Safety confirmation system

Abstract

To provide a system that ensures efficiency and safety of questions and answers using a chatbot.SOLUTION: A safety confirmation system 1 is configured to cause a chatbot to create an answer text to a question from a questioner 102, and give an answer in chat. The system includes: an answer text generation unit 157 for creating an answer text to the question from the questioner 102; and an implication determination unit 160 which determines whether the answer text generated by the answer text generation unit 157 includes safety issues or not.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a safety confirmation system that provides workers performing work such as facility maintenance and installation work with knowledge to enable them to work safely. [Background technology]

[0002] Infrastructure equipment such as generators, elevators, mining and construction equipment are essential to society. These facilities are becoming more and more complex as improvements are made to meet the diverse needs of society. This has resulted in a need for a large amount of specialized knowledge about these facilities among the workers who maintain them and the salespeople who sell them.

[0003] For this reason, maintenance companies and sales companies have been providing specialized training on equipment to new employees, but due to a shortage of personnel to provide training, there are an increasing number of company departments where specialized training is insufficient. Therefore, there is a demand for a system that can quickly respond to questions that arise in daily work. Systems that answer such questions are called chatbot systems, but although they use AI (Artificial Intelligence), they do not always provide appropriate responses to questions.

[0004] To solve this problem, for example, the technology described in Patent Document 1 has been proposed. Patent Document 1 discloses not only a simple automatic answer, but also a method of automatically requesting a human to answer if it is determined that the chatbot system cannot answer, and further a method of assigning an answerer to seamlessly take over. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2022-189237 A Summary of the Invention [Problem to be solved by the invention]

[0006] However, for questions and answers about work that requires consideration of safety, such as construction and maintenance, even if a chatbot can answer, there is a possibility that the answers may pose safety risks, such as physical danger. On the other hand, there are a large number of questions and answers about administrative tasks that pose almost no safety risks, so it is not realistic to have humans check all of them.

[0007] An object of the present invention is to provide a system that can achieve both efficiency and safety in question-and-answer sessions using chatbots. [Means for solving the problem]

[0008] In order to achieve the above-mentioned object, the present invention provides a safety confirmation system that uses a chatbot to create a response to a user's inquiry and responds via chat, the system being characterized by comprising an answer generation unit that creates a response to the user's inquiry, and an implication determination unit that determines whether the answer generated by the answer generation unit poses any safety issues. Effect of the Invention

[0009] According to the present invention, a system can be provided that can achieve both efficiency and safety in question and answer sessions using chatbots. [Brief description of the drawings]

[0010] [Figure 1] 1 is an overall configuration diagram of a safety confirmation system 1 according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram showing an example of an internal data structure of a work procedure database 165 according to the embodiment of the present invention. [Diagram 3] FIG. 2 is a diagram showing an example of an internal data structure of an accident example database 170 according to an embodiment of the present invention. [Figure 4] 1 is a flowchart showing an overall process of a safety confirmation system 1 according to an embodiment of the present invention. [Diagram 5] 13 is a flowchart showing the process of a subroutine SUB01. [Figure 6] 13 is a flowchart showing the process content of a sub-subroutine SUB01-01. [Figure 7] 13 is a flowchart showing the processing contents of a sub-subroutine SUB01-02. [Figure 8] 13 is a flowchart showing the process of a subroutine SUB02. [Figure 9] FIG. 2 is a diagram showing an example of a question displayed on the display unit 105 of the terminal device 100. [Figure 10] FIG. 13 is a diagram showing an example of the display content displayed on display unit 135 of terminal device 130. [Figure 11] FIG. 10 is a diagram showing an example of the display content after input in an answer input field 1040. [Figure 12] FIG. 11 is a diagram showing an example of a reply text displayed on display unit 105 of terminal device 100. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In principle, the same elements are given the same reference numerals in all drawings. Furthermore, explanations of parts having the same functions will be omitted. Note that the configurations described below are merely examples, and it is not intended that the embodiment of the present invention is limited to the specific embodiments described below.

