Processor system

Abstract

To provide a processor system which reduces the burden of answering similar questions.SOLUTION: In a question answering system 10, a processor system 100 comprising a memory 110, processing unit 120, Input / output interface 130, and transmission interface 140 is configured to use a questionnaire including a plurality of predetermined questions and information on answers to the questions to supplement information on answers to questions in a second questionnaire different form the questionnaire.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a processor system.

Background Art

[0002] In recent years, the ESG (Environment, Social, Governance) investment market has been on an expanding trend. In the ESG investment market, investors evaluate companies based on ESG factors and determine investment destinations. This is because companies that emphasize ESG are expected to have stability in future management and high growth potential.

[0003] Evaluations related to ESG, including environmental issues and child labor issues, etc., are being conducted not only within the company but also across the entire supply chain in recent years. For this reason, buyers often hope to conduct ESG evaluations of business entities (also referred to as suppliers) participating in the supply chain system.

[0004] Such supplier ESG evaluations for buyers mainly involve obtaining information by conducting questionnaire surveys on suppliers.

[0005] On the other hand, for suppliers, the burden of answering questionnaire surveys is large. The number of questions can range from dozens to hundreds, and some questions require attaching evidence data as proof for the answers. Also, there are multiple evaluation institutions, and each sets similar but not completely identical questions. As a result, the response rate from suppliers tends to be low.

[0006] Also, the tendency of questions is that many of the underlying standards, such as ISO (International Organization for Standardization) 26000, ISO14000 series, and the United Nations Global Compact, etc., are common, and the differences in questions are often differences in resolution in many cases.

[0007] In this context, a common technique for assisting in the creation of answers involves sequentially acquiring the content entered in the answer fields, retrieving predetermined keywords and related information corresponding to those keywords from a storage unit, and displaying them on a display unit. For example, there is a technique described in Japanese Patent Publication No. 2020-012946.

[0008] Furthermore, in the field of natural language processing, feature extraction methods such as Bag of Words, TF-IDF, BM-25, and N-gram are generally known as techniques for recognizing commonalities in texts. [Prior art documents] [Patent Documents]

[0009] [Patent Document 1] Japanese Patent Publication No. 2020-012946 [Overview of the Initiative] [Problems that the invention aims to solve]

[0010] In the technology described in Patent Document 1 mentioned above, relevant information is displayed according to the content entered when the answer is entered. However, since the answer must be entered manually, it cannot be said that the burden of answer input is significantly reduced. This is because assistance in entering the answer is only provided at the stage when the user has read the question, accurately understood the question, considered the content of the answer, and constructed an outline of the answer.

[0011] Furthermore, the vocabulary used in the questions provided for ESG evaluation includes expressions unique to ESG evaluation. For example, "Greenhouse Gas (GHG)" means greenhouse gases. Various expressions such as "GHGs," "GHG emissions," "GHG emissions," and "Scope 1, 2, and 3 emissions" are used in the questions, but all of them mean greenhouse gas emissions. In addition, carbon dioxide emissions, which are one of the greenhouse gases, are also included in this. Also, expressions such as "green procurement" and "sustainable procurement," and "corruption prevention" and "maintaining integrity" are considered to have the same meaning. However, because some or all of the sentence components or the meaning of individual words do not match, existing feature extraction methods will not recognize them as the same and will classify them as separate questions.

[0012] The objective of this invention is to reduce the burden of answering similar questions. [Means for solving the problem]

[0013] To solve the above problems, this application employs, for example, the means described in the claims. The present invention includes multiple means for solving the above problems, but to give one example, a processor system having at least one memory and one or more processors, wherein the processor uses a questionnaire containing a plurality of predetermined questions and answer information for those questions to supplement the answer information for questions in a second questionnaire that is different from the first questionnaire.

[0014] Furthermore, the above processor system links words or sentences that are not originally recognized as synonyms by registering them in the synonym dictionary memory area. In addition, words or sentences that are of particular priority in the synonym dictionary memory area are learned as priority words or priority sentences. Moreover, when these priority words or sentences appear, the system prioritizes supplementing the answer information to the question. [Effects of the Invention]

[0015] According to the present invention, it is possible to provide a technique for reducing the burden of answering similar questions. Other problems, configurations, and effects than those described above will be clarified by the description of the embodiments for carrying out the following invention.

