wired. Examples of the input device include a keyboard, a mouse, and a microphone. Examples of the output device include a display device, a lighting device, and an audio output device that outputs audio. Examples of the input / output device include a touch panel display. The input device, the output device, and the input / output device may be built into the computer 80 or may be external.

[0122] The hardware configuration of the computer 80 is an example. The computer 80 may have some of the components shown in FIG. 21. The computer 80 may have components other than those shown in FIG. 21. For example, the computer 80 may have a drive device or the like. The processor 801 may read out a program or data stored in a recording medium attached to the drive device or the like to the RAM 803. Examples of non-transient tangible recording media include optical disks, flexible disks, magneto-optical disks, and USB (Universal Serial Bus) memories. As described above, the computer 80 may have input devices such as a keyboard and a mouse. The computer 80 may have an output device such as a display. The computer 80 may have an input device, an output device, and an input / output device.

[0123] The computer 80 may also include various sensors (not shown). The type of the sensor is not particularly limited. The computer 80 may also include an imaging device capable of capturing images or videos.

[0124] This concludes the description of the hardware configuration of each device. There are various variations in the method of realizing each device. For example, each device may be realized by any combination of a computer and a program that are different for each component. Furthermore, multiple components included in each device may be realized by any combination of a single computer and a program.

[0125] Also, some or all of the components of each device may be realized by circuits for specific applications. Also, some or all of the components of each device may be realized by general-purpose circuits such as an FPGA (Field Programmable Gate Array). Also, some or all of the components of each device may be realized by a combination of circuits for specific applications and general-purpose circuits. Also, these circuits may be a single integrated circuit. Or, these circuits may be divided into multiple integrated circuits. And, the multiple integrated circuits may be configured by being connected via a bus or the like.

[0126] Furthermore, when some or all of the components of each device are realized by a plurality of computers, circuits, etc., the plurality of computers, circuits, etc. may be arranged in a centralized manner or in a distributed manner.

[0127] The search support method described in the embodiment may be implemented by a computer such as the search support device 10.

[0128] Each program, such as the formulation program and the optimization program described in the embodiment, is recorded on a computer-readable recording medium, such as an HDD, an SSD, a flexible disk, an optical disk, a magneto-optical disk, or a USB memory. Each program is executed by being read from the recording medium by a computer. Each program may also be distributed via a communication network NT.

[0129] The functions of each of the components of the search support device 10 described above may be realized by dedicated hardware, such as a computer. Alternatively, each component may be realized by software. Alternatively, each component may be realized by a combination of hardware and software.

[0130] Although the present disclosure has been described above with reference to the embodiments, the present disclosure is not limited to the above embodiments. The configuration and details of each of the present disclosures may include embodiments to which various modifications that a person skilled in the art may understand within the scope of the present disclosure are applied. The present disclosure may include embodiments in which the matters described in this specification are appropriately combined or replaced as necessary. For example, matters described using a specific embodiment may also be applied to other embodiments to the extent that no contradiction occurs. For example, although multiple operations are described in order in the form of a flowchart, the order of description does not limit the order in which the multiple operations are performed. Therefore, when implementing the embodiment, the order of the multiple operations can be changed to the extent that the content is not impaired.

[0131] A part or all of the above-described embodiments can be described as follows: However, a part or all of the above-described embodiments is not limited to the following.