[0012] In the following embodiment, it is assumed that a questioner 102, such as a worker who is about to install equipment, uses a chatbot system to which the present invention is applied. This questioner 102 is about to install equipment X, but is having trouble with a lack of space at the installation site, and is about to ask about the matter through chat. The chat inquiry from the questioner 102 is received by the center-side system 150, which is a chatbot, and an answer is automatically generated and answered to the questioner 102 through chat. If the center-side system 150 detects a safety concern in the automatically generated answer, it issues a confirmation request to the answerer 132.

[0013] 〔system configuration diagram〕 Fig. 1 is an overall configuration diagram of a safety confirmation system 1 according to an embodiment of the present invention. In Fig. 1, the safety confirmation system 1 of the embodiment is composed of a terminal device 100 (questioner terminal device) carried by a questioner 102 who is a worker, a terminal device 130 (responder terminal device) carried by a respondent 132, and a center-side system 150 that connects the terminal device 100 and the terminal device 130 via communication.

[0014] The terminal device 100 and the terminal device 130 are preferably lightweight tablet terminals or smartphones that are easy for the questioner 102 and the answerer 132 to carry around. The terminal device 100 of the questioner 102 is equipped with a display unit 105 such as a liquid crystal display and an input unit 120 configured with a touch panel or the like, and communicates with a communication unit 180 of the center-side system via the communication unit 120. The terminal device 130 of the answerer is similar to the terminal device 100 of the questioner 102, and is equipped with a display unit 135 such as a liquid crystal display and an input unit 140 configured with a touch panel or the like, and communicates with the communication unit 180 of the center-side system via the communication unit 145.

[0015] In this embodiment, since it is assumed that the questioner 102 and the answerer 132 carry the terminal device 100 and the terminal device 130, the terminal device 100, the terminal device 130, and the center side system 150 are configured separately, but they may also be configured as an integrated system.

[0016] The center side system 150 includes an answer sentence generation unit 157 which generates an answer sentence to a question from the questioner 102 (an inquiry from a user), an implication determination unit 160 which determines whether the answer sentence generated by the answer sentence generation unit 157 has any safety problems, a work procedure database 165 which stores work procedures that are the basis for determining whether there is a problem, and an accident case database 170 which stores accident cases. A safety problem in this embodiment means, for example, a danger to the user's body.

[0017] The center-side system 150 also includes an answerer selection unit 175 that selects which answerer to send a confirmation request to when the implication determination unit 160 determines that there is a safety problem, and a confirmation request generation unit 177 that creates the text of the confirmation request for the answerer 132 selected by the answerer selection unit 175. Furthermore, the center-side system 150 includes a primary storage unit 155 that temporarily stores the information output by these functions so that it can be used as input for the next process. The center-side system 150 constitutes a chatbot for inquiries.

[0018] [Data structure of work procedure database 165] Next, the data structure of the work procedure database 165 will be described with reference to Fig. 2. Fig. 2 is a diagram showing an example of the internal data structure of the work procedure database 165 according to the embodiment of the present invention.

[0019] In the work procedure database 165, a plurality of existing work procedure manuals used in infrastructure facilities and the like are digitized and stored. The work procedure database 165 is divided into a plurality of items, and stores, for example, a document name 212 indicating the name of the procedure manual, a document URL 215 which is an address for downloading the document name, a page number 220 of each document, a paragraph ID 225 indicating the paragraph present on that page, a sentence 230 which is the content of the paragraph, and a writer contact address 235 for searching for a respondent. The work procedure database 165 breaks down various work procedure manuals into paragraph units and stores each paragraph in units of one line. In this embodiment, one procedure is written in one paragraph, and the implication determination unit 160 can determine whether there is any omission in the answer sentence by comparing it with the sentence 230. In this embodiment, the writer contact address 235 is an email address, but it can also be substituted with a user ID of a chat system or the like.