Brief Description of the Drawings

[0016] [Figure 1] It is a diagram showing a configuration example of a question-and-answer system. [Figure 2] It is a diagram showing an example of the data structure of a question material storage area. [Figure 3] It is a diagram showing an example of the data structure of a thesaurus dictionary storage area. [Figure 4] It is a diagram showing an example of the data structure of an answer history storage area. [Figure 5] It is a diagram showing an example of the hardware configuration of a processor system. [Figure 6] It is a diagram showing an example of a flowchart of answer support processing (survey agency). [Figure 7] It is a diagram showing an example of a flowchart of answer completion processing. [Figure 8] It is a diagram showing an example of a flowchart of answer support processing (assisting the respondent). [Figure 9] It is a diagram showing an example of answer completion. [Figure 10] It is a diagram showing an example of presenting an answer. [Figure 11] It is a diagram showing another configuration example of a question-and-answer system. [Figure 12] It is a diagram showing an example of a flowchart of answer support processing (using a learned model). [Figure 13] It is a diagram showing an example of a flowchart of answer completion processing (using a learned model). [Figure 14] It is a diagram showing an example of answer completion (using a learned model).

Embodiments for Carrying Out the Invention

[0017] Embodiments of the present invention will be described below with reference to the drawings. The embodiments are illustrative examples for explaining the present invention, and have been omitted and simplified as appropriate for clarity of explanation. The present invention can also be carried out in various other forms. Unless otherwise specified, each component may be singular or plural.

[0018] The positions, sizes, shapes, and ranges of the components shown in the drawings may not represent their actual positions, sizes, shapes, and ranges in order to facilitate understanding of the invention. Therefore, the present invention is not necessarily limited to the positions, sizes, shapes, and ranges disclosed in the drawings.

[0019] Examples of various types of information may be described using expressions such as "table," "list," and "queue," but these types of information may also be represented by data structures other than these. For example, various types of information such as "XX table," "XX list," and "XX queue" may be referred to as "XX information." When describing identification information, expressions such as "identification information," "identifier," "name," "ID," and "number" are used, but these are interchangeable. Furthermore, the identification information described using these expressions is represented in the examples using symbols, numbers, natural language, or combinations thereof, but the identification information may also be in other formats.

[0020] When there are multiple components with the same or similar function, they may be described using the same symbol but with different subscripts. Furthermore, if it is not necessary to distinguish between these multiple components, the subscripts may be omitted in the description.

[0021] In the examples, the processes performed by executing a program may be described. Here, the computer executes the program using a processor (e.g., CPU, GPU) and performs the processing defined in the program using memory resources (e.g., memory) and interface devices (e.g., communication ports). Therefore, the main entity performing the processing by executing the program may be the processor. Similarly, the main entity performing the processing by executing the program may be a controller, device, system, computer, or node having a processor. The main entity performing the processing by executing the program may be an arithmetic unit, and may include dedicated circuits that perform specific processing. Here, dedicated circuits include, for example, FPGAs (Field Programmable Gate Arrays), ASICs (Application Specific Integrated Circuits), CPLDs (Complex Programmable Logic Devices), etc.

[0022] The program may be installed on the computer from the program source. The program source may be, for example, a program distribution server or a storage medium readable by the computer. If the program source is a program distribution server, the program distribution server includes a processor and storage resources for storing the program to be distributed, and the processor of the program distribution server may distribute the program to other computers. In addition, in the embodiment, two or more programs may be implemented as one program, or one program may be implemented as two or more programs.

[0023] Figure 1 shows an example of the configuration of a question answering system. For example, the question answering system 10 includes a processor system 100, a network 50, a computer 300 for evaluation organization D, a computer 310 for evaluation organization E, a computer 400 for supplier A, a computer 410 for supplier B, a computer 420 for supplier C, a computer 800 for buyer F, and a computer 810 for buyer G.

[0024] Network 50 is, for example, a communication network that uses public lines such as LAN (Local Area Network), WAN (Wide Area Network), VPN (Virtual Private Network), or the Internet in part or in whole, or a mobile phone communication network, or a combination thereof. Network 50 may also be a wireless communication network such as Wi-Fi (registered trademark) or 5G (Generation).

[0025] Evaluation organizations D and E are examples of organizations that evaluate suppliers, which are entities that provide parts and products in supply chain networks and other parts supply networks. While there are usually more than just these two evaluation organizations, in this embodiment, these two are used as examples of evaluation organizations for the sake of simplicity.

[0026] Suppliers A, B, and C are examples of suppliers that provide parts and products in a parts supply network, such as a supply chain network. Suppliers are not limited to just these three organizations; there are usually many more. However, for the sake of simplicity in this embodiment, these three organizations are referred to as suppliers.

[0027] Buyers F and G are examples of buyers that purchase parts and products in a parts supply network such as a supply chain network. Buyers are not limited to just two organizations; there are usually many more. However, for the sake of simplicity in this embodiment, these two organizations are referred to as buyers. When managing and selecting suppliers, buyers may use the evaluation organizations, or they may evaluate suppliers themselves without using them. In this embodiment, for the sake of simplicity, suppliers and buyers are described separately, but when a supplier purchases parts or products, the supplier may also be a buyer, and when a buyer provides parts or products to another buyer, the buyer may also be a supplier.

