JP7568359B2 - Server device, customer support service providing method and customer support service providing program - Google Patents

Description

The present disclosure relates to a server device, a method for providing customer support services, and a program for providing customer support services.

Interactive customer support services using chatbots that automatically respond to customer inquiries have been known for some time. These services can automatically respond to inquiries from a large number of customers at any time. For this reason, chatbots have recently been introduced not only for products aimed at general consumers, but also for specific equipment (e.g., semiconductor manufacturing equipment, measuring equipment, etc., whose specifications vary depending on the delivery destination) that is delivered to the offices or research laboratories of manufacturers.

On the other hand, chatbots are limited in the amount of information they can provide to customers when solving trouble events. In addition, the information that can be provided varies from customer to customer, so simply providing generic information is not sufficient to solve trouble events. For this reason, in the case of specific devices, it is difficult to provide high-quality customer support services using chatbots alone.

JP 2020-61087 A

The present disclosure provides a server device, a method for providing customer support services, and a program for providing customer support services that improve the quality of customer support services.

According to one aspect, a server device includes:
a first providing unit that searches for first information based on input information input to the chat window and displays a search result in the information window during the first mode;
a second providing unit that searches for second information based on input information input to the chat window and displays at least a part of a search result in the information window during a second mode ;
Either one or both of the first providing unit and the second providing unit classify search results into a plurality of groups based on the input information, and display the groups in different tabs within the information window .

According to the present disclosure, it is possible to provide a server device, a method for providing customer support services, and a program for providing customer support services that improve the service quality in customer support services.

FIG. 1 illustrates an example of an overall configuration of a customer support service providing system. 5A to 5C are diagrams illustrating an example of information stored in each storage unit. FIG. 2 illustrates an example of a hardware configuration of a customer support service device. 13 is a diagram illustrating a specific example of a functional configuration of a public information analysis unit and an analysis process in the analysis phase. FIG. 13 is a diagram illustrating a specific example of a functional configuration of a non-public information analysis unit and an analysis process in the analysis phase. FIG. FIG. 13 is a diagram illustrating an example of a functional configuration of a customer support service device and a CRM terminal in a response phase. FIG. 11 is a first diagram showing an example of a display screen provided in an automatic response mode. FIG. 2 is a second diagram showing an example of a display screen provided in the automatic response mode. FIG. 3 is a third diagram showing an example of a display screen provided in the automatic response mode. FIG. 4 is a fourth diagram showing an example of a display screen provided in the automatic response mode. FIG. 11 is a first diagram showing an example of a display screen provided in an operator response mode. FIG. 2 is a second diagram showing an example of a display screen provided in an operator response mode. FIG. 3 is a third diagram showing an example of a display screen provided in an operator response mode. FIG. 13 is a diagram showing an example of a management screen of a CRM terminal displayed in an automatic response mode. 10 is a flowchart showing a flow of a customer support service providing process in the customer support service providing system. 13 is a flowchart showing the flow of an automatic response process. 13 is a flowchart showing the flow of an operator response process.

Each embodiment will be described below with reference to the accompanying drawings. Note that in this specification and the drawings, components having substantially the same functional configuration are designated by the same reference numerals, and redundant description will be omitted.

[First embodiment]
<Overall configuration of customer support service provision system>
First, the overall configuration of the customer support service providing system will be described. The customer support service providing system in this embodiment is a system that provides an interactive customer support service using a chatbot, with the semiconductor manufacturing equipment and measuring equipment delivered to each business establishment having a semiconductor manufacturing process as specific equipment.

Figure 1 is a diagram showing an example of the overall configuration of a customer support service providing system. As shown in Figure 1, the customer support service providing system 100 has a customer support service device 110, which is an example of a server device, and a CRM (Customer Relationship Management) terminal 120. The customer support service providing system 100 also has terminals 131, 141, and 151 in each of the business offices 130, 140, and 150 (business office names = "Business Office A", "Business Office B", and "Business Office C").

The customer support service device 110 and the terminals 131, 141, and 151 in each of the business locations 130, 140, and 150 are communicatively connected via a network 160 such as a WAN (Wide Area Network). The customer support service device 110 and the CRM terminal 120 are communicatively connected via a network such as a LAN (Local Area Network).

A customer support service providing program is installed in the customer support service device 110. By executing the program, the customer support service device 110:
Public information analysis unit 111,
・Private information analysis unit 112,
・Public information provision department 113,
・Private information provision department 114,
Chat response unit 115,
The customer support service device 110 has an analysis phase and a response phase, and among the above-mentioned units, the public information analysis unit 111 and the private information analysis unit 112 operate in the analysis phase. On the other hand, the public information providing unit 113, the private information providing unit 114, and the chat response unit 115 operate in the response phase.

The public information analysis unit 111 performs natural language processing on the public information, for example, by analyzing the frequency of occurrence of each of the multiple phrases contained in the public information that are related to past trouble events, thereby quantifying the public information.

Public information is an example of the first information, and refers to generic data that does not include either customer-specific data or development data. In this embodiment, the public information used in the analysis process by the public information analysis unit 111 includes customer information stored in the customer information storage unit 116 (an example of a storage unit) and chatbot history information stored in the chatbot history information storage unit 117 (an example of a storage unit).

The non-public information analysis unit 112 performs natural language processing on the non-public information, for example, by analyzing the frequency of occurrence of each of the multiple phrases contained in the non-public information that are related to past trouble events, thereby quantifying the non-public information.

The non-public information is an example of the second information, and refers to data that includes either customer-specific data or development data, or both. In this embodiment, the non-public information used in the analysis process by the non-public information analysis unit 112 includes internal information stored in the internal information storage unit 118 (an example of a storage unit).

The public information and private information stored in each storage unit are both non-standard documents, and the public information has been processed so that specific customer names, etc. cannot be identified. The private information includes both information that has been processed so that specific customer names, etc. cannot be identified, and information that has not been processed.

The chat response unit 115 is an example of a control unit, and switches between an automatic response mode, which is an example of a first mode, and an operator response mode, which is an example of a second mode. In the automatic response mode, the chat response unit 115 accepts customer inquiries about trouble events via chat from terminals 131-151 in each of the business offices 130-150. The chat response unit 115 also automatically responds to accepted inquiries at any time using a chatbot. A chatbot is an automatic conversation program that utilizes AI (Artificial Intelligence) to perform natural language processing.

In addition, in order to search for public information corresponding to the received inquiry, the chat response unit 115 extracts specific input information (referred to as customer input information) from the received inquiry content and notifies the public information provision unit 113. The customer input information is any information related to the trouble event, and includes, for example, the device name, device model, alarm ID, details of the trouble event, etc. As a result, the customer support service device 110 can immediately provide the customer with public information corresponding to the customer input information in response to the customer's inquiry in order to resolve the trouble event.