[0020] [Data structure of accident case database 170] Next, the data structure of the accident example database 170 will be described with reference to Fig. 3. Fig. 3 is a diagram showing an example of the internal data structure of the accident example database 170 according to the embodiment of the present invention.

[0021] Accident case database 170 stores electronic reports of accidents that have occurred in infrastructure facilities, etc. However, when a new accident occurs, it is advisable to add the accident case to accident case database 170 to update it.

[0022] The accident case database 170 is divided into multiple items, such as a case URL 310 where the report can be downloaded, page numbers 315 on which the paragraphs are written, paragraph IDs 320 indicating the paragraphs on that page, text 325 which is the content of the paragraph, and writer contact information 330 for searching for respondents.

[0023] In the process described below, it is determined that there is a safety concern when the reply text of the chatbot system is related to text 325. The accident case database 170 also has writer contact information 330, and if it is determined that there is a related accident case, a confirmation request is sent to the writer contact information.

[0024] [Answer sentence generation unit 157] The answer generation unit 157 is a function of the chatbot system that generates an answer to a question from the questioner 102.

[0025] This function can be used in a variety of ways, such as by creating a large number of pairs of expected questions and answers in advance, and when a question is given, searching for similar expected questions to find the answer, or by generating sentences from past history learned by a generation AI. Regardless of which method is used in this embodiment, it will be treated as existing technology.

[0026] [Implication determination unit 160] The implication determination unit 160 is a function that uses an implication determination technique, which is one of the natural language processing techniques, for determining whether there is a safety concern in the answer sentence generated by the answer sentence generation unit 157. The implication determination technique is a technique that interprets the contents of, for example, text data A, "I have a pen and a notebook," and text data B, "I have a notebook," and automatically performs the following three types of determinations. (1) The content of A is contained in B. (2) The content of B is contained in A. (3) A and B are unrelated.

[0027] In the above example, it is determined that "(2) the content of B is included in A."

[0028] In this embodiment, the implication determination technology is treated as an existing technology. For details, see, for example, the references listed below. https: / / en.wikipedia.org / wiki / Textual_entailment Using the above technology, the implication determination unit 160 compares the answer sentence output by the answer sentence generation unit 157 with sentences 230 in the work procedure database 165 to perform implication determination. This determines whether the answer sentence has any omissions, or whether the answer sentence is related to an accident by comparing it with sentences 325 in the accident case database 170.

[0029] [Main processing flow] FIG. 4 is a flowchart showing the overall process of the safety confirmation system 1 according to the embodiment of the present invention.

[0030] The process flow from start to finish (steps S400 to S445) will be described with reference to Fig. 4. Subroutines SUB01 and SUB02 are called in steps S415 and S430 along the way, and the detailed processing of these subroutines will be described later.

[0031] First, the questioner 102, who is a worker, inputs a question from the input unit 110 of the terminal device 100 and transmits it to the center-side system 150 (step S400). This is done by displaying a screen as shown in FIG. 9 on the display unit 105, and prompting the user to enter the question. FIG. 9 is a diagram showing an example of the question displayed on the display unit 105 of the terminal device 100. When the question is input from the input unit 110, a question 910 is displayed on the display unit 105. In FIG. 9, the question is displayed in a dialogue format. The questioner 102 (worker) checks the displayed question 910 and then presses the send button 920. When the send button 920 is pressed, the terminal device 100 transmits the question to the center-side system 150 via the communication unit 120. The communication unit 180 of the center-side system 150 receives the question.

[0032] When the question is received, the answer generator 157 of the center system 150 generates an answer to the received question (step S405). The answer may be generated from the expected question and answer table as described above, or may be generated by a generation AI.

