Input support device, input support method, and input support program

Abstract

To improve the information coverage of an input sentence after complementation.SOLUTION: An input unit that inputs an input sentence and a question sentence related to a determination target, a determination unit that determines whether a predetermined condition defined in a rule related to the determination target is satisfied by a combination of the input sentence and the question sentence, a generation unit that generates, when the determination unit determines that the predetermined condition is not satisfied, missing information that is necessary for satisfying the predetermined condition and is not included in the input sentence, and a management unit that complements the input sentence using the missing information generated by the generation unit; And an output unit configured to output a determination result determined by the determination unit.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to an input support device, an input support method, and an input support program for supporting input. [Background technology]

[0002] In information systems that provide information or perform navigational functions in response to user input, such as procedure guidance systems and question-answering systems, users have traditionally had to input the information required for each function item by item, placing a burden on the user. Large-scale language models, on the other hand, are expected to enable users to input information directly into the system in natural language.

[0003] In order to accept information input in natural language, which was previously done by users on a per-item basis, it is important for users to include the information necessary for the action in their own input sentences. However, since users do not necessarily understand the information necessary for the action in advance, there is a possibility that the input sentences will be ambiguous or necessary information will be omitted. Therefore, technology is needed to support information input by assuming that the user's input sentences will be incomplete.

[0004] One method for realizing information supplementation of an input sentence is the technology of Patent Document 1 below. Patent Document 1 discloses "a question generation device having a search unit that searches a document collection for related documents related to an input question sentence based on the input question sentence, and a revised question sentence generation unit that generates a revised question sentence by supplementing the question sentence with words included in a predetermined vocabulary set based on the question sentence and the related documents." [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2022-111261 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the technology of Patent Document 1 adds new information to the input sentence using a probabilistic generative model. Therefore, even after the information is supplemented, it is not guaranteed that the input sentence contains all the information necessary for the answer. In order for the supplemented input sentence to satisfy the conditions necessary for the answer, it is necessary to guarantee that the supplemented input sentence contains all the necessary information.

[0007] The present invention aims to improve the information coverage of the input sentence after completion. [Means for solving the problem]

[0008] An input support device according to one aspect of the invention disclosed in the present application comprises an input unit that inputs an input sentence and a question sentence related to an object to be judged, a judgment unit that judges whether a predetermined condition defined in a rule related to the object to be judged is satisfied for a combination of the input sentence and the question sentence, a generation unit that generates missing information that is not included in the input sentence but is necessary for satisfying the predetermined condition when the judgment unit judges that the predetermined condition is not satisfied, a management unit that complements the input sentence using the missing information generated by the generation unit, and an output unit that outputs the judgment result judged by the judgment unit. [Effects of the Invention]

[0009] According to a representative embodiment of the present invention, it is an object to improve the information comprehensiveness of an input sentence after completion. Problems, configurations, and effects other than those described above will become clear from the following description of the embodiment. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a block diagram illustrating an example of the hardware configuration of an input support device. [Figure 2] FIG. 2 is a block diagram illustrating an example of a functional configuration of the input support device. [Figure 3] FIG. 3 is an explanatory diagram illustrating an example of the rule DB. [Figure 4] FIG. 4 is an explanatory diagram illustrating an example of the process management DB. [Figure 5] FIG. 5 is an explanatory diagram showing an example of the initial input screen. [Figure 6] FIG. 6 is an explanatory diagram showing an example of the answer screen. [Figure 7] FIG. 7 is a flowchart illustrating an example of an input support process procedure performed by the input support device. DETAILED DESCRIPTION OF THE INVENTION

[0011] In this embodiment, an example of an information supplementing method in the input support device 100 that supplements information necessary for an operation with respect to a user's input in natural language will be described.