[0028] The processor system 100 includes a memory 110, a processing unit 120, an input / output interface 130, and a transmission interface 140. The memory 110 includes a question data storage area 111, a synonym dictionary storage area 112, and an answer history storage area 113. The processing unit 120 includes a question data receiving unit 121, an answer support unit 122, an answer receiving unit 123, and an optimization unit 124. The answer support unit 122 includes a classifier 1221 and a translator 1222. The processor system 100 is a system having one or more processors.

[0029] Figure 2 shows an example of the data structure of the question data storage area. The question data storage area 111 stores information about questions to suppliers. Specifically, the question data storage area 111 contains the issuing organization ID 111a, the document name 111b, the response period 111c, the target supplier 111d, and the question 111e. The issuing organization ID 111a, the document name 111b, the response period 111c, the target supplier 111d, and the question 111e are all associated with each other.

[0030] Issuer ID 111a stores information identifying the issuer ID, which is identification information that identifies the issuer of the question. In this embodiment, in the case of non-financial information questionnaires and CSR questionnaires, the issuer is evaluation organization D and evaluation organization E, and in the case of SAQ (Self-Assessment Questionnaire), etc., the issuer is buyer F. However, evaluation organizations may provide self-assessment questionnaires, and buyers may issue non-financial information questionnaires and CSR questionnaires.

[0031] The document name 111b stores the name of the document containing the questions. The document name may vary depending on the issuing organization, but in this embodiment, it is such as "Non-Financial Information Questionnaire," "CSR (Corporate Social Responsibility) Questionnaire," or "Self-Assessment Questionnaire." However, it is not limited to these, and any document that requests answers regarding non-financial information such as ISO 26000 or ISO 14000 series is acceptable. Many of these documents include questions issued by evaluation organizations or buyers with suppliers as respondents.

[0032] The response period 111c contains information that identifies the period for which the document containing the question is intended to be answered. Since evaluation organizations often require suppliers to submit results for the previous year once a year, the response period 111c stores information that identifies the previous year. However, if the question documents are issued at different frequencies, the response period 111c stores information that identifies the period corresponding to the response period (first half, second half, 1Q, etc.).

[0033] The responding supplier 111d stores information that identifies the entity responsible for responding to the document specified by the document name 111b.

[0034] Question 111e stores the question text (natural language, indicators, or mathematical formulas) contained in the document identified by document name 111b. In Figure 2, there are only two questions for simplification, but in reality there are several to several hundred questions.

[0035] Figure 3 shows an example of the data structure of the synonym dictionary memory area. The synonym dictionary memory area 112 stores the relationships between words used in the questions that are unlikely to be considered synonyms because some or all of the sentence components or the meaning of individual words do not match, but are actually synonyms or contain the same meaning. The synonym dictionary memory area 112 also stores the relationships between words that are considered synonyms in general dictionaries, etc. Specifically, the synonym dictionary memory area 112 has a meaning ID 112a, a word 112b, a registration year 112c, and a priority word / sentence 112d. Although we will continue to refer to them as words or phrases, the synonym dictionary can store not only words but also sentences.

[0036] The meaning ID 112a, word 112b, registration year 112c, and priority word / sentence 112d are associated with each other.

[0037] Semantic ID 112a stores an identifier assigned according to the meaning of a specific word. Words that share this identifier are considered synonyms.

[0038] Word 112b stores words. Words are not limited to linguistic terms such as words and morphemes, but also broadly include neologisms and phrases and expressions used as a set. Differences between different languages ​​are also treated as synonyms. For example, "carbon dioxide emissions" in Japanese and "CO2Emissions" in English are both synonymous, meaning carbon dioxide emissions. Similarly, "GHGs" and "GHG emissions" are originally abbreviations for greenhouse gases and greenhouse gas emissions, respectively, but since carbon dioxide emissions, which are one of the greenhouse gases, are also included, defining them as synonyms prevents omissions.

[0039] Furthermore, "Scope 1, 2, and 3" are terms commonly used to indicate the range of greenhouse gas emissions to be measured, as defined in the GHG Protocol, the international standard for calculating and reporting greenhouse gas emissions. Therefore, these terms are also used in the question, and "Scope 1, 2, and 3 emissions" are synonymous with greenhouse gas emissions. In this embodiment, these terms are assigned the semantic ID "7654321" to indicate their common meaning. In addition, although not shown in this figure, terms such as "green procurement" and "sustainable procurement," and "corruption prevention" and "maintaining integrity" do not have some or all of the meaning of the sentence components or individual words, and are not usually considered synonymous by existing feature extraction methods. However, in the synonym dictionary memory area 112 of this embodiment, they have the same semantic ID.

[0040] The registration year 112c stores information that identifies the year in which the word identified by word 112b was registered in the synonym dictionary memory area.

