Computer program, information processing method, information processing device, and information processing system

Abstract

Provided are a computer program, an information processing method, an information processing device, and an information processing system that can be expected to assist in the collection of information pertaining to an event in a device.　A computer program according to the present embodiment causes a computer to execute a process for: acquiring information pertaining to the history of a handling of an event that has occurred in a device; using a language model to generate a natural-language question sentence relating to the acquired information; outputting the generated question sentence; acquiring an answer to the question sentence; and storing information pertaining to the acquired answer in a first database.

Description

Computer program, information processing method, information processing device, and information processing system

[0001] The present disclosure relates to a computer program, an information processing method, an information processing device, and an information processing system.

[0002] Patent Literature 1 proposes a bot system that uses avatars to converse with users in natural language and coordinate a team of autonomous helper bots. The helper bots can perform various tasks, such as ranking content based on a user's profile and customizing search functions to suit the user's preferences, utilizing a digital library architecture to provide content.

[0003] US Patent Application Publication No. 2016 / 0044380

[0004] The present disclosure provides a computer program, an information processing method, an information processing device, and an information processing system that are expected to assist in collecting information about device events.

[0005] A computer program according to one embodiment causes a computer to acquire information relating to the process of responding to an event that occurred in a device, generate a natural language question relating to the acquired information using a language model, output the generated question, acquire an answer to the question, and store information relating to the acquired answer in a first database.

[0006] The present disclosure is expected to assist in collecting information about device events.

[0007] FIG. 1 is a schematic diagram for explaining an overview of an information processing system according to the present embodiment. FIG. 2 is a block diagram showing an example of a configuration of an information processing device according to the present embodiment. FIG. 3 is a schematic diagram showing an example of a trouble response DB. FIG. 4 is a block diagram showing an example of a configuration of a trouble history recording device according to the present embodiment. FIG. 5 is a schematic diagram showing an example of trouble history information stored in a trouble history DB. FIG. 6 is a flowchart showing an example of a processing procedure performed by the information processing device according to the present embodiment. FIG. 7 is a schematic diagram showing an example of information input to a large-scale language model and information output by the large-scale language model. FIG. 8 is a schematic diagram showing an example of information input to a large-scale language model and information output by the large-scale language model. FIG. 9 is a schematic diagram showing an example of information input to a large-scale language model and information output by the large-scale language model. FIG. 10 is a schematic diagram showing an example of information input to a large-scale language model and information output by the large-scale language model. FIG. 11 is a schematic diagram showing an example of information display by a terminal device.

[0008] Specific examples of information processing systems according to embodiments of the present disclosure will be described below with reference to the drawings. Note that the present disclosure is not limited to these examples, but is defined by the claims, and is intended to include all modifications within the meaning and scope of the claims.

[0009] <System Overview> The information processing system according to this embodiment stores information about the history of events occurring in various devices, such as substrate processing apparatuses or transport apparatuses, in a database. Events occurring in devices can include, for example, device abnormalities, breakdowns, maintenance, setting changes, part replacements, or function additions. Information about the history of events occurring in devices can include, for example, the occurrence of an event, such as an abnormality or breakdown, the troubleshooting measures taken by engineers in response to the event, and the causes identified for the event. Information about the history of events occurring in devices can also include the opinions of one or more engineers regarding the event, discussions among multiple engineers, or questions and answers regarding the event. The information processing system according to this embodiment appropriately reorganizes the various pieces of information stored in the database and creates a new database by supplementing missing information, thereby supporting the utilization of past knowledge and insights of engineers. The following description will be given as an example of a system that handles problems, such as abnormalities or breakdowns, occurring in substrate processing apparatuses, as device events.

[0010] FIG. 1 is a schematic diagram illustrating an overview of an information processing system according to this embodiment. The information processing system according to this embodiment includes an information processing device 1, a trouble history recording device 3, and a terminal device 5. The terminal device 5 is used by a user who uses or manages an apparatus such as a substrate processing apparatus or a transport apparatus. Various general-purpose devices, such as a personal computer, a smartphone, or a tablet terminal device, may be used. When a trouble such as an abnormality or a breakdown occurs in a target apparatus, the user who handles the trouble inputs information about the process of dealing with the trouble to the terminal device 5, such as the nature of the trouble and the details of the measures taken to resolve the trouble. The information about the process of dealing with the trouble input by the user is transmitted from the terminal device 5 to the trouble history recording device 3. For example, the user inputs information each time the user handles a trouble with the apparatus, and the terminal device 5 transmits the information to the trouble history recording device 3 each time it receives information input from the user. A user may input information multiple times for one trouble, and the input and transmission of information may be repeated until the trouble is resolved. Furthermore, multiple users may input information for one trouble.

[0011] The trouble history recording device 3 is a device that records information about the history of responses to troubles that have occurred in various devices, such as substrate processing devices or transport devices, in a trouble history DB (database) 102. The trouble history recording device 3 acquires information about troubles transmitted from the terminal device 5 and stores the acquired information in the trouble history DB 102 together with information such as the date and time the trouble occurred, the date and time the response to the trouble was made, identification information of the device in which the trouble occurred, identification information that identifies the trouble, and identification information of the user who responded to the trouble. The trouble history recording device 3 can acquire information about one trouble multiple times and store it in the trouble history DB 102, and stores the multiple pieces of information about one trouble in association with each other so that the multiple pieces of information about the trouble can be traced in chronological order as the history of responses to the trouble. In other words, the information about the history of responses to one trouble can include multiple pieces of information from the occurrence of the trouble to its resolution.