Furthermore, the chat response unit 115 checks whether the trouble has been resolved based on the response using the chatbot or the publicly available information provided. If the chat response unit 115 determines that the trouble has not been resolved (including cases where publicly available information cannot be found), it switches from automatic response mode to operator response mode with the customer's approval. This makes it possible for an operator (e.g., a developer of semiconductor manufacturing equipment or measuring equipment) to respond by chat via the CRM terminal 120.

The chat response unit 115 continues to extract customer-input information from the received inquiry even after switching to the operator response mode. However, after switching to the operator response mode, the extracted customer-input information is notified to the private information provision unit 114.

The public information providing unit 113 is an example of a first providing unit, and searches for public information corresponding to the customer input information notified by the chat response unit 115, and provides the searched public information to the customer. The public information providing unit 113 searches for public information corresponding to the customer input information from among the public information stored in the customer information storage unit 116 and the chatbot history information storage unit 117.

The non-public information providing unit 114 is an example of a second providing unit, and searches for non-public information corresponding to the customer input information notified by the chat response unit 115, and transmits the searched non-public information to the CRM terminal 120. This allows the operator to select appropriate non-public information from the non-public information notified by the non-public information providing unit 114 and immediately provide it to the customer when switching to the operator response mode. Furthermore, when providing the selected non-public information to the customer, the operator checks the non-public information, and if it contains information that should be concealed (such as a customer name, factory name, individual name, or information that corresponds to a trade secret), processes the non-public information and provides it to the customer with part of the information concealed. On the other hand, if it does not contain information that should be concealed, the operator provides the non-public information to the customer without processing it.

In this way, in the operator response mode, at least a portion of the search results (non-public information selected by the operator from the search results, and non-public information selected by the operator other than information that should be concealed) are provided to the customer.

The non-public information providing unit 114 searches for non-public information corresponding to the customer input information from among the non-public information stored in the internal information storage unit 118.

In the automatic response mode, the CRM terminal 120 manages responses using a chatbot by the chat response unit 115. In addition, if the trouble event is not resolved in the automatic response mode and the CRM terminal 120 switches to the operator response mode, it accepts chat input from the operator and responds on behalf of the chatbot.

The CRM terminal 120 also provides the customer with private information that is notified by the private information provision unit 114 and selected by the operator. The CRM terminal 120 acquires equipment log information collected by the semiconductor manufacturing equipment from the terminals 131-151 in each of the business locations 130-150, and stores the equipment log information together with the analysis results in the equipment log information storage unit 121. The analysis results stored in the equipment log information storage unit 121 are provided to the customer.

The CRM terminal 120 may acquire device log information, for example, under a maintenance contract with a customer, or may acquire device log information attached by the customer to chat input information.

Semiconductor manufacturing equipment that executes the semiconductor manufacturing process is installed in business establishment 130 (business establishment name = "business establishment A"). Also installed in business establishment 130 is measuring equipment that measures measurement data in the semiconductor manufacturing process. Furthermore, also installed in business establishment 130 are terminals 131 that constitute customer support service providing system 100, and a database that stores various information (including equipment log information).

The semiconductor manufacturing equipment executes the semiconductor manufacturing process. The semiconductor manufacturing equipment also stores various information (including equipment log information) generated by executing the semiconductor manufacturing process in a database.

The measurement device measures measurement data during or after the semiconductor manufacturing process is performed by the semiconductor manufacturing device, and stores the data in a database.

When a trouble event occurs in the semiconductor manufacturing equipment or the measuring equipment, the terminal 131 accesses the customer support service device 110 and transmits the details of the trouble event (including the type of equipment, alarm ID, etc.) using chat. In addition, in response to transmitting the details of the trouble event, the terminal 131 displays the response content, public information, and private information transmitted from the customer support service device 110. This allows the worker at business establishment A to resolve the trouble event.

Furthermore, the terminal 131 reads the equipment log information of the semiconductor manufacturing equipment from the database and transmits it to the CRM terminal 120.

At business establishment 140 (business establishment name = "business establishment B") and business establishment 150 (business establishment name = "business establishment C"), the same semiconductor manufacturing process as at business establishment 130 is carried out. For this reason, business establishments 140 and 150 are equipped with the same equipment as at business establishment 130. Note that a description of the equipment installed at business establishments 140 and 150 will be omitted.

<Specific examples of information stored in each storage unit>
Next, a specific example of information stored in each storage unit (customer information storage unit 116, chatbot history information storage unit 117, internal information storage unit 118) of the customer support service device 110 will be described. FIG. 2 is a diagram showing an example of information stored in each storage unit. As shown in FIG. 2, the customer information storage unit 116 stores delivered device information, trouble history information, and Call Log as customer information. The delivered device information is information about devices (e.g., semiconductor manufacturing devices, measuring devices, etc.) delivered to customers. The trouble history information is information about past trouble events for devices delivered to customers. The Call Log is information that a call center (not shown) has received in the past about devices delivered to customers. The call center is a customer support service center where operators respond to inquiries from customers by phone.

All of this information stored as customer information in the customer information storage unit 116 is public information, and is provided to the customer when the chat response unit 115 automatically responds using a chatbot in the automatic response mode.

As shown in FIG. 2, the chatbot history information storage unit 117 stores inquiry history information as chatbot history information. The inquiry history information is information that records the contents of inquiries from customers and the contents of responses when the chat response unit 115 automatically responded in the past using a chatbot.

The information stored as chatbot history information in the chatbot history information storage unit 117 is public information, like the customer information storage unit 116, and is information provided to the customer when the chat response unit 115 automatically responds using a chatbot in the automatic response mode.

Furthermore, as shown in FIG. 2, the internal information storage unit 118 stores specifications, procedures, requirement data, and defect management data as internal information. Specifications are information about the specifications of equipment (e.g., semiconductor manufacturing equipment, measuring equipment, etc.) delivered to a customer. Procedures are information about the operating procedures for equipment (e.g., semiconductor manufacturing equipment, measuring equipment, etc.) delivered to a customer. Requirement data and defect management data are data describing the requirement specifications for equipment (e.g., semiconductor manufacturing equipment, measuring equipment, etc.) delivered to a customer and data describing the cause of a defect (alarm details) when a defect occurs.

All of this information stored as internal information in the internal information storage unit 118 is confidential information, and is provided to the customer when the operator responds via the CRM terminal 120 in the operator response mode.

The information stored in each storage unit shown in FIG. 2 is configured to be updated periodically, with the latest information being added successively. Furthermore, the information stored in each storage unit shown in FIG. 2 is merely illustrative of specific examples of public or private information, and is not limited to these specific examples.

<Hardware Configuration of Customer Support Service Device>
Next, a description will be given of the hardware configuration of the customer support service device 110. Fig. 3 is a diagram showing an example of the hardware configuration of the customer support service device.