[0041] Priority words / phrases 112d are assigned an ID only if word 112b is a priority word / phrase as described below.

[0042] Figure 4 shows an example of the data structure of the response history storage area. The response history storage area 113 stores response information obtained from survey respondents. In this embodiment, the respondent is one of supplier A, supplier B, or supplier C. The response history storage area 113 includes the respondent supplier 113a, the question 113b, the response period 113c, and the response 113d.

[0043] The answer supplier 113a, question 113b, response period 113c, and answer 113d are all related to each other.

[0044] The response supplier 113a stores information that uniquely identifies the entity that answered the question.

[0045] Question 113b contains the question text (natural language, indicators, or mathematical formulas).

[0046] The response period 113c contains information that identifies when the response to the provided document was made. Since evaluation organizations often require suppliers to submit results for the previous year once a year, the response period 113c stores information that identifies the year in which the response was submitted. However, if the document is issued at different frequencies, the response period 113c will store information that identifies the corresponding period (first half, second half, 1Q, etc.) for the response.

[0047] Answer 113d contains information that identifies the answer to the question regarding the provided materials. If the answer includes evidence data, that evidence data is also stored in Answer 113d.

[0048] Returning to the explanation of Figure 1, the Question Data Reception Unit 121 of the Processing Unit 120 receives Question Data from Evaluation Organization D Computer 300, Evaluation Organization E Computer 310, Buyer F Computer 800, and Buyer G Computer 810. The Question Data Reception Unit 121 breaks down the received Question Data into question units, reconstructs them in the Question Data Storage Area 111, and stores them there.

[0049] In this embodiment, the pattern in which the processor system 100 receives questionnaire materials from an evaluation organization or buyer is referred to as the survey outsourcing method or simply survey outsourcing. In this embodiment, the pattern in which the processor system 100 receives questionnaire materials from suppliers A, B, and C is referred to as the respondent assistance method or simply respondent assistance. The difference between the two is whether the questionnaire answering system 10 handles the tasks from receiving the questionnaire materials to conducting the survey and submitting the answers, or whether the suppliers themselves take the lead in these tasks. Despite this difference in method, the questionnaire answering system 10 can reduce the burden of answering similar questions. However, this difference in method causes some variation in the flow of the response support processing, which will be described in more detail later.

[0050] The response support unit 122 uses a questionnaire containing multiple predetermined questions and the response information to those questions to supplement the response information for questions in a second questionnaire that is different from the first questionnaire. Here, the questionnaire includes questions that investigate company information. In supplementing the information, the response support unit 122 uses various existing feature extraction methods, such as Bag of Words, TF-IDF, BM-25, and N-gram, either individually or in combination, to recognize commonalities between the questions and generate appropriate feature vectors from the question strings. Furthermore, the response support unit 122 classifies the feature vectors generated from the questions into the same class if the feature vectors can be considered to have a common meaning with the original questions, and assigns a classification ID to them.

[0051] The answer support unit 122 has a classifier 1221 that learns multiple patterns of the relationship between this feature vector and classification ID, and can predict classification IDs for feature vectors newly created from unknown questions. Within the answer support unit 122, this classifier 1221 predicts classification IDs using methods such as a Support Vector Machine, decision tree, or k-nearest neighbors. In other words, the answer support unit 122 determines that the questions have the same meaning when the classifier 1221 determines that they have the same classification ID. By using a thesaurus, the answer support unit 122 can recognize words or sentences that are not considered synonyms by existing feature extraction methods, even if some or all of the meanings of the sentence components or individual words do not match.

[0052] Furthermore, the response support unit 122 learns words or phrases that are of particular priority in the synonym dictionary memory area 112 as priority words (phrases). Specifically, the response support unit 122 assigns a common ID to the priority words / phrases 112d shown in Figure 3, which are particularly priority words or phrases. This allows the response support unit 122 to prioritize the presence or absence of the word (phrase) when it appears and perform common recognition. Using these methods, the response support unit 122 identifies similar questions and transcribes the answers. Even if all or part of the question is in a different language, the response support unit 122 uses its built-in translator 1222 to translate it and perform the same processing. In addition, if there is document information related to the answer information, the response support unit 122 associates and supplements that related document information when processing the answer information to supplement the questions in the second questionnaire.

[0053] Furthermore, the response support unit 122 supplements the data by identifying and transcribing response information from the source questionnaire that is similar to the questions in the second questionnaire. In particular, if the questions in the second questionnaire to be supplemented are questions concerning non-financial information related to management and conform to a prescribed international standard, the response support unit 122 transcribes the information in this manner.

[0054] However, the response support unit 122 excludes questions in the questionnaire that meet certain conditions from being included in the supplementation of response information. These conditions are that the response period of the original questionnaire and the second questionnaire to which the information is being supplemented are not the same, the response information is numerical, and the response information requires related documentary information as evidence. Questions that meet all of these conditions are clearly not suitable for supplementation.