[0012] The information processing device 1 performs a process of monitoring the recording of information on the history of troubleshooting (hereinafter referred to as "trouble history information") by the trouble history recording device 3. The information processing device 1 periodically checks the trouble history information recorded in the trouble history DB 102 of the trouble history recording device 3, searching for cases in which the history of the troubleshooting from the occurrence of the trouble to its resolution has not been recorded and a predetermined period of time (e.g., one month or more) has passed since the last information entry on the history of the case. If such incomplete trouble history information is stored in the trouble history DB 102, the information processing device 1 sends a message to one or more users (or terminal devices 5 used by the users) involved in the trouble, asking about the history of the subsequent response. The information processing device 1 receives a response message from the user in response to the question message and completes the incomplete information on the history of the trouble based on the received response message. Note that the message exchange between the information processing device 1 and the users may be repeated multiple times until a sufficient amount of trouble history information is collected.

[0013] The information processing device 1 converts the trouble-related information stored in the trouble-related DB 102 of the trouble-related recording device 3 and information obtained through message exchange with the user into information in a predetermined format, and stores the converted trouble-related information in the predetermined format in the trouble response DB 101. Furthermore, when the trouble-related information stored in the trouble-related DB 102 of the trouble-related recording device 3 includes all the information up to the completion of response to the trouble, the information processing device 1 acquires the information stored in the trouble-related DB 102 of the trouble-related recording device 3, converts it into a predetermined format, and stores it in the trouble response DB 101 without exchanging messages with the user.

[0014] The information relating to troubleshooting stored in the troubleshooting DB 101 is made public to a plurality of users. A user can search for the information stored in the troubleshooting DB 101 using various conditions and view desired information via the terminal device 5. Users who can search for and view the information stored in the troubleshooting DB 101 include not only users who have dealt with troubles in devices such as substrate processing apparatuses or transport devices, but also various users who seek knowledge of what kind of response should be made to troubles in these devices.

[0015] Furthermore, the information processing device 1 according to this embodiment utilizes a trained large language model (LLM) when exchanging messages with a user and when formatting information about a trouble into a predetermined format. The information processing device 1 inputs incomplete trouble history information stored in the trouble history DB 102 into the large language model and causes the large language model to generate a question in natural language inquiring about the situation related to the trouble. Based on the question generated by the large language model, the information processing device 1 transmits a question message to one or more user terminal devices 5 corresponding to the trouble and obtains a reply message from the user. The reply message from the user may be written in natural language, or may be information such as which option was selected, or may be in any other format. The information processing device 1 inputs the reply message from the user into the large language model, determines whether the necessary information has been collected, and if the necessary information has not been collected, creates an additional question and repeatedly exchanges messages with the user.

[0016] When the necessary information is collected, the information processing device 1 provides the trouble history information acquired from the trouble history DB 102 and the information acquired through message exchange with the user to the large-scale language model and issues a command to format this information into a predetermined format. In the information processing system according to this embodiment, the information stored in the trouble history DB 102 and the information acquired through message exchange with the user are information in natural language or information including information in natural language. The large-scale language model accepts information expressed in natural language as input, formats the information into a specified predetermined format, in this embodiment, a format suitable for storage in a database, and outputs the information formatted in the predetermined format. The information processing device 1 acquires the information output by the large-scale language model and stores the information in the predetermined format formatted by the large-scale language model in the trouble response DB 101.

[0017] <Device Configuration> Fig. 2 is a block diagram showing an example configuration of an information processing device 1 according to this embodiment. The information processing device 1 according to this embodiment can be realized by installing a predetermined application program or the like in a general-purpose information processing device such as a personal computer or a server computer. The information processing device 1 according to this embodiment is configured to include a processing unit 11, a storage unit 12, a communication unit 13, etc. Note that in this embodiment, the processing will be described as being performed by a single information processing device 1, but the processing of the information processing device 1 may be distributed among multiple devices.

[0018] The processing unit 11 is configured using an arithmetic processing device such as a CPU (Central Processing Unit), an MPU (Micro-Processing Unit), a GPU (Graphics Processing Unit) or a quantum processor, a ROM (Read Only Memory), a RAM (Random Access Memory), etc. The processing unit 11 reads and executes a program 12a stored in the storage unit 12 to perform various processes such as monitoring the information recorded by the trouble history recording device 3, exchanging messages with the user, and arranging the obtained information into a predetermined format and storing it in the trouble response DB 101.

[0019] The storage unit 12 is configured using a large-capacity storage device such as a hard disk or an SSD (Solid State Drive). The storage unit 12 stores various programs executed by the processing unit 11 and various data required for the processing of the processing unit 11. In this embodiment, the storage unit 12 stores a program 12a executed by the processing unit 11. The storage unit 12 also has a model information storage unit 12b that stores information related to a trained learning model used in the processing performed by the information processing device 1. The storage unit 12 also has a trouble response DB 101 that stores information related to responses to troubles that occur in devices such as the substrate processing device or the transport device.