As shown in FIG. 3, the customer support service device 110 has a processor 301, a memory 302, an auxiliary storage device 303, an I/F (Interface) device 304, a communication device 305, and a drive device 306. Each piece of hardware in the customer support service device 110 is connected to each other via a bus 307.

The processor 301 has various computing devices such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit). The processor 301 reads various programs (e.g., a customer support service provision program, etc.) onto the memory 302 and executes them.

The memory 302 has a primary storage device such as a ROM (Read Only Memory) or a RAM (Random Access Memory). The processor 301 and the memory 302 form a so-called computer, and the processor 301 executes various programs read onto the memory 302, causing the computer to realize various functions (for example, from the public information analysis unit 111 to the chat response unit 115).

The auxiliary storage device 303 stores various programs and various data used when the various programs are executed by the processor 301. For example, the customer information storage unit 116, the chatbot history information storage unit 117, and the internal information storage unit 118 are realized in the auxiliary storage device 303.

The I/F device 304 is a connection device that connects to the CRM terminal 120, which is an example of an external device.

The communication device 305 is a communication device for communicating with the terminals 131-151 in each of the business locations 130-150 via the network 160.

The drive device 306 is a device for setting the recording medium 310. The recording medium 310 here includes media that record information optically, electrically, or magnetically, such as CD-ROMs, flexible disks, and magneto-optical disks. The recording medium 310 may also include semiconductor memory that records information electrically, such as ROMs and flash memories.

The various programs to be installed in the auxiliary storage device 303 are installed, for example, by setting the distributed recording medium 310 in the drive device 306 and reading the various programs recorded on the recording medium 310 by the drive device 306. Alternatively, the various programs to be installed in the auxiliary storage device 303 may be installed by downloading them from the network 160 via the communication device 305.

<Functional configuration of the customer support service device in the analysis phase>
Next, a specific example of the functional configuration and analysis process of the public information analysis unit 111 and the private information analysis unit 112 that operate in the analysis phase among the units of the customer support service device 110 will be described.

(1) Public Information Analysis Unit First, a specific example of the functional configuration and analysis process of the public information analysis unit 111 will be described. Fig. 4 is a diagram showing a specific example of the functional configuration and analysis process of the public information analysis unit in the analysis phase. As shown in Fig. 4, the public information analysis unit 111 has an information acquisition unit 410, a trained analysis model 420, an occurrence frequency digitization unit 430, and an analysis result output unit 440.

The information acquisition unit 410 acquires public information and inputs it to the trained analysis model 420. Specifically, the information acquisition unit 410 reads delivered device information, trouble history information, and call logs from the customer information storage unit 116, and inputs them sequentially to the trained analysis model 420. The information acquisition unit 410 also reads inquiry history information from the chatbot history information storage unit 117, and inputs them sequentially to the trained analysis model 420.

The trained analysis model 420 is a model that has been trained in advance, and performs natural language processing on the public information input by the information acquisition unit 410. By performing natural language processing on the public information, the trained analysis model 420 analyzes, for example, the frequency of occurrence of each of the multiple phrases included in the public information that are related to past trouble events.

The occurrence frequency quantification unit 430 quantifies (e.g., normalizes) the occurrence frequency of each word related to past trouble events analyzed by the trained analysis model 420.

The analysis result output unit 440 stores the occurrence frequency of each word related to past trouble events, quantified by the occurrence frequency quantification unit 430, as the analysis result, in association with the public information.

The example of Fig. 4 shows how the analysis result output unit 440 associates the analysis results _S11 , _S12 , ... with each of a plurality of pieces of delivered device information and stores them in the customer information storage unit 116. The example of Fig. 4 also shows how the analysis result output unit 440 associates the analysis results _S21 , _S22 , ... with each of a plurality of pieces of trouble history information and stores them in the customer information storage unit 116. The example of Fig. 4 also shows how the analysis result output unit 440 associates the analysis results _S31 , _S32 , ... with each of a plurality of Call Logs and stores them in the customer information storage unit 116. Furthermore, the example of Fig. 4 shows how the analysis result output unit 440 associates the analysis results _S41 , _S42 , ... with each of a plurality of pieces of inquiry history information and stores them in the chatbot history information storage unit 117.

The analysis process by the public information analysis unit 111 is performed periodically, and the analysis results are periodically associated with newly added public information to the customer information storage unit 116 and the chatbot history information storage unit 117.

(2) Non-public information analysis unit Next, a specific example of the functional configuration and analysis process of the non-public information analysis unit 112 will be described. Fig. 5 is a diagram showing a specific example of the functional configuration and analysis process of the non-public information analysis unit. As shown in Fig. 5, the non-public information analysis unit 112 has an information acquisition unit 510, a trained analysis model 520, an occurrence frequency quantification unit 530, and an analysis result output unit 540.

The information acquisition unit 510 acquires non-public information and inputs it to the trained analysis model 520. Specifically, the information acquisition unit 510 reads specifications, procedures, requirements data, and defect management data from the internal information storage unit 118, and inputs them sequentially to the trained analysis model 520.

The trained analysis model 520 is a model that has been trained in advance, and performs natural language processing on the private information input by the information acquisition unit 510. By performing natural language processing on the private information, the trained analysis model 520 analyzes, for example, the frequency of occurrence of each of the multiple phrases included in the private information that are related to past trouble events.

The occurrence frequency quantification unit 530 quantifies (e.g., normalizes) the occurrence frequency of each word related to past trouble events analyzed by the trained analysis model 520.

The analysis result output unit 540 stores the occurrence frequency of each word related to past trouble events, quantified by the occurrence frequency quantification unit 530, as the analysis result, in association with the non-public information.

The example of Fig. 5 shows how the analysis result output unit 540 associates analysis results _S51 , _S52 , ... with each of a plurality of specifications and stores them in the internal information storage unit 118. The example of Fig. 5 also shows how the analysis result output unit 540 associates analysis results _S61 , _S62 , ... with each of a plurality of procedure manuals and stores them in the internal information storage unit 118. Furthermore, the example of Fig. 5 shows how the analysis result output unit 540 associates analysis results _S71 , _S72 , ... with each of a plurality of requirement data and defect management data and stores them in the internal information storage unit 118.

The analysis process by the non-public information analysis unit 112 is performed periodically, and the analysis results are periodically associated with newly added non-public information to the internal information storage unit 118.

<Functional configuration of the customer support service device and CRM terminal in the response phase>
Next, the functional configurations of the public information providing unit 113, the private information providing unit 114, and the chat response unit 115, which operate in the response phase among the units of the customer support service device 110, and the functional configuration of the CRM terminal 120 in the response phase will be described.