[0012] <Figure 1. Example of hardware configuration for input support device> FIG. 1 is a block diagram showing an example of the hardware configuration of an input assistance device. The input assistance device 100 includes a processor 101, a storage device 102, an input device 103, an output device 104, and a communication interface (communication IF) 105. The processor 101, the storage device 102, the input device 103, the output device 104, and the communication IF 105 are connected via a bus 106. The processor 101 controls the input assistance device 100. The storage device 102 serves as a working area for the processor 101. The storage device 102 is a non-transitory or temporary recording medium that stores various programs and data. Examples of the storage device 102 include a read-only memory (ROM), a random access memory (RAM), a hard disk drive (HDD), and a flash memory. The input device 103 inputs data. Examples of the input device 103 include a keyboard, a mouse, a touch panel, a numeric keypad, a scanner, a microphone, and a sensor. The output device 104 outputs data. The output device 104 includes, for example, a display, a printer, and a speaker. The communication IF 105 connects to a network and transmits and receives data.

[0013] <Figure 2: Example of functional configuration of input support device 100> 2 is a block diagram showing an example of the functional configuration of the input support device 100. The input support device 100 has a rule DB (Database) 201 and a process management DB 202. Specifically, the rule DB 201 and the process management DB 202 are realized by, for example, the storage device 102 shown in FIG.

[0014] The input support device 100 also includes a data transmission / reception unit 211, a management unit 212, a conversion unit 213, a determination unit 214, a generation unit 215, and an input / output unit 216. Specifically, the data transmission / reception unit 211, the management unit 212, the conversion unit 213, the determination unit 214, the generation unit 215, and the input / output unit 216 are realized, for example, by causing the processor 101 to execute a program stored in the storage device 102 shown in FIG.

[0015] The rule DB 201 is a DB for holding rules for determining whether or not information supplementation is required for input information related to a determination target from the user terminal 230. The process management DB 202 is a DB for recording the process by which input information from the user terminal 230 is supplemented within the input support device 100 and for determining whether to continue or end the supplementation process.

[0016] The data transmission / reception unit 211 receives data to be used in the input support device 100 from outside the input support device 100, and transmits data generated by the input support device 100 to outside the input support device 100. The data transmission / reception unit 211 transmits and receives data via a network 220 such as the Internet, a LAN (Local Area Network), or a WAN (Wide Area Network).

[0017] The management unit 212 manages the iterative process of rule conversion by the conversion unit 213, rule determination by the determination unit 214, and information generation by the generation unit 215. The management unit 212 determines the operation required in the next stage using input information recorded in the process management DB 202, and controls the conversion unit 213, determination unit 214, or generation unit 215 to operate based on the determination result.

[0018] Furthermore, the management unit 212 stores the determination result by the determination unit 214 in the process management DB 202 , and stores the missing information generated by the generation unit 215 in the process management DB 202 .

[0019] The conversion unit 213 converts natural sentences written in a natural language into logical formulas that can be evaluated by a computer. The logical formulas are written in, for example, a logic programming language. If the natural sentences are rules written in a natural language, they are converted into if-then syntax, that is, logical formulas consisting of conditions and their results. The conversion unit 213 executes a process of converting the natural sentences recorded in the rule DB 201 into logical formulas, and records the logical formulas that are the conversion results in the rule DB 201. The process of conversion into logical formulas is executed using a machine learning model.

[0020] This machine learning model is a model trained using a natural language rule 301 related to the object to be determined and its logical expression rule 302 as a training data set. Therefore, when the natural language rule 301 is input to the machine learning model, the logical expression rule 302 is output. The machine learning model may be implemented in the conversion unit 213, or may be implemented on another computer that can communicate with the input support device 100 via the network 220.

[0021] The determination unit 214 determines whether or not the input information related to the determination target recorded in the process management DB 202 satisfies the condition defined in the rule of the determination target. Specifically, for example, the determination unit 214 uses a logical inference engine to determine whether or not the condition defined in the rule is met with a combination of a fact that occurred for the user obtained from the input information (hereinafter, "occurring fact") and the question, and generates a determination result indicating whether the condition is met or not. Specifically, for example, the determination unit 214 uses a logical inference engine to determine whether or not the condition defined in the rule is met with the occurring fact for the question. The determination unit 214 passes the determination result to the management unit 212.

[0022] The logical inference engine is trained using a training data set in which a combination of occurrence facts and question sentences is used as training data, and whether a condition is satisfied or not is used as correct answer data. Specifically, for example, the logical inference engine is implemented as a generative AI (artificial intelligence). The generative AI includes a language model trained by natural language processing using a large-scale training data set, and generates sentences using the language model.