[0020] In this embodiment, the program (computer program, program product) 12a is provided in a form recorded on a recording medium 99 such as a memory card or an optical disc, and the information processing device 1 reads the program 12a from the recording medium 99 and stores it in the storage unit 12. However, the program 12a may also be written to the storage unit 12, for example, during the manufacturing stage of the information processing device 1. Alternatively, the program 12a may be distributed by a remote server device or the like and acquired by the information processing device 1 via communication. For example, the program 12a may be read from the recording medium 99 by a writing device and written to the storage unit 12 of the information processing device 1. The program 12a may be provided in a form distributed via a network or in a form recorded on the recording medium 99.

[0021] The model information storage unit 12b stores information about a learning model that has been previously subjected to machine learning. The information about the learning model may include, for example, information indicating the configuration of the learning model and information such as the values ​​of internal parameters determined by machine learning. In this embodiment, the learning model storage unit 12b stores information about a large-scale language model for exchanging messages in natural language with a user and shaping information into a predetermined format. Note that in this embodiment, the information processing device 1 stores information about the learning model and performs processing using the learning model, but this is not limited to this. The information about the learning model may be stored in a device different from the information processing device 1, and this device may perform processing using the learning model, and the information processing device 1 may obtain the processing results from this device. Furthermore, the machine learning processing of the learning model may be performed by the information processing device 1 or by a device different from the information processing device 1.

[0022] The trouble response DB 101 is a database that stores various information related to responses taken by users to troubles in devices such as substrate processing apparatuses or transport devices. Fig. 3 is a schematic diagram showing an example of the trouble response DB 101. The trouble response DB 101 according to this embodiment is a database in which values ​​are associated with item names such as "trouble name," "date and time of occurrence," "customer name," "Fab No.", "device information," "recipe type," "(presumed) cause," "emergency treatment," "whether or not an experiment was conducted to investigate the cause, and if so, the details of the experiment," "permanent measures," "information transfer destination for the permanent measures," "related data," "related materials," "responder," and "recording date and time."

[0023] The "Trouble Name" field stores a value such as "ESC Trouble." The "Date and Time of Occurrence" field stores a value such as "September 15, 2022, 8:15." The "Customer Name" field stores a value such as "Company M." The "Fab No." field stores a value such as "X." The "Equipment Information" field stores a value such as "T Platform, CP Chamber." The "Recipe Type" field stores a value such as "ON Etching." The "(Probable) Cause" field stores a value such as "Uneven Wear Within the Surface." The "Temporary Solution" field stores a value such as "ESC Replacement." The "Whether Experiments to Investigate the Cause Have Been Conducted, and If So, What?," field stores a value such as "No Experiments Have Been Conducted." The "Permanent Solution" field stores a value such as "The Wear Mechanism Must Be Identified First." In the "Information transfer destination for permanent measures" field, for example, information such as "Device Development Department" is stored as a value. In the "Related data" and "Related materials" fields, although not shown in FIG. 3, information such as an address or link to a location where files of related data or materials are stored is stored. In the "Responder" field, for example, information such as "A" is stored as a value. In the "Recording date and time" field, for example, information such as "October 17, 2022, 0:00" is stored as a value. Note that these field names and values ​​are merely examples and are not limited to these.

[0024] In this embodiment, the information processing device 1 is configured to include the trouble-shooting DB 101, but this is not limiting. The trouble-shooting DB 101 may be included in a device different from the information processing device 1. In this case, the information processing device 1 communicates with the device that includes the trouble-shooting DB 101, and reads information from the trouble-shooting DB 101 and writes information to the trouble-shooting DB 101.

[0025] The communication unit 13 transmits and receives data to and from devices such as the trouble history recording device 3 and the terminal device 5 via a wired or wireless network N. In this embodiment, the information processing device 1 can acquire information stored in the trouble history DB 102 provided in the trouble history recording device 3 by communicating with the trouble history recording device 3 via the communication unit 13. Furthermore, the information processing device 1 can transmit question messages to users who use the terminal device 5 and receive answer messages from the users by communicating with the terminal device 5 via the communication unit 13. The communication unit 13 transmits data provided by the processing unit 11 to other devices, receives data from other devices, and provides the received data to the processing unit 11.

[0026] The storage unit 12 may be an external storage device connected to the information processing device 1. The information processing device 1 may also be a multi-computer including multiple computers, or may be a virtual machine virtually constructed by software. The information processing device 1 is not limited to the above configuration, and may also include, for example, a reading unit that reads information stored in a portable storage medium, an input unit that accepts operation input, or a display unit that displays images.

[0027] In the information processing device 1 according to this embodiment, the processing unit 11 reads and executes the program 12a stored in the storage unit 12, thereby realizing a trouble history monitoring unit 11a, a question generation unit 11b, a response acquisition unit 11c, an information shaping unit 11d, a DB processing unit 11e, and the like as software functional units in the processing unit 11. Note that in the figure, functional units related to processing for handling information about device troubles are shown as functional units of the processing unit 11, and functional units related to other processing are not shown.