Figure 6 is a diagram showing an example of the functional configuration of a customer support service device and a CRM terminal in the response phase. As shown in Figure 6, in the response phase, the chat response unit 115 functions as a chat window control unit 601 and a customer input information acquisition unit 602.

The chat window control unit 601 provides chat windows to terminals 131-151 in response to access from terminals 131-151 in each of the business locations 130-150.

In addition, in the automatic response mode, the chat window control unit 601 uses a chatbot to generate chat response information in response to chat input information entered by a customer in the chat window, and provides the chat response information to the customer.

In addition, in the operator response mode, the chat window control unit 601 provides the customer with chat response information input by the operator via the CRM terminal 120 in response to chat input information input by the customer in the chat window.

Furthermore, the chat window control unit 601 extracts customer input information from the chat input information regardless of whether the mode is the automatic response mode or the operator response mode, and notifies the customer input information acquisition unit 602.

When the customer input information acquisition unit 602 is notified of the customer input information from the chat window control unit 601, in the automatic response mode, the notified customer input information is processed by the public information analysis unit 611 and the similarity search unit 612 of the public information providing unit 113.
A private information analysis unit 621 and a similarity search unit 622 of the private information providing unit 114;
In the operator response mode, the customer input information acquisition unit 602 notifies both of the customer input information,
A private information analysis unit 621 and a similarity search unit 622 of the private information providing unit 114;
Notify.

Also, as shown in FIG. 6, in the response phase, the public information providing unit 113 functions as a public information analysis unit 611, a similarity search unit 612, and an information window control unit 613.

The public information analysis unit 611 has the same functions as the public information analysis unit 111 used in the analysis process in the analysis phase. However, in the case of the public information analysis unit 611, when customer input information notified by the customer input information acquisition unit 602 is input, the analysis result is notified to the similarity search unit 612.

The similarity search unit 612 filters (narrows) the public information stored in the customer information storage unit 116 and the chatbot history information storage unit 117 based on the customer input information. The similarity search unit 612 filters based on customer input information such as device name, device model, alarm ID, etc.

The similarity search unit 612 searches the filtered public information for public information associated with an analysis result similar to the analysis result notified by the public information analysis unit 611 (for example, the analysis result of customer input information such as the content of the trouble event). The analysis result used when searching for public information may be an analysis result of customer input information including the device name, device model, alarm ID, etc. in addition to the content of the trouble event, etc.

The similarity search unit 612, for example, compares the frequency of occurrence (quantified) of each word contained in the analysis results to calculate the degree of similarity between them, and then calculates the degree of similarity with the public information by weighting and adding up the calculated degrees of similarity.

Furthermore, the similarity search unit 612 notifies the information window control unit 613 of the top N pieces of public information in terms of similarity. The similarity search unit 612 may notify the information window control unit 613 of the top N pieces of public information whose similarity meets a predetermined condition (has a similarity equal to or greater than a predetermined threshold) from the entire filtered public information. Alternatively, the similarity search unit 612 may classify the filtered public information into a plurality of groups, and notify the information window control unit 613 of the top N pieces of public information whose similarity meets a predetermined condition (has a similarity equal to or greater than a predetermined threshold) for each group. The public information belonging to each group may be stored in a different database in advance in each storage unit, or may be stored in the same database and notified separately when notifying the information window control unit 613.

The information window control unit 613 provides an information window adjacent to the chat window that displays public information in the automatic response mode. The information window control unit 613 also displays the public information notified by the similarity search unit 612 in the information window.

Also, as shown in FIG. 6, in the response phase, the private information providing unit 114 functions as a private information analysis unit 621, a similarity search unit 622, and an information window control unit 623.

The private information analysis unit 621 has the same functions as the private information analysis unit 112 used in the analysis process in the analysis phase. However, in the case of the private information analysis unit 621, when customer input information notified by the customer input information acquisition unit 602 is input, the analysis result is notified to the similarity search unit 622.

The similarity search unit 622 filters (narrows) the non-public information stored in the internal information storage unit 118 based on customer-input information. The similarity search unit 622 filters based on customer-input information such as device name, device model, and alarm ID.

The similarity search unit 622 searches the filtered private information for private information associated with an analysis result similar to the analysis result notified by the private information analysis unit 621 (for example, the analysis result of customer input information such as the content of the trouble event). The analysis result used when searching for private information may be the analysis result of analysis of customer input information including the device name, device model, alarm ID, etc. in addition to the content of the trouble event, etc.

The similarity search unit 622, for example, compares the frequency of occurrence (quantified) of each word contained in the analysis results to calculate the degree of similarity between them, and then calculates the degree of similarity with non-public information by weighting and adding up the calculated degrees of similarity.

Furthermore, the similarity search unit 622 notifies the CRM terminal 120 of the private information with the top N similarities. The similarity search unit 622 may notify the CRM terminal 120 of the private information with the top N similarities from the entire filtered private information. Alternatively, the similarity search unit 622 may classify the filtered private information into multiple groups and notify the CRM terminal 120 of the private information with the top N similarities for each group. The private information belonging to each group may be stored in different databases in the internal information storage unit 118, or may be stored in the same database and notified separately when notifying the information window control unit 623.

The information window control unit 623 provides an information window adjacent to the chat window that displays private information that is searched and selected in the operator response mode. The information window control unit 623 also displays private information notified by the CRM terminal 120 in the information window.

Also, as shown in FIG. 6, in the response phase, the CRM terminal 120 functions as a chat management unit 631, a chat input unit 632, and a private information management unit 633.

In the automatic response mode, the chat management unit 631 manages the state of the chat response unit 115 of the customer support service device 110 and displays a management screen to the operator. In addition, when the customer support service device 110 switches from the automatic response mode to the operator response mode, the chat management unit 631 activates the chat input unit 632 and the private information management unit 633.

The chat input unit 632 is activated when the mode is switched to the operator response mode. The chat input unit 632 accepts chat response information entered by the operator in response to chat input information entered in the chat window, and transmits it to a terminal (for example, one of terminals 131 to 151) via the chat window control unit 601.

The non-public information management unit 633 is activated when the mode is switched to the operator response mode. The non-public information management unit 633 notifies the information window control unit 623 of the non-public information selected by the operator from among the non-public information notified by the similarity search unit 622. As a result, the non-public information selected by the operator is provided to the customer. When providing the selected non-public information to the customer, the operator checks the non-public information, and if it contains information that should be concealed (such as a customer name, factory name, individual name, or information that corresponds to a trade secret), the operator processes the non-public information and provides it to the customer with part of the information concealed. On the other hand, if it does not contain information that should be concealed, the operator provides the non-public information to the customer without processing it.