[0033] When the answer sentence is created by the answer sentence generation unit 157, the implication determination unit 160 determines whether or not there is a safety concern in the created answer sentence (step S415). Details of this process will be described later in the explanation of the subroutine SUB01. In the process of the subroutine SUB01, for example, one of the following four patterns is returned as a return value. (1) No safety concerns (2) No implication of the procedure (3) There are omissions (4) There have been similar accidents In addition, in the case of (3), the paragraph ID of the work procedure manual database 170 is attached as the missing work procedure, and in the case of (4), the paragraph ID of the accident case database 170 is attached to the return value as a similar accident case.

[0034] When the implication determination unit 160 determines that "(1) there is no safety concern" (Yes in step S420), the implication determination unit 160 transmits the answer sentence generated by the answer sentence generation unit 157 to the terminal device 100 of the questioner 102 (worker) and presents it (step S425). A display example of the answer sentence is shown in Fig. 12. Fig. 12 is a diagram showing an example of the answer sentence displayed on the display unit 105 of the terminal device 100.

[0035] In step S425, if the answer does not pose a safety concern, the answer 1250 is displayed on the display unit 105 of the terminal device 100 together with the question content 1210. After the answer 1250 is displayed on the display unit 105, this process ends.

[0036] If the implication determination unit 160 determines that there is any one of "(2) no implication relationship with the procedure, (3) there is an omission, or (4) there is a similar accident case" (No in step S420), the implication determination unit 160 instructs the answerer selection unit 175 to select an answerer.

[0037] Upon receiving an instruction from the implication determination unit 160, the answerer selection unit 175 searches for the contact information of the answerer 132 making the safety confirmation request (step S430). This is obtained by searching the writer contact information 235 of the work procedure database 165 or the writer contact information 330 of the accident case database 170, the details of which will be described later in the section on subroutine SUB02. After selecting the answerer 132, the answerer selection unit 175 transmits the selection result to the confirmation request generation unit 177.

[0038] The confirmation request generation unit 177 transmits (a) the question text, (b) the generated answer text, and (c) the accident case on which the safety concern is based or the paragraph ID of the work procedure manual to the terminal device 130 of the respondent 132 selected by the answerer selection unit 175 (step S435). This can be done by simply sending all the output results up to that point together. Display examples of the terminal device 130 of the respondent 132 are shown in Figs. 10 and 11. Fig. 10 is a diagram showing an example of the display content displayed on the display unit 135 of the terminal device 130. Fig. 11 is a diagram showing an example of the display content after input into the answer input field 1040.

[0039] A question 1010, a proposed answer 1020, and a concern 1030 are displayed on the display unit 135 of the terminal device 130, as well as an answer input field 1040.

[0040] The answerer 132 reads and confirms the contents displayed on the display unit 135 (question 1010, draft answer 1020, concern 1030), and inputs the answer in the answer input field 1040. The answerer 132 reads the contents displayed on the display unit 135, determines whether there is a safety problem, and if there is a problem, enters that fact in the answer input field 1040. The screen after entry is as shown in FIG. 11. Then, after entering the answer in the answer input field 1040, the answerer 132 presses the send button 1050. The terminal device 130 transmits the answer entered in the answer input field 1040 to the center side system 150 via the communication unit 145 (step S440).

[0041] The center-side system 150 transmits the answer text of the answerer 132 transmitted from the terminal device 130 to the terminal device 100 of the questioner 102 (worker) via the communication unit 180. The terminal device 100 receives the answer text at the communication unit 120 and presents (displays) the answer text on the display unit 105 (step S445). The concern 1030 and the answer text entered in the answer input field 1040 are displayed on the display unit 105 of the terminal device 100. After displaying, the center-side system 150 ends this process.

[0042] [Processing flow of subroutine SUB01] Next, the process flow of the subroutine SUB01 called in step S415 will be described with reference to Fig. 5. Fig. 5 is a flowchart showing the process contents of the subroutine SUB01.

[0043] In the subroutine SUB01 of this embodiment, the accident case database 170 and the answer sentence 160, and the work procedure database 165 and the answer sentence 160 are compared to determine implication. The implication determination between the accident case database 170 and the answer sentence 160 is performed in the sub-subroutine SUB01-01, and the implication determination between the work procedure database 165 and the answer sentence 160 is performed in the sub-subroutine SUB01-02.