[0028] The trouble history monitoring unit 11a communicates with the trouble history recording device 3 via the communication unit 13 at appropriate intervals, such as once every few hours or once every few days, and performs processing to monitor the status of storage (collection) of trouble history information in the trouble history DB 102 of the trouble history recording device 3. The trouble history monitoring unit 11a monitors whether the many pieces of trouble history information stored in the trouble history DB 102 include trouble history information for which a response to a problem that occurred in the device has not been completed and for which a predetermined period (e.g., one month) has passed since the last addition of the trouble history information that is added chronologically. If information that meets the above conditions is included, the trouble history monitoring unit 11a acquires the corresponding trouble history information from the trouble history recording device 3.

[0029] The question generation unit 11b performs a process of generating a question to fill in the missing information based on the trouble history information acquired by the trouble history monitoring unit 11a as meeting the above conditions. In this embodiment, the question generation unit 11b generates a question in natural language for the user using a large-scale language model stored in the model information storage unit 12b. The question generation unit 11b provides the trouble history information acquired by the trouble history monitoring unit 11a to the large-scale language model and also provides a command (prompt) to generate a question to fill in the missing information based on this information. The question generation unit 11b appropriately edits or processes the question output by the large-scale language model and transmits the question (question message) to the terminal device 5 used by one or more users related to the trouble history information.

[0030] The answer sentence acquisition unit 11c acquires the answer sentence (answer message) from the user in response to the question message sent by the question sentence generation unit 11b to the terminal device 5 by receiving it. If the answer message acquired by the answer sentence acquisition unit 11c is insufficient, the information processing device 1 may repeatedly send question messages to the user and acquire answer messages from the user. In this case, in the information processing device 1, the question sentence generation unit 11b determines whether an additional question is necessary based on the answer message acquired by the answer sentence acquisition unit 11c, and if an additional question is necessary, causes the large-scale language model to generate a question sentence. The answer acquired from the user in response to the question sentence does not necessarily have to be a sentence written in natural language. For example, the answer may be a number indicating which option to select in response to a question presenting multiple options, or a numerical value, character, or word that does not fully constitute a sentence, such as a number in response to a question asking about a setting value of a device.

[0031] When trouble history information related to the device is collected or collected through message exchange with the user, the information shaping unit 11d performs a process of shaping this information into information in a predetermined format. In this embodiment, the information shaping unit 11d uses a large-scale language model stored in the model information storage unit 12b to shape the information into a predetermined format. The information shaping unit 11d inputs the trouble history information acquired by the trouble history monitoring unit 11a from the trouble history DB 102 and information such as the user's reply message acquired by the reply message acquisition unit 11c into the large-scale language model, provides information regarding the predetermined format for shaping, and issues a command to shape the information into this predetermined format. The information shaping unit 11d acquires the formatted information output by the large-scale language model.

[0032] The DB processing unit 11e performs processes such as storing and reading information from the trouble response DB 101. The DB processing unit 11e stores information in a predetermined format formatted by the information formatting unit 11d in the trouble response DB 101. The DB processing unit 11e communicates with the user's terminal device 5 via the communication unit 13, accepts a request to read information from the user, reads the desired information from the trouble response DB 101, and transmits it to the terminal device 5. The DB processing unit 11e also accepts an information search request from the terminal device 5, searches for information that meets the search criteria from the information stored in the trouble response DB 101, and transmits a list of information such as the name of the trouble and the date and time of occurrence of the relevant information to the terminal device 5.

[0033] 4 is a block diagram showing an example of the configuration of the trouble history recording device 3 according to this embodiment. The trouble history recording device 3 according to this embodiment can be realized by installing a predetermined application program or the like in a general-purpose information processing device such as a personal computer or a server computer. The trouble history recording device 3 according to this embodiment is configured to include a processing unit 31, a storage unit 32, a communication unit 33, and the like. Note that in this embodiment, the processing of the trouble history recording device 3 is described as being performed by a single trouble history recording device 3, but the processing of the trouble history recording device 3 may be distributed among multiple devices. Furthermore, the processing of the trouble history recording device 3 may be performed by the information processing device 1. In other words, the information processing device 1 and the trouble history recording device 3 may be a single device.

[0034] The processing unit 31 is configured using an arithmetic processing device such as a CPU, MPU, GPU, or quantum processor, a ROM, and a RAM. The processing unit 31 performs processing such as collecting and recording information about the history of a trouble by reading and executing a program 32a stored in the storage unit 32. The storage unit 32 is configured using a large-capacity storage device such as a hard disk or SSD. The storage unit 32 stores the program 32a executed by the processing unit 31 and is also provided with a trouble history DB 102 that stores information about the history of a trouble. In this embodiment, the program (computer program, program product) 32a is provided in a form recorded on a recording medium 98 such as a memory card or optical disk, and the trouble history recording device 3 reads the program 32a from the recording medium 98 and stores it in the storage unit 32. However, the program 32a may also be provided in a form distributed via a network. The communication unit 33 transmits and receives data between the information processing device 1 and devices such as the terminal device 5 via a wired or wireless network N.

[0035] The trouble history DB 102 is a database that stores, as trouble history information, information obtained by one or more users regarding a trouble occurring in an apparatus such as a substrate processing apparatus or a transport apparatus, or information regarding the history of the user's response, etc. FIG. 5 is a schematic diagram showing an example of trouble history information stored in the trouble history DB 102. In this embodiment, the trouble history information stored in the trouble history DB 102 is information that chronologically lists messages in natural language exchanged between a group of multiple users responding to a trouble in an apparatus such as a substrate processing apparatus or a transport apparatus. The trouble history recording device 3 receives a message regarding a trouble response input by one user on the terminal device 5 and transmits this message to the terminal devices 5 of other users in the group to which the user belongs. At this time, the trouble history recording device 3 stores the message received from the terminal device 5 in the trouble history DB 102 as trouble history information.