[0023] A language model is a type of probabilistic model used in natural language processing, and is a model that probabilistically predicts how likely a given word or sentence is to occur in natural language. Specifically, in the field of natural language processing, it is a mathematical model that learns language patterns and grammatical rules, and generates and understands natural language.

[0024] For example, generative AI uses a language model to calculate the probability of occurrence of a given word sequence or sentence, or to compare the probabilities of occurrence of multiple word sequences or sentences, and when predicting the next word or sentence, automatically generates the most likely word or sentence based on the context. In this way, when generative AI receives a query called a prompt, it outputs an answer to that query using a language model that has been trained on a huge amount of data set.

[0025] The logical inference engine may be implemented in the determination unit 214, or may be implemented on another computer that can communicate with the input support device 100 via the network 220. When implemented on another computer, the determination unit 214 generates a prompt that inquires of the other computer whether or not the question is valid under the occurrence facts and rules. Upon receiving the prompt, the other computer uses the logical inference engine to return the determination result to the determination unit 214.

[0026] When it is determined that the input information recorded in the process management DB 202 is insufficient, that is, when the determination unit 214 determines that the condition is not satisfied, the generation unit 215 references the rules in the rule DB 201 and generates missing information. Specifically, for example, the generation unit 215 inputs the rules in the rule DB 201 and the input information recorded in the process management DB 202 into a machine learning model, generates missing information that is not included in the input information among the information necessary to comply with the rules, and passes the generated information to the management unit 212.

[0027] The machine learning model is trained using a training data set in which, for example, a combination of input information related to the object of determination, a question sentence, and a rule is used as training data, and missing information that is not included in the facts and is a condition specified in the rule required to answer the question is used as correct answer data. The machine learning model may also be a generative AI. By implementing the generative AI, the generation unit 215 completes the input sentence with fewer iterations based on the knowledge possessed by the generative AI. The machine learning model may be implemented in the generation unit 215 or on another computer that can communicate with the input assistance device 100 via the network 220.

[0028] Furthermore, when the generation AI is implemented on another computer, the generation unit 215 generates a prompt from the other computer requesting missing information for a combination of an occurrence, a rule, and a question. Upon receiving the prompt, the other computer returns the missing information to the generation unit 215 via the generation AI.

[0029] When the input / output unit 216 receives input information from the user terminal 230 via the network 220, it transmits the data to the management unit 212 of the input support device 100. The received input information is recorded in the process management DB 202. The input / output unit 216 may accept input information from the input device 103 of the input support device 100, or may output output information to the output device 104 of the input support device 100.

[0030] The user terminal 230 is a terminal through which a user makes settings for the input support device 100, and includes, for example, a personal computer, a mobile phone, or a tablet. The user connects to the input support device 100 using the user terminal 230. The user terminal 230 also has the same hardware configuration as that shown in FIG. 1.

[0031] <Figure 3 Rule DB201> 3 is an explanatory diagram showing an example of the rule DB 201. The rule DB 201 has the following fields: a rule ID 300, a natural language rule 301, and a logical expression rule 302. A combination of values ​​in the fields in the same row becomes an entry that defines one rule. A rule is a standard that must be observed when an occurrence occurs in relation to a subject of judgment.

[0032] The rule ID 300 is identification information that uniquely identifies a rule related to a judgment target. For example, the judgment target is system A if the rule ID 300 is "1," system B if the rule ID 300 is "2," procedure C if the rule ID 300 is "3," and work if the rule ID 300 is "4." The natural language rule 301 is a character string that describes the content of the rule related to the judgment target in natural language. The logical expression rule 302 is a logical expression that converts the natural language rule 301 into content that can be evaluated by a computer. Note that it is sufficient for each rule to store at least one of the natural language rule 301 and the logical expression rule 302.

[0033] <Figure 4 Process Management DB202> FIG. 4 is an explanatory diagram showing an example of the process management DB 202. The process management DB 202 has fields such as a repetition count 400, an occurrence fact 401, a question 402, missing information 403, an applied rule ID 404, and a judgment result 405. A combination of values ​​in each field in the same row forms an entry that defines one process. In the initial state, no entries exist in the process management DB 202. When the process of FIG. 7, which will be described later, is executed, entries are generated in ascending order of the value of the repetition count 400.