[0044] In FIG. 5, when an answer sentence is generated by the answer sentence generation unit 157, the implication determination unit 160 calls the sub-subroutine SUB01-01 and searches the accident case database 170 to determine whether there are any safety concerns in the generated answer sentence, and performs an implication determination (step S500).

[0045] The implication determination unit 160 searches the accident case database 170 and performs implication determination to determine whether or not there is a sentence of an accident case similar to the answer sentence generated by the answer sentence generation unit 157 (step S510). The detailed processing of the sub-subroutine SUB01-01 will be described later, but as a return value of the determination result of the implication determination unit 160, one of the following is returned. (a) There is a similar accident case (however, the paragraph ID of the accident case is added to the return value) (b) No accident case In step S510, if the return value is "(a) similar accident cases exist" (Yes in step S510), the implication determination unit 160 returns the paragraph ID of the accident case together with the information that similar accident cases exist as a return value (step S560).

[0046] In step S510, if the return value is "(b) not related to an accident case" (No in step S510), the implication determination unit 160 calls the sub-subroutine SUB01-02, searches the work procedure database 165, and performs an implication determination to check whether there are any omissions in the answer sentence generated by the answer sentence generation unit 157 (step S520). The detailed processing of the sub-subroutine SUB01-02 will be described later, but one of the following is returned as the return value of the determination result of the implication determination unit 160 (step S530). (a) No omissions (b) There are omissions (c) No implication In step S530, if the return value is “(a) no omissions” or “(c) no implication relationship” (Yes in step S530), implication determination unit 160 returns a value indicating that there are no safety concerns and terminates this subroutine (step S540).

[0047] In step S530, if the return value is “(b) omission or omission exists” (No in step S530), implication determination unit 160 returns a message indicating that omission or omission exists, and the paragraph ID of the work procedure manual in which the omission or omission has been confirmed, and ends this subroutine (step S550).

[0048] [Processing flow of sub-subroutine SUB01-01] Next, the processing flow of the sub-subroutine SUB01-01 will be described with reference to Fig. 6. Fig. 6 is a flowchart showing the processing contents of the sub-subroutine SUB01-01.

[0049] The sub-subroutine SUB01-01 is called in step S500, and executes a process of determining whether or not there is an implication relationship between the reply sentence generated by the reply sentence generating unit 157 and the sentence 325 in the accident example database 170, and returning the result.

[0050] First, the implication determination unit 160 loads the top line of a paragraph of text P from the accident case database 170 into the primary storage unit 155 (step S600).

[0051] Next, the implication determination unit 160 executes a determination of an implication relationship between the sentence P loaded in the primary storage unit 155 and the answer sentence (step S605). As a return value of the determination result of the implication determination unit 160, one of the following is returned. (a) Sentence P is included in the answer (b) The answer is contained in sentence P (c) No implication If the implication determination result is "(a) sentence P is included in the answer sentence" or "(b) the answer sentence is included in sentence P," the implication determination unit 160 determines that there is a relationship with the accident case (Yes in step S610). Then, the implication determination unit 160 returns a set of information that there is a similar accident case and the paragraph ID of the accident case to the main routine, and ends this sub-subroutine (step S627).

[0052] Furthermore, if the implication determination result is "(c) no implication relationship," the implication determination unit 160 determines whether the implication determination of all sentences in the accident case database 170 has been completed (step S630), and if not completed, returns to step S600 (No in step S630) and performs implication determination of the paragraphs from the second line onwards in the accident case database 170.

[0053] In step S630, if all implication determinations have been completed (Yes in step S630), the implication determination unit 160 determines that the answer sentence is not related to the accident case, returns a value indicating that the answer sentence is not related to the accident case to the main routine, and terminates this sub-subroutine (step S635).