[0036] The trouble history information stored in the trouble history DB 102 is information in which one or more pieces of message information are linked in chronological order to header information such as a "trouble ID," a "group ID," and a "completion flag." The "trouble ID" and "group ID" are appropriate character string information, and the "completion flag" is binary information that is set to either "completed" or "incomplete." In the illustrated example, the "trouble ID" in the header information is "A1234," the "group ID" is "group G," and the "completion flag" is "incomplete."

[0037] Each piece of message information linked to the header information may include information such as a timestamp, user ID, and message text. In the illustrated example, two pieces of message information are linked to the header information in chronological order. The first piece of message information has a timestamp of "September 15, 2022, 15:26," a user ID of "Person in Charge A," and a message text that reads, "A problem YY has occurred on device PP of M Company's FabX. I think we should take action kk. What do you think?" The second piece of message information has a timestamp of "September 16, 2022, 8:15," a user ID of "Person in Charge B," and a message text that reads, "In that case, wouldn't action II be more appropriate? There is a possibility that ZZ may occur."

[0038] A user who is the first to send a message regarding a trouble that has occurred in a device such as a substrate processing device or a transport device selects, for example, a group with which to share the message, and notifies the trouble history recording device 3 via the terminal device 5 that they will begin responding to the trouble. Upon receiving this notification from the terminal device 5, the trouble history recording device 3 generates a trouble ID and stores header information in the trouble history DB 102, and associates information about messages subsequently sent and received between users with this header information and stores them in chronological order. Furthermore, when the user has completed responding to the trouble, the user notifies the trouble history recording device 3 via the terminal device 5 that the trouble has been completed. Upon receiving this notification from the terminal device 5, the trouble history recording device 3 sets the "completion flag" in the header information to "completed."

[0039] In this embodiment, the trouble history recording device 3 is configured to include the trouble history DB 102, but this is not limiting. The trouble history DB 102 may be included in a device different from the trouble history recording device 3. In this case, the trouble history recording device 3 communicates with the device that includes the trouble history DB 102, and reads information from the trouble history DB 102 and writes information to the trouble history DB 102.

[0040] The storage unit 32 may be an external storage device connected to the trouble history recording device 3. The trouble history recording device 3 may be a multi-computer including multiple computers, or may be a virtual machine virtually constructed by software. The trouble history recording device 3 is not limited to the above configuration, and may include, for example, a reading unit that reads information stored in a portable storage medium, an input unit that accepts operation input, or a display unit that displays images.

[0041] In the trouble history recording device 3 according to this embodiment, the processing unit 31 reads and executes a program 32a stored in the storage unit 32, thereby realizing a message exchange processing unit 31a, a DB processing unit 31b, etc. as software functional units in the processing unit 11. In the figure, functional units related to the processing for collecting trouble history information are shown as functional units of the processing unit 31, and functional units related to other processing are not shown.

[0042] The message exchange processing unit 31a communicates with one or more terminal devices 5 via the communication unit 33 and processes message exchanges between multiple users to respond to device problems. The message exchange processing unit 31a receives a message from one user and transmits the received message to other users in the group to which this user belongs. Information such as the users belonging to the group and the message destinations for each user is registered in advance by the user and stored in the storage unit 32 as registered information.

[0043] The DB processing unit 31b performs processing to store information about messages sent and received between users by the message exchange processing unit 31a in the trouble history DB 102 as trouble history information. The DB processing unit 31b acquires messages received from users by the message exchange processing unit 31a, attaches information such as the date and time the message was received (timestamp) and the ID of the user who sent the message to the message text, and stores the message information in chronological order in association with the header information. In addition, the DB processing unit 31b searches and extracts information stored in the trouble history DB 102 in response to a request from the information processing device 1 or the terminal device 5, and transmits the resulting information to the requestor. The DB processing unit 31b can extract trouble history information whose "completion flag" is "incomplete" from the information stored in the trouble history DB 102, for example, and transmit the extracted information to the information processing device 1.

[0044] <Trouble Information Processing> Figure 6 is a flowchart showing an example of a processing procedure performed by the information processing device 1 according to this embodiment. The trouble history monitoring unit 11a of the processing unit 11 of the information processing device 1 according to this embodiment monitors the information stored in the trouble history DB 102 of the trouble history recording device 3 at a predetermined cycle, for example, once per hour or once per day. The trouble history monitoring unit 11a determines whether this predetermined cycle has elapsed and it is time to perform monitoring (step S1). If it is not time to perform monitoring (S1: NO), the trouble history monitoring unit 11a waits until it is time to perform monitoring.

[0045] When the time to perform monitoring has arrived (S1: YES), the trouble history monitoring unit 11a communicates with the trouble history recording device 3 via the communication unit 13, and acquires from the trouble history DB 102 trouble history information whose "completion flag" is "incomplete" and for which a predetermined period (e.g., one month) has passed since the timestamp of the last message (step S2). The process of searching the trouble history DB 102 for incomplete trouble history information for which a predetermined period has passed may be performed by the trouble history recording device 3 or may be performed directly by the information processing device 1.