<Specific examples of display screens provided in the automatic response mode>
Next, the display screen in the automatic response mode provided to a terminal (e.g., any one of terminals 131 to 151) that has accessed customer support service device 110 will be described with reference to Figures 7 to 9. Figures 7 to 9 are first to third diagrams showing an example of a display screen provided in the automatic response mode.

As shown in FIG. 7, the display screen 700 has a chat window 710 and an information window (public information) 720. The chat window 710 displays chat input information entered by the customer and chat response information generated by the chat response unit 115 using a chatbot.

In the chat window 710, the information written in the speech bubble located on the left side is chat response information generated by the chat window control unit 601 of the chat response unit 115. Also, in the chat window 710, the information written in the speech bubble located on the right side is chat input information entered by the customer.

The example in FIG. 7(a) shows the chat window control unit 601 of the chat response unit 115 asking the customer if they have an alarm ID, and the customer responds that they have an alarm ID. The example in FIG. 7(b) shows the chat window control unit 601 of the chat response unit 115 prompting the customer to input an alarm ID. Furthermore, the example in FIG. 7(b) shows the chat response unit 115 repeating the alarm ID input by the customer, and the customer inputting into the chat window 710 that the repeated content is correct.

In this case, the customer input information acquisition unit 602 of the chat response unit 115 extracts, for example, "It's alarm αα" as customer input information and notifies the public information analysis unit 611 and the private information analysis unit 621. As a result, the similarity search unit 612 searches for public information (filtered public information) associated with analysis results similar to the analysis results output by inputting customer input information including "It's alarm αα" into the public information analysis unit 611. The similarity search unit 612 also notifies the information window control unit 613 of the search results.

Figure 8(a) shows how the information window control unit 613 displays public information in the information window 720. The example in Figure 8(a) shows how the similarity search unit 612 searches for the top N pieces of public information that satisfy a specified condition (have a similarity level equal to or greater than a specified threshold) from the public information classified into multiple groups. As shown in the information window 720 in Figure 8(a), the search results for each group are displayed in separate tabs.

This is because the information that customers require regarding trouble events varies widely, and the level of information that customers require varies greatly depending on the customer's level of experience with trouble events. For this reason, the information window control unit 613 classifies each piece of information into multiple groups according to the level of information that the customer requires, and displays the search results for each group in separate tabs. This allows the customer to immediately access the requested information simply by selecting the tab that corresponds to the level of information required.

The example in FIG. 8(a) shows that three pieces of public information were searched from the public information classified in the first group (type 1), and one piece of public information was searched from the public information classified in the second group (type 2). The example in FIG. 8(a) also shows that zero pieces of public information were searched from the public information classified in the third group (type 3).

The example in Figure 8(a) shows that the three pieces of public information searched from the public information classified into the first group (type 1) were Delivered Device Information 13, Delivered Device Information 11, and Delivered Device Information 18, and that the similarities were 45%, 20%, and 15%, respectively.

Figure 8 (b) shows how the delivered equipment information 13 has been selected by the customer from the three pieces of public information classified into the first group (type 1) displayed in the information window 720. As shown in Figure 8 (b), by selecting the delivered equipment information 13, the customer can view the details of the delivered equipment information 13 and resolve the trouble event based on the details of the delivered equipment information 13 that they have viewed.

In this way, by displaying multiple pieces of information that are associated with analysis results that have a certain degree of similarity or higher, the customer can quickly resolve the problem. This is because it is not easy to uniquely identify the information needed to resolve the problem through dialogue using a chatbot alone (providing multiple pieces of information and allowing the customer to select is more effective in resolving the problem).

Figure 9 (a) shows the situation in which the customer is asked whether the trouble has been resolved based on the details of the public information (delivered device information 13) displayed in the information window 720, and the customer inputs a response indicating that the trouble has been resolved.

When the chat response unit 115 recognizes that the trouble event has been resolved, it transitions to the display screen 700 shown in FIG. 9(b). Specifically, the chat response unit 115 queries the customer in the chat window 710 about useful information that helped resolve the issue. Furthermore, when the customer inputs useful information that helped resolve the issue, the chat response unit 115 records the useful information that helped resolve the issue in association with the customer input information that was input by the customer when the useful information that helped resolve the issue was searched for.

The chat response unit 115 also instructs the public information providing unit 113 to end the display of public information in the information window 720. Furthermore, the chat response unit 115 outputs a message to the chat window 710 indicating that the customer support service is to be terminated, and ends the customer support service provision process.

<Specific examples of display screens provided in the operator response mode>
Next, the display screens in the operator response mode provided to a terminal (e.g., any one of terminals 131 to 151) that has accessed customer support service device 110 will be described with reference to Figures 10 to 13. Of these, Figure 10 is the fourth diagram showing an example of a display screen provided in the automatic response mode, and Figures 11 to 13 are the first to third diagrams showing an example of a display screen provided in the operator response mode.

As described above, the chat response unit 115
If the problem is not resolved by the chat response information generated by the chatbot and the public information searched based on the chat input information entered by the customer, or
When public information that satisfies a predetermined condition (has a similarity degree equal to or greater than a predetermined threshold) cannot be found based on the chat input information entered by the customer,
Then, the mode is switched to the operator response mode, and the information window control section 623 displays non-public information.

Figure 10 shows the display screen in the automatic response mode before switching to the operator response mode. Of these, the display screen 700 in Figure 10 (a) is the same as the display screen 700 in Figure 7 (a). However, in the case of Figure 10 (a), the chat window control unit 601 of the chat response unit 115 asks the customer whether or not there is an alarm ID, and the customer inputs in the chat window 710 that there is no alarm ID.

If there is no alarm ID, the chat window control unit 601 of the chat response unit 115 prompts the customer to input the details of the trouble event, as shown in FIG. 10(b). The example of FIG. 10(b) shows the customer inputting "The wafer is scratched and there is an abnormal noise coming from the loader" as the details of the trouble event. The example of FIG. 10(b) also shows the chat window control unit 601 of the chat response unit 115 repeating the details of the trouble event input by the customer. Furthermore, the example of FIG. 10(b) shows the customer inputting in the chat window 710 that there are no errors in the details repeated by the chat window control unit 601 of the chat response unit 115.

Here, assume that the similarity search unit 612 of the public information providing unit 113 is unable to search for public information that has a similarity equal to or exceeds a predetermined threshold value as public information that corresponds to the content of the trouble event entered by the customer. In this case, the chat window control unit 601 of the chat response unit 115 determines to switch to the operator response mode, and transitions to the display screen of FIG. 11(a).

Figure 11 (a) is an example of a display screen 1100 provided in the operator response mode. As shown in Figure 11 (a), the display screen 1100 has a chat window 710 and an information window (private information) 1120.