[0054] [Processing flow of sub-subroutine SUB01-02] Next, the processing flow of the sub-subroutine SUB01-02 will be described with reference to Fig. 7. Fig. 7 is a flowchart showing the processing contents of the sub-subroutine SUB01-02.

[0055] The sub-subroutine SUB01-02 is called in step S520, and executes a process of searching the work procedure database 165, determining whether there is an implication relationship between the answer sentence generated by the answer sentence generation unit 157 and the sentence 230 in the work procedure database 165, and returning the result.

[0056] First, implication determination unit 160 loads text P of one paragraph on the top line from work procedure database 165 into primary storage unit 155 (step S700).

[0057] Next, the implication determining unit 160 executes a determination of an implication relationship between the sentence P loaded in the primary storage unit 155 and the answer sentence (step S705). As a return value of the determination result of the implication determining unit 160, one of the following is returned. (a) Sentence P is included in the answer (b) The answer is contained in sentence P (c) No implication If the implication determination result is "(c) no implication relationship" (No in step S710), the implication determination unit 160 determines whether the implication determination of all sentences in the work procedure database 165 has been completed (step S730), and if not, returns to step S700 (No in step S730) and performs implication determination on the paragraphs from the second line onwards in the work procedure database 165.

[0058] In step S730, if all implication determinations have been completed (Yes in step S730), implication determination unit 160 determines that the answer sentence is not related to the work procedure, returns a value indicating that there is no implication relationship with the work procedure to the main routine, and terminates this sub-subroutine (step S735).

[0059] If the implication determination result is "(a) sentence P is included in the answer sentence" or "(b) the answer sentence is included in sentence P" (Yes in step S710), the implication determination unit 160 executes a determination as to whether sentence P is included in the answer sentence (step S720). In the case of a work procedure, it is necessary to determine not only whether there is an implication relationship but also whether there is an omission, so the implication determination unit 160 executes the process of step S720.

[0060] In step S720, if the determination result is "(a) sentence P is included in the answer sentence", implication determination unit 160 determines that "the answer sentence has no omissions" (step S725) and ends this sub-subroutine. If the determination result is "(a) sentence P is included in the answer sentence", this means that sentence P, which indicates the entire work procedure, is included in its entirety in the answer sentence. In other words, there is no omission.

[0061] On the other hand, in step S720, if the determination result is "(b) the answer sentence is included in sentence P", the implication determination unit 160 determines that "the answer sentence has an omission" (step S727), and returns a set of information that there is an omission and the paragraph ID of sentence P that has the omission to the main routine (step S727). After that, this sub-subroutine is terminated.

[0062] [Processing flow of subroutine SUB02] Next, the process flow of the subroutine SUB02 called in step S430 will be described with reference to Fig. 8. Fig. 8 is a flowchart showing the process contents of the subroutine SUB02.

[0063] Subroutine SUB02 is a process for searching for a respondent who will send a safety confirmation request from the accident case database 170 and the work procedure database 165. This process passes the return value of subroutine SUB01 as an argument to be passed to the routine. The argument is one of the following: (a) Set information that there is a similar accident case and the paragraph ID of the accident case database 170 (b) Set the information that there is an omission and the paragraph ID of the work procedure database 165 where the omission was determined. The answerer selection unit 175 judges whether the argument corresponds to either of the above (a) or (b) from the judgment result of the subroutine SUB01 (step S800).

[0064] If the result of the determination in subroutine SUB01 does not correspond to "similar accidents exist" (No in step S800), the answerer selection unit 175 acquires the paragraph ID 225 of the writer contact information 235 in the work procedure database 165 as a search condition. Then, the paragraph ID included in the above argument (b) of this subroutine is used as the paragraph ID 225 that serves as the search key (step S810). After that, the answerer selection unit 175 returns the writer contact information to the confirmation request generation unit 177 and ends the process.