[0132] (Appendix 1) A receiving means for receiving a first input related to an item to be searched for; A specifying means for specifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in a hierarchical structure of the statistical number group; an output means for outputting, when a plurality of statistical number candidates can be identified by the identification means, first information requesting a second input related to the search target to be additionally input, and the statistical number candidates; A search support device comprising: (Appendix 2) the first information represents a question used to determine whether the search target is classified as the statistical number candidate; the second input being an answer to the question; 2. A search support device as described in appendix 1. (Appendix 3) The output means outputs second information indicating a reason why the second input requested by the first information is necessary. 3. The search support device according to claim 1 or 2. (Appendix 4) The second information includes at least one of the statistical number candidate related to the first information, the name of the statistical number candidate, and the explanation of the statistical number candidate item related to the first information, 4. A search support device as described in appendix 3. (Appendix 5) The output means includes: outputting the second information in response to an operation on the first information; 4. A search support device as described in appendix 3. (Appendix 6) the output means displays a first area capable of accepting the first input and a second area different from the first area capable of accepting the second input; the receiving means receives information input into the second area as the second input; The identification means identifies a new statistical number candidate based on the second input and classification information of the statistical number candidate related to the first information requesting the second input. 7. A search support device according to any one of claims 1 to 6. (Appendix 7) The identification means further identifies the candidate statistical number into which the search target is classified based on information regarding statistical numbers previously assigned to items by users. 7. A search support device according to any one of claims 1 to 6. (Appendix 8) The output means displays the statistical number candidate and the first information on the same screen. 8. A search support device according to any one of claims 1 to 7. (Appendix 9) When a plurality of statistical number candidates can be identified from the statistical number of a hierarchy of an item in the hierarchical structure of the statistical number group, the output means outputs the first information and the statistical number candidates. 9. A search support device according to any one of supplements 1 to 8. (Appendix 10) The identification means identifies the statistical number candidates and the first information by using a large-scale language model. 10. A search support device according to any one of claims 1 to 9. (Appendix 11) the identification means inputs a prompt including the first input, classification information of each statistical number included in a hierarchically structured group of statistical numbers, and question information as to whether the search target is classified into the statistical number, to a large-scale language model, and acquires the statistical number candidates and the first information from the large-scale language model. 11. The search support device according to claim 10. (Appendix 12) The computer Accepting a first input regarding an item to be searched; Identifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in the hierarchical structure of the statistical number group; When a plurality of statistical number candidates can be identified, first information requesting a second input related to the search target to be additionally input and the statistical number candidates are output. A search aid method for performing a process. (Appendix 13) the first information represents a question used to determine whether the search target is classified as the statistical number candidate; the second input being an answer to the question; A search assistance method as described in Appendix 12. (Appendix 14) In the outputting process, second information indicating a reason why the second input requested by the first information is necessary is output. 14. A search assistance method according to claim 12 or 13. (Appendix 15) The second information includes at least one of the statistical number candidate related to the first information, the name of the statistical number candidate, and the explanation of the statistical number candidate item related to the first information, A search aid method as described in Appendix 14. (Appendix 16) In the output process, the second information is output in response to an operation on the first information. A search aid method as described in Appendix 14. (Appendix 17) In the output process, a first area capable of accepting the first input and a second area different from the first area capable of accepting the second input are displayed; In the receiving process, information input in the second area is received as the second input; In the process of specifying, a new statistical number candidate is specified based on the second input and classification information of the statistical number candidate related to the first information requesting the second input. A search support method according to any one of appendix 12 to appendix 17. (Appendix 18) The process of specifying further includes specifying the candidate statistical number into which the search target is classified based on information regarding statistical numbers previously assigned to items by users. A search support method according to any one of appendix 12 to appendix 17. (Appendix 19) In the output process, the statistical number candidate and the first information are displayed on the same screen. A search support method according to any one of appendix 12 to appendix 18. (Appendix 20) In the output process, when a plurality of statistical number candidates can be identified from the statistical number of a hierarchy of an item in the hierarchical structure of the statistical number group, the first information and the statistical number candidates are output. A search support method according to any one of appendix 12 to appendix 19. (Appendix 21) In the process of identifying, the statistical number candidates and the first information are identified using a large-scale language model. A search support method according to any one of appendix 12 to appendix 20. (Appendix 22) In the process of identifying, a prompt including the first input, classification information of each statistical number included in a hierarchically structured group of statistical numbers, and question information as to whether the search target is classified into the statistical number is input to a large-scale language model, and the statistical number candidates and the first information are obtained from the large-scale language model. A search assistance method as described in Appendix 21. (Appendix 23) On the computer, Accepting a first input regarding an item to be searched; Identifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in the hierarchical structure of the statistical number group; When a plurality of statistical number candidates can be identified, first information requesting a second input related to the search target to be additionally input and the statistical number candidates are output. A program that executes a process. (Appendix 24) the first information represents a question used to determine whether the search target is classified as the statistical number candidate; the second input being an answer to the question; 23. The program according to claim 22. (Appendix 25) In the outputting process, second information indicating a reason why the second input requested by the first information is necessary is output. 25. The program according to claim 23 or 24. (Appendix 26) The second information includes at least one of the statistical number candidate related to the first information, the name of the statistical number candidate, and the explanation of the statistical number candidate item related to the first information, 26. The program according to claim 25. (Appendix 27) In the output process, the second information is output in response to an operation on the first information. 26. The program according to claim 25. (Appendix 28) In the output process, a first area capable of accepting the first input and a second area different from the first area capable of accepting the second input are displayed; In the receiving process, information input in the second area is received as the second input; In the process of specifying, a new statistical number candidate is specified based on the second input and classification information of the statistical number candidate related to the first information requesting the second input. 28. A program according to any one of appendix 23 to appendix 27. (Appendix 29) The process of specifying further includes specifying the candidate statistical number into which the search target is classified based on information regarding statistical numbers previously assigned to items by users. 29. A program according to any one of appendix 23 to appendix 28. (Appendix 30) In the output process, the statistical number candidate and the first information are displayed on the same screen. 29. The program according to any one of claims 23 to 29. (Appendix 31) In the output process, when a plurality of statistical number candidates can be identified from the statistical number of a hierarchy of an item in the hierarchical structure of the statistical number group, the first information and the statistical number candidates are output. 31. The program according to any one of appendix 23 to appendix 30. (Appendix 32) In the process of identifying, the statistical number candidates and the first information are identified using a large-scale language model. 32. The program according to any one of appendix 23 to appendix 31. (Appendix 33) In the process of identifying, a prompt including the first input, classification information of each statistical number included in a hierarchically structured group of statistical numbers, and question information as to whether the search target is classified into the statistical number is input to a large-scale language model, and the statistical number candidates and the first information are obtained from the large-scale language model. 33. The program according to claim 32. (Appendix 34) On the computer, Accepting a first input regarding an item to be searched; Identifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in the hierarchical structure of the statistical number group; When a plurality of statistical number candidates can be identified, first information requesting a second input related to the search target to be additionally input and the statistical number candidates are output. A non-transitory recording medium readable by the computer, which records a program for executing a process. [Explanation of symbols]