The example in FIG. 11(a) shows how the chat window control unit 601 of the chat response unit 115 outputs a message to the customer in the chat window 710 to indicate that the mode will be switched to the operator response mode. Note that the chat window control unit 601 of the chat response unit 115 may be configured to inquire of the customer whether or not it is OK to switch to the operator response mode before outputting the message to indicate that the mode will be switched to the operator response mode. Also, when the customer inputs that it is OK to switch to the operator response mode, the chat window control unit 601 of the chat response unit 115 may output a message to the effect that the mode will be switched to the operator response mode and switch to the operator response mode.

When switched to the operator response mode, chat response information input by the operator via the CRM terminal 120 is displayed in the chat window 710. In addition, private information (filtered private information) searched based on customer input information extracted from chat input information input by the customer and selected by the operator is displayed.

Figure 11 (b) shows how non-public information retrieved based on customer-entered information, selected by the operator and partially edited if it contains information that should be hidden, is displayed in the information window (non-public information) 1120.

The example in FIG. 11(b) shows how the similarity search unit 622 searches for the top N pieces of private information that satisfy a specified condition (have a similarity level equal to or greater than a specified threshold) from the private information in each of multiple groups. As shown in the information window 1120 in FIG. 11(b), the search results for each group are displayed in separate tabs.

This is because the information that customers require regarding trouble events varies widely, and the level of information that customers require varies greatly depending on the customer's level of experience with trouble events. For this reason, the information window control unit 623 classifies each piece of information into multiple groups according to the level of information that the customer requires, and displays at least a portion of the search results for each group in separate tabs. This allows the customer to immediately access the requested information simply by selecting the tab that corresponds to the level of information required.

The example in FIG. 11(b) shows that three pieces of private information were retrieved from the private information classified into the first group (type 5), and three pieces of private information were retrieved from the private information classified into the second group (type 6). The example in FIG. 11(b) also shows that two pieces of private information were retrieved from the private information classified into the third group (type 7).

The example in Figure 11(b) shows that the three pieces of non-public information searched from the non-public information classified in the first group (type 5) were specification 52, specification 51, and specification 57, and that the similarities were 62%, 23%, and 13%, respectively.

The example in Figure 11(b) shows a standard message ("Displaying ... from the customer. *This message is being displayed ...") being displayed in the chat window 710 as chat input from the operator.

Figure 12(a) shows how specifications 52 has been selected by the customer from the three pieces of non-public information classified into the first group (type 5) displayed in the information window 1120. As shown in Figure 12(a), by selecting specifications 52, the customer can view the details of specifications 52 and solve the trouble based on the details of specifications 52 that he or she has viewed.

In this way, by displaying multiple pieces of information that are associated with analysis results that have a certain degree of similarity or higher, the customer can quickly resolve the problem. This is because it is not easy to uniquely identify the information needed to resolve the problem through dialogue using a chatbot alone (providing multiple pieces of information and allowing the customer to select is more effective in resolving the problem).

Figure 12 (b) shows the situation in which the customer is asked whether the trouble has been resolved based on the details of the non-public information (specification 52) displayed in the information window 1120, and the customer inputs a response indicating that the trouble has been resolved.

When the operator recognizes that the trouble event has been resolved, the display screen transitions to the display screen 1100 shown in FIG. 13. Specifically, the operator uses the chat window 710 to inquire of the customer about useful information that helped resolve the issue. In addition, when the customer inputs useful information that helped resolve the issue, the operator records the useful information that helped resolve the issue in association with the customer input information that was input by the customer when the useful information that helped resolve the issue was searched for.

The operator also instructs the private information providing unit 114 to end the display of private information in the information window 1120. Furthermore, the operator outputs a message to the chat window 710 indicating that the customer support service is to be terminated, and ends the customer support service provision process.

<CRM device management screen>
Next, the management screen of the CRM terminal 120 displayed during the automatic response mode will be described. Fig. 14 is a diagram showing an example of a display screen of the CRM terminal displayed during the automatic response mode. In the automatic response mode, a management screen 1400 showing a list of the usage status of customer support services is displayed on the CRM terminal 120.

As shown in FIG. 14, the management screen 1400 includes the following information items: "Start time," "Elapsed time," "Device type," "Customer," "Person in charge," "Last conversation," and "Operation."

"Start time" records the date and time when the customer started using the customer support service. "Elapsed time" records the time that has elapsed since the customer started using the customer support service. "Device type" records the type of device on which the trouble occurred.

"Customer" records the name of the customer who uses the customer support service. "Person in Charge" records the name of the operator who last responded using the operator response mode.

"Last Conversation" records the last conversation among the series of conversations entered in the chat window 710.

In "Operation", the operation contents to be performed by the operator when switching from the automatic response mode to the operator response mode are displayed in a selectable manner. The example of the management screen 1400 shows that the operation contents include "Participate".

When switching from automatic response mode to operator response mode, the operator operates "Join" and the CRM terminal 120 switches from the management screen 1400 to the display screen 1100 (Figure 11 (a)). As a result, the operator can respond to chat messages.

<Customer support service provision process flow>
Next, a description will be given of the flow of a customer support service providing process in the customer support service providing system 100. Fig. 15 is a flowchart showing the flow of a customer support service providing process in the customer support service providing system. When an access to the chat response unit 115 is accepted from a customer, the customer support service providing process shown in Fig. 14 is started.

In step S1501, the customer support service device 110 executes an automatic response process. The detailed flow of the automatic response process will be described later with reference to FIG. 16.

In step S1502, the customer support service device 110 determines whether public information has been searched for and whether the customer's trouble has been resolved by the automatic response process. If it is determined in step S1502 that public information has not been searched for (NO in step S1502), the process proceeds to step S1503. If it is determined that public information has been searched for but the trouble has not been resolved (NO in step S1502), the process proceeds to step S1503.

In step S1503, the customer support service device 110 and the CRM terminal 120 execute an operator response process, and after executing the operator response process, the customer support service provision process ends. Note that the detailed flow of the operator response process will be described later with reference to FIG. 17.

On the other hand, if public information is searched for and it is determined in step S1502 that the trouble event has been resolved (YES in step S1502), the customer support service provision process is terminated.

<Detailed flow of automatic response processing>
Next, the detailed flow of the automatic response process (step S1501) will be described with reference to Fig. 16, which is a flow chart showing the flow of the automatic response process.

In step S1601, the chat window control unit 601 of the chat response unit 115 accepts chat input information from the customer.

In step S1602, the chat window control unit 601 of the chat response unit 115 generates and outputs chat response information using a chatbot in response to chat input information from the customer.

In step S1603, the chat window control unit 601 of the chat response unit 115 extracts customer input information from the chat input information from the customer and notifies the customer input information acquisition unit 602.

In step S1604, the similarity search unit 612 of the public information providing unit 113 filters the public information based on the customer input information.