[0065] If the result of the determination in subroutine SUB01 corresponds to "similar accidents have occurred" (Yes in step S800), the answerer selection unit 175 acquires the paragraph ID 320 of the writer contact information 330 in the accident case database 170 as a search condition. Then, the paragraph ID included in the above argument (a) of this subroutine is used as the paragraph ID 320 that serves as the search key (step S820). After that, the answerer selection unit 175 returns the writer contact information to the confirmation request generation unit 177 and ends the process.

[0066] As described above, according to this embodiment, answers that pose safety concerns during a question and answer session by a chatbot are revised, thereby achieving both efficiency and safety in the question and answer session by the chatbot.

[0067] In the above embodiment, an example of an interactive format in which the questioner asks a question to the answerer and the answerer answers the question has been described, but the present invention can also be applied to a request format such as "Please create XX." The present invention can also be applied not only to chatbots but also to checks when creating new manuals or new procedure manuals.

[0068] The present invention is not limited to the above-mentioned embodiment, and various modifications are included. For example, the above-mentioned embodiment has been described in detail to easily explain the present invention, and is not necessarily limited to those including all of the configurations described. Also, it is possible to replace a part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment. Also, it is possible to add, delete, or replace a part of the configuration of each embodiment with another configuration. [Explanation of symbols]

[0069] 1...Safety confirmation system, 100...Terminal device, 102...Questioner, 105...Display unit, 110...Input unit, 120...Communication unit, 130...Terminal device, 132...Answerer, 135...Display unit, 140...Input unit, 145...Communication unit, 150...Center side system, 155...Primary memory unit, 157...Answer sentence generation unit, 160...Implication determination unit, 165...Work procedure database, 170...Accident case database, 175...Answerer selection unit, 177...Confirmation Request generation unit, 180... communication unit, 212... document name, 215... document URL, 220... page number, 230... text, 235... author contact information, 310... case example URL, 315... page number, 320... paragraph ID, 325... text, 330... author contact information, 910... question text, 920... send button, 1010... question text, 1020... answer draft, 1040... answer input field, 1050... send button, 1210... question content, 1250... answer text

Claims

1. In a safety confirmation system that uses a chatbot to create responses to user inquiries and respond via chat, A safety confirmation system comprising an answer sentence generation unit that creates an answer sentence to a query from the user, and an implication determination unit that determines whether the answer sentence generated by the answer sentence generation unit poses any safety issues.

2. The safety confirmation system according to claim 1, Equipped with an accident case database to store accident cases, The implication determination unit searches the accident case database, and if the answer sentence generated by the answer sentence generation unit contains a sentence similar to a sentence stored in the accident case database, determines that the answer sentence contains a safety issue.

3. The safety confirmation system according to claim 2, Equipped with a work procedure database for storing work procedures, A safety confirmation system characterized in that the implication determination unit searches the work procedure database and determines whether there is an implication relationship between the answer sentence generated by the answer sentence generation unit and the work procedure sentence.

4. The safety confirmation system according to claim 3, A safety confirmation system characterized in that, when it is determined that there is an implication relationship between the answer sentence generated by the answer sentence generation unit and the work procedure sentence, the implication determination unit determines whether the content of the work procedure sentence is included in the answer sentence, and if the content of the work procedure sentence is included in the answer sentence, it determines that there is no omission in the answer sentence, and if the content of the work procedure sentence is not included in the answer sentence, it determines that there is an omission in the answer sentence.

5. The safety confirmation system according to claim 4, A safety confirmation system comprising an answerer selection unit which selects which answerer to send a confirmation request to when it is determined that the answer sentence contains a safety problem or that the answer sentence contains an omission.

6. The safety confirmation system according to claim 5, A safety confirmation system comprising a confirmation request generation unit that generates a confirmation request for the answerer selected by the answerer selection unit.

7. The safety confirmation system according to any one of claims 1 to 6, The safety confirmation system is characterized in that it is an interactive system in which the user asks questions and the chatbot answers them.

Images