[0034] The number of repetitions 400 is the number of times rule determination is repeated. The initial value of the number of repetitions 400 is "0", and is incremented by 1 each time rule determination is performed by the determination unit 214.

[0035] Occurrence fact 401 is a fact that occurred to a user using user terminal 230, for example, a character string indicating the user's situation. Question sentence 402 is, for example, a character string in the form of a question indicating the content of a user's inquiry about something unclear. Initially, that is, when the number of repetitions 400 is "0", occurrence fact 401 is a character string input from user terminal 230, but when the number of repetitions 400 is "1" or more, missing information 403 at that number of repetitions 400 is added.

[0036] Missing information 403 is information that is necessary for the rule condition to be met, i.e., for complying with the rule, but is not included in occurrence fact 401 and question sentence 402. Missing information 403 is generated by generation unit 215. Note that when the number of repetitions 400 is "0", no rule judgment has been performed before, so there is no missing information 403. When rule judgment is performed when the number of repetitions 400 is n (n is an integer equal to or greater than 0) and judgment result 405 shows that the condition is not met, missing information 403 with the number of repetitions 400 being n+1 is generated.

[0037] The applied rule ID 404 is the rule ID 300 of the rule that was applied to the determination result 405, which is the answer to the question 402. The determination result 405 is the result of the determination made by the determination unit 214 as to whether the occurrence fact 401 and the question 402 comply with the rule, i.e., whether they satisfy the conditions defined in the rule.

[0038] Also, the occurring fact 401 is updated until the judgment result 405 satisfies the condition. Here, the occurring fact 401 when the number of repetitions 400 is n is referred to as occurring fact 401-n, and the missing information 403 when the number of repetitions 400 is n is referred to as missing information 403-n. The occurring fact 401-1 is updated based on the occurring fact 401-0 and the missing information 403-1, and the occurring fact 401-2 is updated based on the occurring fact 401-0 and the missing information 403-2.

[0039] <Figure 5 Initial input screen> 5 is an explanatory diagram showing an example of an initial input screen. The initial input screen 500 is displayed on the user terminal 230 and accepts input information from the user terminal 230. The initial input screen 500 has a rule input area 501, a user status input area 502, a question input area 503, and an input completion button 504.

[0040] The rule input area 501 is an area for accepting input of rules, and specifically includes, for example, an additional input section 511 and a rule group application target designation section 512. The additional input section 511 is an area for accepting additional input of natural language rules 510 by the user. The rule group application target designation section 512 is a checkbox for designating a rule group recorded in the rule DB 201 as an application target.

[0041] The user situation input area 502 is an area that accepts an occurrence fact 520 as an input sentence. The question input area 503 is an area that accepts input of a question sentence 530. The input completion button 504 is a user interface that, when pressed, transmits the occurrence fact 520 input in the user situation input area 502 and the question sentence 530 input in the question input area 503 to the input support device 100.

[0042] In the rule input area 501, if the input completion button 504 is pressed when no natural language rule 510 has been input in the additional input section 511 and no check mark has been input in the rule group application target designation section 512, an error occurs. In this case, warning information (not shown) is displayed on the initial input screen 500 to prompt input in either the additional input section 511 or the rule group application target designation section 512.

[0043] Similarly, if the user status input area 502 is left blank or if the question input area 503 is left blank, an error occurs, and warning information (not shown) prompting input is displayed on the initial input screen 500.

[0044] <Figure 6 Answer screen> 6 is an explanatory diagram showing an example of an answer screen. The answer screen 600 is displayed on the user terminal 230 and accepts input information from the user terminal 230. The answer screen 600 has an answer result display area 601 and an input support information display area 602.

[0045] The answer result display area 601 displays the answer result 610. The answer result 610 displays, for example, the number of repetitions 400, the occurrence fact 520, the question sentence 530, and the determination result 405.

[0046] The input support information display area 602 is an area that displays input support information 620. The input support information display area 602 has, as the input support information 620, input support sentences 621, missing information candidates 622, an addition check box 623, and a re-input completion button 624.