[0046] Next, the question generator 11b of the processing unit 11 generates a question for the user based on the incomplete trouble history information acquired in step S2 (step S3). At this time, the question generator 11b provides the acquired incomplete trouble history information to the large-scale language model stored in the model information storage unit 12b, and also provides a command (prompt) to generate a question section that inquires about the subsequent situation, etc., thereby generating the question. Figures 7 to 9 are schematic diagrams showing examples of information input to the large-scale language model and information output by the large-scale language model.

[0047] The example shown in Figure 7 is a query generated based on the incomplete trouble history information shown in Figure 5. For example, the query generator 11b inputs the following text information into the large-scale language model: "In the following response, please create a query inquiring about the current status of the person in charge. Please output only the relevant question. ... September 15, 2022, 15:26 Staff A: "Trouble YY has occurred at M Company's FabX device PP. I think we should take action kk. What do you think?" September 16, 2022, 8:15 Staff B: "In that case, I think action II would be preferable. ZZ may occur." ~No response thereafter~." The first half of this input information, "Please respond as follows~," is a predefined phrase, while the second half is text information that expands the timestamp, user ID, and message information included in the trouble history information and arranges them in chronological order. In response to this input information, the large-scale language model outputs a query such as, "It seems that trouble YY has occurred at M Company's FabX device PP. What is the current situation?"

[0048] The example shown in Figure 8 is a case where a question is generated based on different trouble history information than that shown in Figure 7. For example, the question generator 11b inputs the following text information into the large-scale language model: "In the following response, please create a question to inquire about the current situation of the person in charge. Please output only the question. ... January 15, 2023, 22:26 Personnel A: "TT trouble occurred with device QQ of S Company's FabY. We performed action aa, but the situation did not improve. We ask for your assistance." January 16, 2023, 6:15 Personnel B: "Please check BB. If it is cc, perform action dd, then check uu, and then perform action aa." ~ No response thereafter ~" In response to this input information, the large-scale language model outputs a question such as, "What is the current status of device QQ of S Company's FabY?"

[0049] The example shown in FIG. 9 illustrates a case where a follow-up question is asked after a user's response to a question is obtained. A follow-up question is generated in response to the user's response to the question shown in FIG. 7 . For example, the question generator 11b inputs text information such as "[System Message] Please organize the information and generate a question. [Prompt] Please first organize the following items based on past responses. If there are any missing information, please generate a question to ask the person in charge. Trouble name, customer name, Fab No., equipment information, recipe type, probable cause, emergency measures, whether or not an experiment was conducted to identify the cause and its details, permanent measures, and information transfer destination for the permanent measures. <Response history including the most recent user response>" into the large-scale language model. This input information can be obtained, for example, by substituting actual user responses for the <Response history including the most recent user response> portion of a predefined template.

[0050] In response to this input information, the large-scale language model outputs the results of organizing the information, for example, "Trouble name: YY trouble, Customer name: M company, Fab No: FabX, Equipment information: Equipment PP, Recipe type: Unknown, Estimated cause: Unknown, First aid: Considered performing kk treatment, but performed II treatment, Whether or not an experiment was conducted to identify the cause and its details: Unknown, Permanent measure: Unknown, Where should information be transferred for the permanent measure: Unknown." The large-scale language model also outputs questions, for example, "Questions: 1. What are the specific symptoms of YY trouble? 2. What is the difference between kk treatment and II treatment? 3. What would be the results if an experiment were conducted? 4. Do you have any information about permanent measures? 5. Where should information be transferred in case a similar trouble occurs again?"

[0051] The question generator 11b, which generated these questions in step S3 of the flowchart shown in FIG. 6, transmits a question message containing the generated question to one or more terminal devices 5 used by one or more users in the group associated with the trouble (step S4). The answer acquisition unit 11c of the processing unit 11 receives a response message to the question message transmitted in step S4 from the terminal device 5 (step S5) and acquires the user's response. Based on the response message received in step S5, the answer acquisition unit 11c determines whether or not the information required to be stored in the trouble response DB 101 is complete (step S6). At this time, the answer acquisition unit 11c inputs information similar to that shown in FIG. 9 into the large-scale language model, and determines whether or not the required information is complete, provided that the information sorting results output by the large-scale language model do not include "unknown." If the required information is not complete (S6: NO), the answer acquisition unit 11c returns to step S3 and repeatedly exchanges questions and answers with the user.

[0052] When the necessary information is collected (S6: YES), the information shaping unit 11d of the processing unit 11 shapes the information acquired from the trouble history DB 102 and the information obtained through message exchanges with the user into a predetermined format for storage in the trouble response DB 101 (step S7). At this time, the information shaping unit 11d can shape the information by providing the information to be shaped to the large-scale language model stored in the model information storage unit 12b and by providing a command (prompt) to shape the information into the predetermined format. Figures 10 and 11 are schematic diagrams showing examples of information input to the large-scale language model and information output by the large-scale language model.