[0055] The response reception unit 123 receives from suppliers A, B, and C whether or not there have been any corrections to the supplementary response information for the questions in the second questionnaire, and the corrected response information.

[0056] The optimization unit 124 uses the response information received by the response reception unit 123 to generate or modify a thesaurus that links similar words between the words in the questions of the first questionnaire and the words in the questions of the second questionnaire, and also modifies the classifier 1221.

[0057] The input / output interface 130 accepts various types of data input and output. Specifically, the input / output interface 130 accepts input from suppliers regarding the question materials.

[0058] The transmission interface 140 communicates with other devices via the network 50. These other devices include evaluation organization D's computer 300, evaluation organization E's computer 310, supplier A's computer 400, supplier B's computer 410, supplier C's computer 420, buyer F's computer 800, and buyer G's computer 810.

[0059] Figure 5 shows an example of the hardware configuration of a processor system. The processor system 100 can be realized as a general-purpose computer 900 or a network system comprising multiple such computers 900, which includes a processor (for example, a CPU: Central Processing Unit or a GPU: Graphics Processing Unit) 901, hardware memory 902 such as RAM (Random Access Memory), external storage devices 903 such as a Hard Disk Drive (HDD) or SSD (Solid State Drive), a reader 905 that reads information from a portable storage medium 904 such as a CD (Compact Disk) or DVD (Digital Versatile Disk), an input device 906 such as a keyboard, mouse, barcode reader, or touch panel, an output device 907 such as a display, and a communication device 908 that communicates with other computers via a communication network such as a LAN or the Internet. Note that the reader 905 may be capable of not only reading but also writing to the portable storage medium 904.

[0060] The processor 901 performs various processes by executing a predetermined question-answering program loaded into memory 902 from the external storage device 903. The question-answering program is, for example, an application program that can be executed on an OS (Operating System) program. The question-answering program may be installed in the external storage device 903 from a portable storage medium 904 via a reader device 905, or it may be downloaded from a network via a communication device 908 and executed by the processor 901.

[0061] For example, the question data reception unit 121, the answer support unit 122, the answer reception unit 123, and the optimization unit 124 can be implemented by loading the question answer program stored in the external storage device 903 into memory 902 and executing it with the processor 901. The input / output interface 130 can be implemented by the processor 901 using the input device 906, the output device 907, and the communication device 908. The memory 110 can be implemented by the processor 901 using memory 902 or the external storage device 903. The transmission interface 140 can be implemented by the processor 901 using the communication device 908.

[0062] Figure 6 shows an example flowchart of the response support process (survey outsourcing). The response support process (survey outsourcing) is initiated when a start instruction is received from the evaluation organization, buyer, etc. Alternatively, the response support process (survey outsourcing) may be initiated at a predetermined date and time (for example, 6:00 a.m. every day) or at predetermined intervals (for example, every 12 hours). The response support process (survey outsourcing) is performed when the question response system 10 handles the tasks from receiving the question materials to conducting the survey and submitting the responses.

[0063] First, the question material receiving unit 121 receives and stores the question material from the evaluation organization or buyer (step S101). Specifically, the question material receiving unit 121 receives the question material from evaluation organization computer D 300, evaluation organization computer E 310, buyer computer F 800, and buyer computer G 810. The question material receiving unit 121 breaks down the received question material into question units, reconstructs them in the question material storage area 111, and stores them there.

[0064] Then, the response support unit 122 performs response completion processing based on the response history for each supplier (step S102). The specific processing will be described later using Figure 7.

[0065] Then, the response support unit 122 sends the response, including the proposed answer, to the supplier (step S103).

[0066] Then, the response receiving unit 123 receives responses from the suppliers (step S104). Specifically, the response receiving unit 123 receives responses, attached documents, and revisions to the proposed responses from supplier A's computer 400, supplier B's computer 410, and supplier C's computer 420.

[0067] The optimization unit 124 then stores the received answers as answer history in the answer history storage area 113 (step S105). The optimization unit 124 then analyzes the modifications to the answer proposals and corrects one or more of the following: any deficiencies or errors in the word similarity relationships in the synonym dictionary storage area 112, or the classifier 1221. For example, if there is an error in the word similarity relationships in the synonym dictionary storage area 112, for questions where the answer proposal has been modified, words that are considered to have the same meaning in the synonym dictionary storage area 112 may not actually have the same meaning. Therefore, the optimization unit 124 reassigns meaning IDs so that they do not refer to the same meaning ID, and updates the synonym dictionary storage area 112 so that they are not considered synonyms thereafter. If there is an error in the classifier 1221, the error may not be in the word itself, but rather in the classification of the feature vectorized question, for example. Therefore, the optimization unit 124 corrects the classification (classification ID) of the question that was determined to be incorrect, and adds a new classification ID if necessary.