[0047] The input support sentence 621 is a template for confirming to the user that the answer 611 is insufficient in information when the determination result 405 indicates that the condition is not satisfied, and that the condition is satisfied, that is, the rule is complied with, if the missing information candidate 622 exists. Note that if the determination result 405 indicates that the condition is satisfied, the input support sentence 621 is not displayed.

[0048] Missing information candidate 622 is a candidate that could become missing information 403. Add check box 623 is a user interface for specifying missing information candidate 622 as missing information 403. Re-input completion button 624 is a user interface for adopting missing information candidate 622 specified in add check box 623 as missing information 403 when pressed.

[0049] The user can check the action result for the question sentence 530 entered in the question input area 503 by referring to the answer result display area 601. When the answer result 610 displays, as the determination result 405, that the input information required for the action is insufficient (condition not met), the missing information candidate 622 is automatically displayed in the input support information 620. By using the input support information 620, the user can check the occurrence fact that was not entered in the user situation input area 502 but that should have been entered, and add the missing information candidate 622 as the occurrence fact to the occurrence fact 401 when the number of repetitions 400, which is the original input information, is "0".

[0050] <Figure 7 Input support processing> FIG. 7 is a flowchart showing an example of the input support processing procedure by the input support device 100.

[0051] (Step S701) The input support device 100 acquires input information from the user terminal 230 via the input / output unit 216, and proceeds to step S702. The input information from the user terminal 230 includes, for example, rule information (hereinafter, target rule information) input in the rule input area 501 of the initial input screen 500, an occurring fact 520 (hereinafter, target occurring fact 520) input in the user situation input area 502, and a question sentence 530 (hereinafter, target question sentence 530) input in the question input area 503.

[0052] Here, if a natural language rule 510 is input to the additional input unit 511, the natural language rule 510 is a target rule included in the target rule information. Also, if a check mark is input to the checkbox in the rule group application target designation unit 512, the rule group application target designation information is included in the target rule information. The input information is output to the management unit 212.

[0053] The management unit 212 creates a new entry in the process management DB 202, sets the number of repetitions 400 of the new entry to "0", sets the occurrence fact 401 of the new entry to the target occurrence fact 520, and sets the question 402 of the new entry to the target question 530. At this point, the missing information 403, applied rule ID 404, and judgment result 405 of the new entry have not yet been set.

[0054] (Step S702) The input support device 100 updates the target rule information in the input information using the management unit 212, and proceeds to step S703. Specifically, for example, if the target rule information includes rule group application target designation information, the management unit 212 reads all entries from the rule DB 201 and adds the rules identified by each entry to the target rule information as target rules.

[0055] Therefore, if the target rule information does not include rule group application target designation information, the target rule information will be only the natural language rule 510 input to the additional input unit 511. If the target rule information includes rule group application target designation information and the additional input unit 511 is empty, the target rule information will be the rules specified by each entry in the rule DB 201. If the target rule information includes rule group application target designation information and the natural language rule 510 has been input to the additional input unit 511, the target rule information will be the natural language rule 510 and the rules specified by each entry in the rule DB 201.

[0056] The management unit 212 adds the rule ID 300 of the target rule included in the target rule information to the applied rule ID 404 of the new entry in which the number of repetitions 400 is 0.

[0057] (Step S703) The input support device 100 extracts, by the conversion unit 213, a target rule in which the natural language rule 301 exists but the logical expression rule 302 does not exist for each target rule in the target rule information as a logical expression non-existent rule, and proceeds to step S704.

[0058] Specifically, for example, the natural language rule 510 input to the additional input unit 511 is a target rule for which there is no paired logical expression rule 302, and is therefore extracted as a logical expression non-existence rule. Also, for example, entries with rule IDs 300 of 1 to 3 in the rule DB 201 are target rules for which there is a natural language rule 301 but no logical expression rule 302, and are therefore extracted as logical expression non-existence rules.

[0059] (Step S704) The input support device 100 converts the natural language rule 301 of the logical expression non-existence rule extracted in step S703 into a logical expression rule 302 by the conversion unit 213, and adds the converted logical expression rule 302 to the entry of the logical expression non-existence rule in the rule DB 201. Then, the process proceeds to step S705.