[0133] 10 Search support device 11 Terminal Equipment 80 Computer 101 Reception 102 Specific part 103 Output section 111 Input reception section 112 Display section 801 Processor 802 ROM 803 RAM 804 Storage device 805 Communication Interface 806 Input / Output Interface 807 Bus 1001 Master DB i01 Keyword input field i02 Input field i03 Answer input field NT Communications Network

Claims

1. A receiving means for receiving a first input related to an item to be searched for; A specifying means for specifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in a hierarchical structure of the statistical number group; an output means for outputting, when a plurality of statistical number candidates can be specified by the specifying means, first information requesting a second input related to the search target to be additionally input, and the statistical number candidates; A search support device comprising:

2. the first information represents a question used to determine whether the search target is classified as the statistical number candidate; The second input is an answer to the question. The search support device according to claim 1.

3. the output means outputs second information indicating a reason why the second input requested by the first information is necessary. The search support device according to claim 1.

4. The second information includes at least one of the statistical number candidate related to the first information, the name of the statistical number candidate, and the explanation of the statistical number candidate related to the first information. The search support device according to claim 3.

5. The output means includes: outputting the second information in response to an operation on the first information; The search support device according to claim 3.

6. the output means displays a first area capable of accepting the first input and a second area different from the first area capable of accepting the second input; the receiving means receives information input into the second area as the second input; The specifying means specifies a new statistical number candidate based on the second input and classification information of the statistical number candidate related to the first information requesting the second input. The search support device according to claim 1.

7. The identification means further identifies the candidate statistical number into which the search target is classified based on information regarding statistical numbers previously assigned to items by users. The search support device according to claim 1.

8. The output means displays the statistical number candidate and the first information on the same screen. The search support device according to claim 1.

9. The computer Accepting a first input relating to an item to be searched; Identifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in the hierarchical structure of the statistical number group; When a plurality of statistical number candidates can be identified, first information requesting a second input related to the search target to be additionally input and the statistical number candidates are output. A search aid method for performing a process.

10. On the computer, Accepting a first input relating to an item to be searched; Identifying a candidate statistical number into which the search target is classified based on the first input and classification information of each statistical number included in the hierarchical structure of the statistical number group; When a plurality of statistical number candidates can be identified, first information requesting a second input related to the search target to be additionally input and the statistical number candidates are output. A program that executes a process.

Images