[0068] Then, the response support unit 122 transmits the response received from the supplier to the evaluation institution or the buyer that requested the investigation (step S106). Specifically, the response support unit 122 transmits the response received from the supplier to the evaluation institution D computer 300, the evaluation institution E computer 310, the buyer F computer 800, and the buyer G computer 810.

[0069] The above is an example of the flowchart of the response support process (investigation agency). According to the response support process (investigation agency), the burden of answering similar questions can be reduced.

[0070] FIG. 7 is a diagram showing an example of the flowchart of the response completion process. The response completion process starts in step S102 of the response support process (investigation agency).

[0071] First, the response support unit 122 determines whether there is a response history of the same supplier in the same period (step S201). Specifically, the response support unit 122 refers to the response history storage area 113 and determines whether the response supplier 113a is the same as the respondent of the questionnaire to be supplemented, and the response period 113c is the same period as the response period of the questionnaire to be supplemented, and whether it is a question 113b similar to the questionnaire to be supplemented. If there is a response history of the same supplier in the same period (when "Yes" in step S201), the response support unit 122 proceeds to step S205.

[0072] If there is no response history of the same supplier in the same period (when "No" in step S201), the response support unit 122 determines whether numerical evidence is required for the response (step S202). Specifically, the response support unit 122 determines whether a predetermined word such as "specific basis", "attachment", "data", "numerical value", etc. is included in the question. If it is included, it is determined that numerical evidence is required for the response. If numerical evidence is not required for the response (when "No" in step S202), the response support unit 122 proceeds to step S205.

[0073] If numerical evidence is required for the answer (if "Yes" is answered in step S202), the answer support unit 122 determines whether the question requires a numerical answer (step S203). Specifically, the answer support unit 122 determines whether the question contains predetermined words such as "specific value" or "unit," and if so, determines that the question requires a numerical answer. If the question does not require a numerical answer (if "No" is answered in step S203), the answer support unit 122 proceeds to step S205.

[0074] If a question requires a numerical answer (in the case of "Yes" in step S203), the answer support unit 122 will not provide a sample answer for that question (step S204). This is because it is clear that the question is not suitable for providing a sample answer.

[0075] If there is a response history from the same supplier for the same period (if the answer is "Yes" in step S201), if numerical evidence is not required for the answer (if the answer is "No" in step S202), or if the question does not require a numerical answer (if the answer is "No" in step S203), the response support unit 122 supplements the answer to the question from the response history (step S205). Specifically, the response support unit 122 identifies similar questions in the response history and transcribes the answers to those questions and the associated document information. If there are multiple similar questions, the response support unit 122 transcribes multiple answers so that the respondent can select which answer is appropriate. In this case, the response support unit 122 presents or displays the answers in order of their likelihood of being appropriate. For example, if there is a response history for the same period, the response support unit 122 presents the response history for the same period first, as it is more likely to be appropriate than a response history from a different period. Furthermore, if there is no response history from the same supplier for the same period (i.e., the response was received via "No" in step S201), the response support unit 122 will add supplementary information indicating that the supplemented response is not from the same period. This is because, although the response is supplemented, since the source of the supplement is not from the same period, the response may not be suitable for supplementation.

[0076] The answer support unit 122 then determines whether or not it has made a decision to complete all questions (step S206). If it has made a decision to complete all questions (if the answer is "Yes" in step S206), the answer support unit 122 terminates the answer completion process. If it has not made a decision to complete all questions (if the answer is "No" in step S206), the answer support unit 122 returns control to step S201 with respect to the unprocessed questions.

[0077] The above is an example of a flowchart for answer completion processing. According to the answer completion processing, an answer form containing multiple predetermined questions and the answer information (answer history) for those questions can be used to complete the answer information for questions in a second questionnaire that is different from the first questionnaire.

[0078] Figure 8 shows an example flowchart of the response support process (respondent assistance). The response support process (respondent assistance) is basically the same as the response support process (survey outsourcing), but it is a process that is carried out in the case of a respondent assistance method in which the supplier itself takes the lead in tasks from receiving the question materials to conducting the survey and submitting the responses. Therefore, there is a difference in that the processor system 100 of the question response system 10 receives the question materials from the supplier rather than from the evaluation organization or buyer, and does not send the responses to the evaluation organization.

[0079] The response support process (respondent assistance) is initiated when a start instruction is received from a supplier or the like. First, the question material receiving unit 121 receives and stores the question material from the supplier (step S111). Specifically, the question material receiving unit 121 receives the question material from either supplier A's computer 400, supplier B's computer 410, or supplier C's computer 420. The question material receiving unit 121 breaks down the received question material into question units, reconstructs them in the question material storage area 111, and stores them.