[0060] If the logical expression non-existence rule is the natural language rule 510 input to the additional input unit 511, the conversion unit 213 assigns a new rule ID 330 to generate a new entry in the rule DB 201, and adds to the new entry the natural language rule 510 input to the additional input unit 511 and the logical expression rule 302 converted from the natural language rule 301. In this case, the management unit 212 adds the rule ID 330 assigned in the new entry to the applied rule ID 404 of a new entry in the process management DB 202 whose number of repetitions 400 is 0.

[0061] Here, an example of conversion from natural language rules 301 to logical formula rules 302 is shown. Natural language: Procedure C is necessary when a person who holds a driver's license moves into the city from outside the city. logical_expression:eligible_for_procedure_C :- possess_drivers_license; moved_from_outside_city.

[0062] In the logical expression, the character string before [:-] indicates a conditional clause, and [;] indicates logical conjunction "AND." That is, it is converted into a logical expression rule 302 in which both [possess_drivers_license] and [moved_from_outside_city] must be true for the fact [eligible_for_procedure_C] to be true.

[0063] Note that the logical expression non-existence rules in the rule DB 201 may be converted into logical expression rules 302 in advance by the conversion unit 213 executing steps S703 and S704 before executing the input support process of FIG.

[0064] (Step S705) The input support device 100 reads the occurrence fact 401 and the question 402 recorded in the process management DB 202 by the conversion unit 213, and proceeds to step S706. Specifically, for example, the conversion unit 213 reads the occurrence fact 401 (hereinafter, the target occurrence fact 401) and the question 402 (hereinafter, the target question 402) from the new entry set in the process management DB 202 in step S701.

[0065] (Step S706) The input support device 100 converts each of the target occurrence fact 401 and the target question sentence 402 into a logical formula at the current iteration count 400 by the conversion unit 213, and proceeds to step S707. Specifically, for example, the conversion unit 213 converts each of the target occurrence fact 401 and the target question sentence 402 read out in step S705 into a logical formula. Furthermore, when the target occurrence fact 401 is complemented in step S710, the conversion unit 213 converts the complemented target occurrence fact 401 into a logical formula.

[0066] As for the target question sentence 402, since the content is the same regardless of the number of repetitions 400, the conversion unit 213 converts the target question sentence 402 into a logical formula only when the number of repetitions 400 is "0", and when the number of repetitions 400 is "1" or more, the conversion unit 213 simply copies the converted logical formula from the target question sentence 402.

[0067] Here, an example of conversion from the target occurrence fact 401 to a logical formula is shown. Natural Language: I moved from outside the city. Logical expression: moved_from_outside_city.

[0068] An example of conversion from the target question sentence 402 to a logical formula is also shown. Natural sentence: Should procedure C be performed? Logical expression:?- eligible_for_procedure_C.

[0069] The "?-" at the beginning of a logical expression indicates that the following string is a question.

[0070] (Step S707) The input support device 100 inputs the logical formula of the target occurrence fact 401, the logical formula rule 302 of the target rule, and the logical formula of the target question sentence 402 to the logical inference engine via the determination unit 214, determines whether the condition of the target rule is satisfied for the combination of the target occurrence fact 401 and the target question sentence 402, and generates a determination result 405 indicating whether the condition is satisfied or not. The determination result 405 is recorded in the entry for the current iteration count 400 in the process management DB 202.

[0071] When the logical inference engine is implemented in an external computer, the input support device 100 generates a prompt requesting a determination result as to whether or not the condition of the target rule is satisfied for the combination of the target occurrence fact 401 and the target question sentence 402 by the determination unit 214, and receives the determination result 405 from the external computer.

[0072] (Step S708) The input support device 100 determines whether the determination result 405 indicates that the condition is met or not met by the determination unit 214. The condition being met means that the condition of the target rule is met with the combination of the target occurrence fact 401 and the target question sentence 402, that is, the target rule is complied with, and the condition not being met means that this is not the case, that is, the target rule is violated and there is insufficient information.