In step S1605, the similarity search unit 612 of the public information providing unit 113 determines whether or not the target public information exists as a result of filtering. If it is determined in step S1605 that the target public information does not exist (NO in step S1605), the process proceeds to step S1610.

On the other hand, if it is determined in step S1605 that the relevant public information exists (YES in step S1605), the process proceeds to step S1606.

In step S1606, the public information analysis unit 611 of the public information provision unit 113 receives the customer input information and outputs the analysis result. In addition, the similarity search unit 612 of the public information provision unit 113 calculates the similarity between the output analysis result and the analysis result associated with the public information filtered in step S1604.

In step S1607, the similarity search unit 612 of the public information providing unit 113 determines whether public information exists that is associated with an analysis result having a similarity equal to or greater than a predetermined threshold. If it is determined in step S1607 that public information exists that is associated with an analysis result having a similarity equal to or greater than a predetermined threshold (YES in step S1607), the process proceeds to step S1608.

In step S1608, the information window control unit 613 of the public information providing unit 113 acquires the top N pieces of public information among the pieces of public information associated with analysis results having a similarity equal to or greater than a predetermined threshold.

In step S1609, the information window control unit 613 of the public information providing unit 113 outputs the acquired public information to the information window 720, and the process returns to step S1502 in FIG. 15.

On the other hand, if it is determined in step S1607 that there is no public information associated with an analysis result having a similarity equal to or greater than the predetermined threshold (NO in step S1607), the process proceeds to step S1610.

In step S1610, the public information providing unit 113 notifies the chat response unit 115 that there are no search results, and the process returns to step S1502 in FIG. 15.

<Detailed flow of operator response processing>
Next, the detailed flow of the operator response process (step S1503) will be described with reference to Fig. 17. The flow of the operator response process is shown in Fig. 17.

In step S1701, the chat window control unit 601 of the chat response unit 115 accepts chat input information from the customer.

In step S1702, the chat input unit 632 of the CRM terminal 120 outputs chat response information entered by the operator in response to the chat input information from the customer.

In step S1703, the chat window control unit 601 of the chat response unit 115 extracts customer input information from the chat input information from the customer and notifies the customer input information acquisition unit 602.

In step S1704, the similarity search unit 622 of the non-public information provision unit 114 filters the non-public information based on the customer input information.

In step S1705, the similarity search unit 622 of the non-public information providing unit 114 determines whether or not the target non-public information exists as a result of filtering. If it is determined in step S1705 that the target non-public information does not exist (NO in step S1705), the process proceeds to step S1709.

On the other hand, if it is determined in step S1705 that the relevant non-public information exists (YES in step S1705), the process proceeds to step S1706.

In step S1706, the private information analysis unit 621 of the private information provision unit 114 receives the customer input information and outputs the analysis result. In addition, the similarity search unit 622 of the private information provision unit 114 calculates the similarity between the output analysis result and the analysis result associated with the private information filtered in step S1704.

In step S1707, the similarity search unit 622 of the non-public information providing unit 114 determines whether non-public information exists that is associated with an analysis result having a similarity equal to or greater than a predetermined threshold. If it is determined in step S1707 that non-public information exists that is associated with an analysis result having a similarity equal to or greater than a predetermined threshold (YES in step S1707), the process proceeds to step S1708.

In step S1708, the information window control unit 623 of the non-public information providing unit 114 acquires the top N pieces of non-public information from among the pieces of non-public information associated with analysis results having a similarity equal to or greater than a predetermined threshold.

On the other hand, if it is determined in step S1707 that there is no non-public information associated with an analysis result having a similarity equal to or greater than the predetermined threshold (NO in step S1707), the process proceeds to step S1709.

In step S1709, the private information provision unit 114 notifies the CRM terminal 120 that there are no search results.

In step S1710, the chat input unit 632 of the CRM terminal 120 accepts the chat response information entered by the operator based on the search results and outputs it to the customer. In addition, the private information management unit 633 of the CRM terminal 120 outputs the private information selected by the operator based on the search results to the customer.

<Summary>
As is clear from the above description, the customer support service device 110 according to the first embodiment has the following features:
A chat response unit is provided that switches between an automatic response mode in which a response is automatically made to chat input information about semiconductor manufacturing equipment or the like that is input into the chat window, and an operator response mode in which an operator responds.
The system has a public information providing section that searches for public information based on chat input information entered in the chat window and displays the search results in the information window during the automatic response mode.
The system has a private information providing unit 114 that searches for private information based on chat input information entered in the chat window and displays at least a part of the search results in the information window during the operator response mode.

In this way, by providing an information window in addition to a chat window and configuring it to display public or private information, it is possible to increase the amount of information that can be provided to customers when resolving trouble events, compared to when there is only a chat window.

In addition, by displaying private information in addition to public information, it is possible to provide sufficient information to resolve trouble events, and by limiting the display of private information to operator response mode, it is possible to appropriately manage the information that can be provided to customers.

As a result, the customer support service device according to the first embodiment can improve the quality of customer support services.

Second Embodiment
In the above first embodiment, the chat window control unit 601 of the chat response unit 115 has been described as inquiring about the presence or absence of an alarm ID, and inquiring about the trouble event if the customer inputs in the chat window that there is no alarm ID. However, the content of the inquiry that the chat window control unit 601 of the chat response unit 115 makes to the customer and the order of the inquiries are not limited to this. For example, the chat window control unit 601 may inquire about the type of device in which the trouble event has occurred. Also, a configuration may be made in which a predetermined inquiry is made regardless of the content that the customer inputs in the chat window in response to the inquiry.

In addition, when a predetermined inquiry is made, the system may be configured to make the predetermined inquiry again if the trouble event is not resolved. Alternatively, the system may be configured to make only specific inquiries (e.g., inquiries about the trouble event).

In the above first embodiment, the chat response unit 115 was described as recording the useful information that helped to resolve the issue in association with the customer input information that was input by the customer when the useful information that helped to resolve the issue was searched for.

However, the information recorded by the chat response unit 115 may not only be recorded, but may also be used for additional learning processing of the trained analysis model 420 of the public information analysis unit 111. Specifically, the public information analysis unit 111
The output when the useful information that was useful for the solution is input to the trained analysis model 420,
The output when the chat input information entered by the customer when useful information that was useful for the solution was searched for is input to the trained analysis model 420;
In order to make the trained analytical model 420 match the trained analytical model 420, an additional learning process may be performed on the trained analytical model 420. Note that the additional learning process on the trained analytical model 420 may be performed periodically.

Similarly, the information recorded by the chat response unit 115 may not only be recorded, but may also be used for additional learning processing of the trained analysis model 520 of the private information analysis unit 112. Specifically, the private information analysis unit 112:
The output when the useful information that was useful for the solution is input to the trained analysis model 520,
The output when the chat input information entered by the customer when useful information that was useful for the solution was searched for is input to the trained analysis model 520; and
In order to make the trained analysis model 520 match the trained analysis model 520, an additional learning process may be performed on the trained analysis model 520. Note that the additional learning process on the trained analysis model 520 may be performed periodically.