[0080] Then, the answer support unit 122 performs answer completion processing on the question materials based on the answer history (step S112). The specific processing is the same as the answer completion processing shown in Figure 7.

[0081] Then, the response support unit 122 sends the response, including the proposed answer, to the supplier (step S113).

[0082] Then, the response receiving unit 123 receives the responses from the suppliers via broadcast (step S114). Specifically, the response receiving unit 123 receives the responses, attached documents, and revisions to the draft responses from either supplier A's computer 400, supplier B's computer 410, or supplier C's computer 420. This is because the suppliers broadcast their responses to the processor system 100 when sending them to the evaluation organization and the buyer.

[0083] The optimization unit 124 then stores the received answers as answer history in the answer history storage area 113 (step S115). The optimization unit 124 then analyzes the modifications to the answer proposals and corrects one or more of the following: deficiencies or errors in the word similarity relationships in the synonym dictionary storage area 112, or errors in the classifier 1221. For example, if there is an error in the word similarity relationships in the synonym dictionary storage area 112, for questions where the answer proposal has been modified, words that are considered to have the same meaning in the synonym dictionary storage area 112 may not actually have the same meaning. Therefore, the optimization unit 124 reassigns meaning IDs so that they do not refer to the same meaning ID, and updates the synonym dictionary storage area 112 so that they are not considered synonyms thereafter. If there is an error in the classifier 1221, the error may not be in the word itself, but rather in the classification of the question, for example, in the feature vectorization. Therefore, the optimization unit 124 corrects the classification (classification ID) of the question that was determined to be erroneous, and adds a new classification ID if necessary.

[0084] The above is an example of a flowchart for response support processing (respondent assistance). According to response support processing (respondent assistance), even if the supplier does not conduct the survey on behalf of the respondents, the burden of answering similar questions can be reduced.

[0085] FIG. 9 is a diagram showing an example of answer completion. Answer completion example 500 is an example in which an answer is completed by answer support processing (survey agency proxy) or answer support processing (respondent assistance). In answer completion example 500, processor system 100 uses the history of responses given by supplier A401 to each question in questionnaire 402 issued by evaluation agency D to complete answers to questions in questionnaire 403 issued by evaluation agency E. That is, according to answer support processing (survey agency proxy) or answer support processing (respondent assistance), supplier A401 can obtain answer completion for questionnaire 403 issued by evaluation agency E based on the content of the answers given to questionnaire 402 issued by evaluation agency D.

[0086] FIG. 10 is a diagram showing an example of presenting an answer. Answer presentation example 600 is an example of a screen display of an answer created by answer support processing (survey agency proxy) or answer support processing (respondent assistance). In answer presentation example 600, an answer is presented for each question in questionnaire 403 issued by evaluation agency E, and both an "Accept" button 601a for receiving an instruction to accept the answer for each question and a "Modify" button 601b for receiving an instruction to modify the answer are shown. When processor system 100 receives an input to the "Modify" button 601b, it displays a modification screen 602. On modification screen 602, the question and the answer to be modified are displayed, and a modification field 602a for receiving an input of the modified content of the answer is displayed. The modified content received by modification field 602a becomes the answer to the question in questionnaire 403.

[0087] The above is a configuration example of the questionnaire answering system according to the embodiment of the present invention. According to questionnaire answering system 10, it is possible to reduce the burden of answering similar questions.

[0088] It should be noted that the present invention is not limited to the embodiments described above, and various modifications are included. For example, the embodiments described above are described in detail to make the present invention easier to understand, and are not necessarily limited to those having all the configurations described. It is possible to replace some of the configurations of the embodiments with other configurations, and it is also possible to add configurations from other embodiments to the configurations of the embodiments. Furthermore, it is possible to delete some of the configurations of the embodiments.

[0089] For example, in the embodiment described above, the optimization unit 124 receives the answers and stores them as answer history, and the answer support unit 122 identifies similar questions from the answer history and transcribes the answers, but the embodiment is not limited to this. Such embodiments will be explained with reference to Figures 11 to 14.

[0090] Figure 11 shows another example of the configuration of the question answering system. The question answering system 10' in Figure 11 is basically the same as the question answering system 10 of the embodiment described above, but there are some differences. The differences will be explained below. In summary, the processor system 100' reduces the burden of answering similar questions by generating suggested answers to questions according to the supplier using artificial intelligence such as a neural network.

[0091] The processing unit 120 includes a learning model unit 125. The learning model unit 125 is implemented using artificial intelligence such as a neural network and operates as a text generator that uses question materials and their answer history as training data. Note that the learning model unit 125 is not limited to artificial intelligence; for example, it may use an answer collection that aggregates multiple question materials as answer master information.

[0092] The learning model unit 125 performs additional learning and reconstructs the model using the second questionnaire item to be complemented, with respect to whether there are any corrections to the complemented response information and the information received as the response information after correction. In addition, when the learning model unit 125 receives relevant document information in addition to the response information, it learns that the response to the question is associated with the relevant document information.