[0073] If the condition is met (step S708: Yes), the input support device 100 causes the management unit 212 to register "condition met" in the determination result 405 of the entry for the current repetition count of 400 in the process management DB 202, and proceeds to step S712. If the condition is not met (step S708: No), the input support device 100 causes the management unit 212 to register "condition not met" in the determination result 405 of the entry for the current repetition count of 400 in the process management DB 202, and proceeds to step S709.

[0074] (Step S709) The input support device 100 generates the missing information 403 by the generation unit 215, with reference to the rules in the rule DB 201. Specifically, for example, at the current iteration count 400, the generation unit 215 inputs the rules in the rule DB 201 and the target occurrence fact 401 recorded in the process management DB 202 into a machine learning model, generates the missing information 403 that is not included in the target occurrence fact 401 among the information necessary to comply with the rules, and stores the missing information 403 in the process management DB 202.

[0075] When the machine learning model is implemented in an external computer as a generation AI, the input support device 100 generates, by the generation unit 215, a prompt that requests the other computer to provide missing information 403 for the combination of the rule in the rule DB 201 and the target occurrence fact 401 recorded in the process management DB 202 at the current iteration count 400. Upon receiving the prompt, the other computer returns the missing information 403 to the generation unit 215 by using the generation AI.

[0076] In step S709, if the number of repetitions 400 is "0", "It has been less than three months since the move" is generated as missing information 403 (the number of repetitions 400 is stored in the entry with "1"); if the number of repetitions 400 is "1", "I have a driver's license" is generated as missing information 403 (the number of repetitions 400 is stored in the entry with "2").

[0077] (Step S710) The input support device 100 causes the management unit 212 to complement the target occurrence fact 401 based on the missing information 403. Specifically, for example, the management unit 212 generates a target occurrence fact 401 whose number of repetitions 400 is n+1 by linking the missing information 403 generated in step S709 to the target occurrence fact 401 whose number of repetitions 400 is "0."

[0078] For example, when the number of repetitions 400 is "0," assume that "It has been less than three months since moving." is generated as missing information 403 in step S709. The management unit 212 concatenates "It has been less than three months since moving." as missing information 403 in step S709 to "I recently moved from outside the city.", which is the target occurrence fact 401 with the number of repetitions 400 of "0," and updates the target occurrence fact 401 to "I recently moved from outside the city. It has been less than three months since moving." Therefore, in step S706, the supplemented target occurrence fact 401 is converted into a logical formula.

[0079] (Step S711) The input support device 100 increments the number of repetitions 400 by the management unit 212, updates the process management DB 202, and returns to step S706. Specifically, for example, the management unit 212 increments the number of repetitions 400 and creates a new entry in the process management DB 202. The management unit 212 registers the target occurrence fact 401 completed in step S710 in the new entry, duplicates the target question sentence 402, registers the missing information 403 generated in step S709, and duplicates the applied rule ID 404.

[0080] (Step S712) The input support device 100 causes the management unit 212 to output the target occurrence fact 401 for which the judgment result 405 indicates that the condition is satisfied, in a displayable manner on the user terminal 230. Specifically, for example, the management unit 212 outputs the target occurrence fact 401 for which the number of repetitions 400 is "2" in the process management DB 202, via the input / output unit 216, in a displayable manner on the user terminal 230.

[0081] As a result, the user terminal 230 updates the number of repetitions 400 to "2" in the answer result 610 in the answer result display area 601, updates the occurring fact 520 with the target occurring fact 401 for which the number of repetitions 400 is "2" from the input support device 100, and updates the judgment result 405 to "condition met." In this way, the result of complementing the information deficiency is presented to the user as the target occurring fact 401.

[0082] In step S712, the input support device 100, by the management unit 212, outputs the target occurrence fact 401 for which the judgment result 405 is that the condition is met (step S708: Yes) in a displayable manner to the user terminal 230. However, even if the condition is not met (step S708: No), the input support device 100 may output the judgment result 405 of the current number of repetitions 400 and the target occurrence fact 520 at the current number of repetitions 400, "condition not met", in a displayable manner to the user terminal 230 via the input / output unit 216. This makes it possible to present to the user the intermediate process in which the condition was not met.