[0053] The example shown in FIG. 10 is a case where information shaping is performed in response to the user's answer to the question shown in FIG. 」 English: For example, the information formatting unit 11d inputs the following text information into the large-scale language model: "[System Message] Please organize and output the response history using the format_conversation function. [Prompt] September 15, 2022, 15:26 Staff A "Trouble YY has occurred with the equipment PP of M Company's FabX. I think we should take action kk. What do you think?" September 16, 2022, 8:15 Staff B "In that case, wouldn't action II be more preferable? There is a possibility that ZZ may occur." October 16, 2022, 8:15 reminder_bot "We understand that trouble YY has occurred with the equipment PP of M Company's FabX. What is the current situation?" October 16, 2022, 11:13 Staff A "We took action II on that matter and completed it. However, AA occurred as a result, so we took action pp. As a permanent solution, we believe that a change in the parts material is necessary, and this has been communicated to the m department."

[0054] In this embodiment, the information processing device 1 executes a pre-prepared format_conversation function to store information in the troubleshooting DB 101 in a predetermined format. The format_conversation function receives, as arguments, values ​​corresponding to the item names shown in FIG. 3 , and stores the received values ​​in corresponding locations in the troubleshooting DB 101. For example, the order of multiple arguments is predetermined, such as format_conversation (trouble name, occurrence date and time, ..., recording date and time), and the function specification information is provided in advance to the large-scale language model. The large-scale language model organizes and summarizes input natural language information to extract information to be provided as arguments to the format_conversation function, and outputs text information with arguments set to the format_conversation function, such as "format_conversation('YY Trouble', 'September 15, 2022, 15:26', ...)" in this example. The information processing device 1 obtains the text information of this function call obtained from the large-scale language model as the information formatting result, and executes this function to store information in a predetermined format in the troubleshooting DB 101.

[0055] The example shown in FIG. 11 is a case where information obtained through multiple questions and answers shown in FIG. 9 is shaped. For example, "[System Message] Please organize and output the response history using the format_conversation function. [Prompt] September 15, 2022 15:26 Staff A "YY trouble has occurred with M Company FabX's equipment PP. I think we should take kk action, what do you think?" September 16, 2022 8:15 Staff B "In that case, wouldn't II action be more preferable? There is a possibility that ZZ will occur." October 16, 2022 8:15 reminder_bot "We understand that YY trouble has occurred with M Company FabX's equipment PP. What is the current situation?" October 16, 2022 8:15 Staff A "II action has been taken and completed." October 16, 2022 8:16 reminder_bot "1. What are the specific symptoms of YY trouble? 2. What is the difference between kk action and II action? 3. What would be the results if an experiment were conducted? 4. Do you have any information about permanent measures? 5. Where should we take over this information in case a similar problem occurs again?" October 16, 2022, 11:18 Staff A: "II part is burned out, KK treatment will not be performed on BB, but II treatment will be performed. No specific experiments have been conducted. A permanent change to the material of ii part is required. M department has been contacted." This text information is input into the large-scale language model by the information formatting unit 11d. In response to this input, the large-scale language model outputs text information with arguments set in the format_conversation function, such as "format_conversation("YY trouble", "September 15, 2022, 15:26", ...)."

[0056] 6, the DB processing unit 11e of the processing unit 11 stores the formatted information in the predetermined format in the troubleshooting DB 101 (step S8), and then ends the process. At this time, the DB processing unit 11e can store the information in the predetermined format in the troubleshooting DB 101 by executing the format_conversation function based on the function call information obtained from the large-scale language model in step S7.

[0057] In the information processing system according to this embodiment, for example, a user who wants to obtain existing knowledge about troubleshooting can access the information processing device 1 using the terminal device 5 and search for and view information stored in the troubleshooting DB 101. The user inputs, for example, the name, cause, or device name of a problem into the terminal device 5 as search criteria, and searches for information stored in the troubleshooting DB 101. The terminal device 5 transmits the input search criteria and a request for information search based on the search criteria to the information processing device 1. In response to a request from the terminal device 5, the information processing device 1 extracts information that matches the search criteria from the troubleshooting DB 101 and transmits the extracted information or summary information of the information as search results to the terminal device 5 that made the request. The terminal device 5 receives search result information from the information processing device 1 and displays the received information on a display unit.

[0058] 12 is a schematic diagram showing an example of information display by the terminal device 5. In this embodiment, the terminal device 5 displays, on the display unit, an input box for inputting words or phrases that serve as search criteria, and a search button for receiving an instruction to execute a search, and receives a search operation from the user for information related to troubleshooting. The terminal device 5 also displays a list of search results in a table format based on information obtained from the information processing device 1 as a response to the search request. The example shown in FIG. 12 displays a list of search results when the term "ESC" is specified as a search criterion.

[0059] In this example, the terminal device 5 presents the search results to the user by displaying a list of items such as "relevance," "trouble name," "(estimated) cause," "permanent solution," and "related materials" for information related to troubleshooting that matches the search criteria. Among these items, "relevance" is a numerical value obtained by, for example, using a pre-prepared encoder, converting the words or phrases entered as search criteria and the information related to troubleshooting that matches the search criteria into feature vectors, and then calculating the similarity (e.g., cosine similarity) between the two vectors. The terminal device 5 lists multiple pieces of information, arranging them vertically in descending order of relevance. The four items, "trouble name" to "related materials," are items included in the information stored in the troubleshooting DB 101, and the terminal device 5 can extract some or all of the information from the corresponding items and display it in a table.