[0093] The response support unit 122 uses the text generator of the learning model unit 125 to complement the response and generate a response answer. In addition, when there is relevant document information related to the response information, the response support unit 122 associates the relevant document information in the process of complementing the response information for the questions in the second questionnaire.

[0094] FIG. 12 is a diagram showing an example of a flowchart of the response support process (using a learned model). Since the response support process (using a learned model) is basically the same as the response support process (outsourcing the survey), the explanation will focus on the differences.

[0095] In step S102´, the response support unit 122 performs a response complementing process using the learning model for each supplier (step S102´). Specifically, the response support unit 122 uses the text generator of the learning model unit 125 to complement the response and generate a response answer.

[0096] Then, the learning model unit 125 incorporates the response sent to the evaluation institution or the buyer into the teacher data and performs additional learning of the model (step S107).

[0097] The above is an example of a flowchart of the response support process (using a learned model). According to the response support process (using a learned model), the burden of answering similar questions can be reduced.

[0098] FIG. 13 is a diagram showing an example of a flowchart of the response complementing process (using a learned model). Since the response complementing process (using a learned model) is basically the same as the response complementing process, the explanation will focus on the differences.

[0099] In step S205', the answer support unit 122 supplements the proposed answer to the question with a trained model (step S205').

[0100] The above is an example flowchart for the response completion process (using a pre-trained model). The response completion process (using a pre-trained model) allows for the completion of response information for questions in the questionnaire using a pre-trained model.

[0101] Figure 14 shows an example of answer completion (using a pre-trained model). Answer completion example 700 is an example in which answers are completed by answer support processing (using a pre-trained model). In answer completion example 700, the processor system 100 constructs a pre-trained model 406 using training data 405, which includes the results of web crawling about suppliers, the supplier's answers from the previous year, and question information obtained from evaluation materials of other evaluation organizations for the same period. An example is shown in which the pre-trained model 406 is used to complete answers to the questions in question material 402 issued by evaluation organization D and question material 403 issued by evaluation organization E. In other words, according to the answer support processing (using a pre-trained model), suppliers can obtain completed answers to question material 402 issued by evaluation organization D and question material 403 issued by evaluation organization E using the pre-trained model.

[0102] The above is an example of a modified embodiment. Each of the above parts, configurations, functions, processing units, etc., may be implemented in hardware, in whole or in part, for example, by designing them as integrated circuits. Alternatively, each of the above parts, configurations, functions, etc., may be implemented in software by having a processor interpret and execute programs that realize each function. Information such as programs, tables, and files that realize each function can be stored in memory, a recording device such as a hard disk, or a recording medium such as an IC card, SD card, or DVD.

[0103] It should be noted that the control lines and information lines in the embodiments described above are those deemed necessary for explanation and do not necessarily represent all control lines and information lines in the actual product. In practice, it can be assumed that almost all components are interconnected. The present invention has now been described, focusing on its embodiments. [Explanation of Symbols]

[0104] 10: Question answering system, 50: Network, 100: Processor system, 110: Memory, 111: Question data storage area, 112: Synonym dictionary storage area, 113: Answer history storage area, 120: Processing unit, 121: Question data reception unit, 122: Answer support unit, 123: Answer reception unit, 124: Optimization unit, 130: Input / output interface, 140: Transmission interface, 300: Evaluation organization D computer, 310: Evaluation organization E computer, 400: Supplier A computer, 410: Supplier B computer, 420: Supplier C computer, 800: Buyer F computer, 810: Buyer G computer, 1221: Classifier, 1222: Translator.

Claims

1. A processor system having one or more memory and one or more processors, The aforementioned processor, Using a questionnaire containing multiple predetermined questions stored in a predetermined memory unit, and answer information for those questions, when a question for a second questionnaire different from the first questionnaire is received, a supplementary process is performed to supplement the answer information for the second questionnaire by identifying and transcribing the answer information for questions similar to the questions in the second questionnaire from the answer information for the questions in the first questionnaire. In the aforementioned supplementary processing, it is determined whether or not there is document information associated with the response information of the questionnaire as evidence of attached materials, and if so, the associated document information is associated with the response information to the questions of the second questionnaire as attached materials. A processor system characterized by the following features.

2. A processor system according to claim 1, The aforementioned questionnaire is a questionnaire used to gather company information. A processor system characterized by the following features.

3. A processor system according to claim 1, The aforementioned processor, Questions in the aforementioned questionnaire that meet certain conditions will be excluded from the supplementation of response information. The aforementioned specified conditions are that the response period for the first questionnaire and the second questionnaire are not the same, the response information is numerical, and the response information requires related documentary information as evidence. A processor system characterized by the following.

Images