[0083] As described above, according to this embodiment, the input information is supplemented by generating missing information 403 that is necessary to comply with the rules and is not included in the input information (occurrence fact 520) from the user terminal 230. Therefore, when part or all of the information necessary for an action is not included in the input information (occurrence fact 520) from the user terminal 230, the input information (occurrence fact 520) can be supplemented to include all of the information necessary for the user's action.

[0084] The present invention is not limited to the above-described embodiments, and includes various modifications and equivalent configurations within the spirit and scope of the appended claims. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to configurations including all of the described configurations. Furthermore, part of the configuration of one embodiment may be replaced with the configuration of another embodiment. Furthermore, the configuration of another embodiment may be added to the configuration of one embodiment. Furthermore, part of the configuration of each embodiment may be added to, deleted from, or replaced with other configurations.

[0085] Furthermore, the aforementioned configurations, functions, processing units, processing means, etc. may be realized in part or in whole in hardware, for example by designing them as integrated circuits, or may be realized in software by having a processor interpret and execute a program that realizes each function.

[0086] Information such as programs, tables, files, etc. that realize each function can be stored in storage devices such as memory, hard disks, SSDs (Solid State Drives), or recording media such as IC (Integrated Circuit) cards, SD cards, and DVDs (Digital Versatile Discs).

[0087] In addition, the control lines and information lines shown are those that are considered necessary for explanation, and do not necessarily represent all the control lines and information lines that are necessary for implementation. In reality, it can be assumed that almost all components are interconnected. [Explanation of symbols]

[0088] 100 Input support device 101 processors 102 Storage Devices 211 Data transmission and reception unit 212 Management Department 213 Conversion Unit 214 Judgment section 215 Generation part 216 Input / output section 230 User Terminals

Claims

1. an input unit for inputting an input sentence and a question sentence regarding the object to be determined; a determination unit that determines whether a predetermined condition defined in a rule related to the object to be determined is satisfied for a combination of the input sentence and the question sentence; a generation unit that generates missing information that is not included in the input sentence among information required to satisfy the predetermined condition when the determination unit determines that the predetermined condition is not satisfied; a management unit that completes the input sentence using the missing information generated by the generation unit; an output unit that outputs the determination result determined by the determination unit; An input support device comprising:

2. 2. The input support device according to claim 1, the determination unit determines whether the predetermined condition is met with a combination of the input sentence completed by the management unit and the question sentence; the output unit outputs the completed input sentence, for which the determination unit has determined that the predetermined condition is satisfied, as the determination result. An input support device comprising:

3. 2. The input support device according to claim 1, the determination unit determines whether the predetermined condition is met with a combination of the input sentence completed by the management unit and the question sentence; the output unit outputs the completed input sentence, for which the determination unit has determined that the predetermined condition does not hold, as the determination result. An input support device comprising:

4. 2. The input support device according to claim 1, The system is capable of communicating with a computer that implements a generation AI that generates an answer to a query, the generation unit generates a query requesting the missing information and transmits the query to the computer, and then acquires the missing information from the computer. An input support device comprising:

5. 2. The input support device according to claim 1, the determination unit determines whether or not the predetermined condition is satisfied for a combination of the input sentence and the question sentence, based on a logical formula of the input sentence, a logical formula of the question sentence, and a logical formula of the rule. An input support device comprising:

6. 6. The input support device according to claim 5, a storage unit that stores the rules; a conversion unit that converts natural language sentences into logical expressions, the conversion unit converts each of the input sentence and the question sentence into the logical formula, and if the rule is stored in natural language and not in logical formula, converts the rule into the logical formula; An input support device comprising:

7. An input assistance method executed by an input assistance device having a processor that executes a program and a storage device that stores the program, The processor: an input process for inputting an input sentence and a question sentence regarding the object to be determined; a determination process for determining whether a predetermined condition defined in a rule relating to the determination target is satisfied for a combination of the input sentence and the question sentence; an output process for outputting the determination result determined by the determination process; An input assistance method comprising:

8. The processor an input process for inputting an input sentence and a question sentence regarding the object to be determined; a determination process for determining whether a predetermined condition defined in a rule relating to the determination target is satisfied for a combination of the input sentence and the question sentence; an output process for outputting the determination result determined by the determination process; An input support program characterized by executing the above.

Images