[0060] In this example, the search result for the search criteria "ESC" shows that information such as "ESC trouble" and "ESC abnormality" as "trouble name" is stored in the trouble response DB 101. For example, "ESC trouble" with the highest "relevance" of "0.897" shows that the "(presumed) cause" is "uneven wear within the surface" and the "permanent measure" is "first, it is necessary to identify the wear mechanism." The user can select one of the multiple pieces of information displayed in the list and, for example, click or touch the selected trouble name to display details of this information on the terminal device 5 (not shown).

[0061] <Summary> In the information processing system according to the present embodiment configured as described above, the information processing device 1 acquires trouble history information relating to the history of responses to events that occurred in devices such as substrate processing devices or transport devices from the trouble history recording device 3. The information processing device 1 generates a natural language question relating to the acquired trouble history information using a large-scale language model, and outputs the generated question to the user's terminal device 5. The information processing device 1 acquires the user's answer to the question from the terminal device 5, and stores information relating to the acquired answer in the trouble response DB 101. As a result, the information processing system according to the present embodiment can be expected to support the collection of information relating to events in devices.

[0062] In the information processing system according to this embodiment, the information processing device 1 stores information about the history of the trouble, acquires information about incomplete troubleshooting from the trouble history DB 102, and generates a question about the current state of the trouble. As a result, the information processing system according to this embodiment is expected to collect information about incomplete troubleshooting.

[0063] Furthermore, in the information processing system according to this embodiment, the information processing device 1 formats a reply sentence acquired from a user into information in a predetermined format using a large-scale language model, and stores the formatted information in the trouble response DB 101. This allows the information processing system according to this embodiment to accept a reply sentence in natural language from a user, and to format the accepted reply sentence in natural language into a format suitable for the database and store it.

[0064] Furthermore, in the information processing system according to this embodiment, when the information processing device 1 acquires an answer from a user that lacks information for formatting the answer into information in a predetermined format, the information processing device 1 generates a question regarding the missing information, transmits the generated question to the terminal device 5 to present it to the user, and acquires the user's answer to the question. As a result, the information processing system according to this embodiment can be expected to collect all the information to be stored in the database.

[0065] The embodiments disclosed herein are to be considered as illustrative in all respects and not restrictive. The scope of the present disclosure is defined by the claims, not by the above meaning, and is intended to include all modifications within the meaning and scope of the claims.

[0066] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. A multiple claim (multi-multi claim) that references at least one other multiple claim may also be used.

[0067] REFERENCE SIGNS LIST 1 Information processing device (computer) 3 Trouble history recording device 5 Terminal device 11 Processing unit 11a Trouble history monitoring unit 11b Question sentence generation unit 11c Answer sentence acquisition unit 11d Information shaping unit 11e DB processing unit 12 Storage unit 12a Program (computer program) 12b Model information storage unit 13 Communication unit 31 Processing unit 31a DB processing unit 31b Message exchange processing unit 32 Storage unit 32a Program 33 Communication unit 98, 99 Recording medium 101 Trouble response DB (first database) 102 Trouble history DB (second database) 103 Large-scale language model (language model) N Network

Claims

1. A computer program that causes a computer to execute a process of obtaining information regarding the history of responses to events that have occurred in a device, generating a natural language question sentence regarding the obtained information using a language model, outputting the generated question sentence, obtaining an answer to the question sentence, and storing information regarding the obtained answer in a first database.

2. The computer program according to claim 1, wherein the information regarding the history of responses to events that have occurred in the device includes the opinions of one or more persons regarding the state of the device.

3. The computer program according to claim 1, which obtains information regarding events for which the response is incomplete from a second database storing information regarding the history of responses to events that have occurred in the device, and generates a question sentence for inquiring about the current situation regarding the response to the event for the obtained incomplete information.

4. The computer program according to claim 1, which formats the obtained answer into information in a predetermined format using a language model, and stores the formatted information in the first database.

5. The computer program according to claim 4, which generates a natural language question sentence regarding the information that is lacking when there is insufficient information for formatting the obtained answer into the information in the predetermined format, outputs the generated question sentence, and obtains an answer to the question sentence.

6. The computer program according to claim 4, which generates information for executing a function for storing information in the first database in the predetermined format using a language model, executes the function based on the generated information, and stores the information formatted in the predetermined format in the first database.

7. The computer program according to claim 1, which obtains information regarding the history of responses to troubles that have occurred in a substrate processing device.

8. The computer program according to claim 1, which receives search conditions for information stored in the first database, and extracts and outputs information that matches the received conditions from the first database.

9. An information processing method, wherein an information processing apparatus acquires information regarding the process of responding to an event that has occurred in the apparatus, generates a natural language question sentence regarding the acquired information by means of a language model, outputs the generated question sentence, acquires an answer to the question sentence, and stores information regarding the acquired answer in a first database.

10. An information processing apparatus comprising a processing unit, wherein the processing unit acquires information regarding the process of responding to an event that has occurred in the apparatus, generates a natural language question sentence regarding the acquired information by means of a language model, outputs the generated question sentence, acquires an answer to the question sentence, and stores information regarding the acquired answer in a first database.

11. An information processing system comprising the information processing apparatus according to claim 10, the first database, and a second database that stores information regarding the process of responding to an event that has occurred in the apparatus.

Images