In addition, in the above first embodiment, no particular mention was made about the display order when the information window control unit 613 displays public information in the information window 720. However, the information window control unit 613 may be configured to display public information in the information window 720 in an order according to similarity, for example. Similarly, in the above first embodiment, no particular mention was made about the display order when the information window control unit 623 displays private information in the information window 1120. However, the information window control unit 623 may be configured to display private information in the information window 1120 in an order according to similarity, for example.

In addition, in the above first embodiment, no mention was made of a method for displaying the device log information or the analysis results of the device log information. However, for example, in the operator response mode, the information window control unit 623 may be configured to provide a new tab in the information window (private information) 1120 and display the device log information or the analysis results when the tab is selected by the customer. Alternatively, the operator may select and display the tab when inputting chat response information. In addition, the display of the analysis results is not limited to the operator response mode. For example, in the automatic response mode, when a customer attaches device log information to chat input information, the analysis results of the device log information may be displayed in the information window (public information) 720.

In addition, in the above first embodiment, the public information analysis unit 111 and the private information analysis unit 112 output the analysis results obtained by converting the frequency of occurrence of each word related to a past trouble event into a numerical value, but the method of calculating the analysis results is not limited to this. In addition, in the above first embodiment, the similarity search unit 612 and the similarity search unit 622 compare the frequency of occurrence of each word included in the analysis results to calculate the respective similarities, and calculate the similarities between the public information and the private information by weighting and adding the calculated similarities. However, the method of calculating the similarities between the public information and the private information is not limited to this.

For example, the public information analysis unit 111 and the private information analysis unit 112 may output any vector expression (other than an expression obtained by quantifying the frequency of occurrence of each word) calculated by performing natural language processing on the public information or private information as the analysis result. In this case, the similarity search unit 612 calculates the similarity by, for example, comparing the vector expression calculated by performing natural language processing on the customer input information with the vector expression calculated by performing natural language processing on the public information. Similarly, the similarity search unit 622 calculates the similarity by comparing the vector expression calculated by performing natural language processing on the customer input information with the vector expression calculated by performing natural language processing on the private information.

In the above first embodiment, it has been described that each of the business offices 130, 140, 150 is located in Japan, and the chat response unit 115 responds in Japanese. However, each of the business offices 130, 140, 150 may be located overseas. In this case, the chat response unit 115 is configured to respond in the language of each country. It is also assumed that the public information and private information stored in each storage unit is stored in the language of each country, and that the trained analysis model is also trained in the language of each country.

In addition, in the first embodiment, the chat input information and chat response information are described as being text data, but the chat input information and chat response information may be audio data.

In the above first embodiment, it has been described that in the response phase, the customer input information acquisition unit 602 acquires customer input information each time chat input information is input, and notifies the public information analysis unit 611 and the private information analysis unit 621. However, the timing at which the customer input information acquisition unit 602 notifies the public information analysis unit 611 and the private information analysis unit 621 of the customer input information is not limited to this. For example, it may be configured to notify the public information analysis unit 611 and the private information analysis unit 621 at the timing at which a predetermined amount of customer input information is acquired.

In the first embodiment, the chat response unit 115 was described as accepting inquiries from customers regarding trouble events, but it is not limited to accepting inquiries from customers, and may also accept inquiries from field engineers.

In addition, in the first embodiment, the specific device is a semiconductor manufacturing device and a measurement device, but the specific device is not limited to this and may be any other device.

Note that the present disclosure is not limited to the configurations described in the above embodiments, and may be combined with other elements. These points may be changed without departing from the spirit of the present disclosure, and may be appropriately determined according to the application form.

100: Customer support service providing system 110: Customer support service device 111: Public information analysis unit 112: Private information analysis unit 113: Public information provision unit 114: Private information provision unit 115: Chat response unit 116: Customer information storage unit 117: Chatbot history information storage unit 118: Internal information storage unit 120: CRM terminal 410: Information acquisition unit 420: Learned analysis model 430: Occurrence frequency quantification unit 440: Analysis result output unit 510: Information acquisition unit 520: Learned analysis model 530: Occurrence frequency quantification unit 540: Analysis result output unit 601: Chat window control unit 602: Customer input information acquisition unit 611: Public information analysis unit 612: Similarity search unit 613: Information window control unit 621: Private information analysis section 622: Similarity search section 623: Information window control section 700: Display screen 710: Chat window 720: Information window 1100: Display screen 1120: Information window 1400: Management screen

Claims (13)

a first providing unit that searches for first information based on input information input to the chat window and displays a search result in the information window during the first mode;
a second providing unit that searches for second information based on input information input to the chat window and displays at least a part of a search result in the information window during a second mode;
A server device in which either one or both of the first providing unit and the second providing unit classify search results into a plurality of groups based on the input information and display them in different tabs in the information window. a control unit that switches between the first mode in which an automatic response is made to input information input to the chat window and the second mode in which an operator responds,
The control unit is
2. The server device according to claim 1, wherein the first providing unit switches to the second mode when, as a result of searching for the first information based on input information entered into the chat window during the first mode, information satisfying a predetermined condition is not found. The control unit is
3. The server device according to claim 2, wherein the server device switches to the second mode when the first providing unit determines that the search results displayed in the information window during the first mode by searching for the first information based on input information entered in the chat window during the first mode are invalid. The server device according to claim 1, wherein the first providing unit displays the search results in an order according to the degree of similarity calculated when searching for the first information. The server device according to claim 4, wherein the first providing unit calculates the similarity by comparing a vector representation calculated by performing natural language processing on the input information with a vector representation calculated by performing natural language processing on the first information. The server device according to claim 1, wherein the second providing unit displays the search results in an order according to the degree of similarity calculated when searching for the second information. The server device according to claim 6, wherein the second providing unit calculates the similarity by comparing a vector representation calculated by performing natural language processing on the input information with a vector representation calculated by performing natural language processing on the second information. The server device according to claim 1, wherein the second providing unit displays the analysis results of the analysis of the log information collected by a specific device in the information window. The server device according to claim 1, wherein during the first mode, a chatbot automatically responds to input information about a specific device entered in the chat window. With respect to a model used when the first providing unit searches for information,
search results displayed in the information window during the first mode that are determined to be valid; and
input information input into the chat window when the search results are displayed;
The server device according to claim 1 , wherein the additional learning process is performed using With respect to a model used when the second providing unit searches for information,
search results displayed in the information window during the second mode that are determined to be valid; and
input information input into the chat window when the search results are displayed;
The server device according to claim 1 , wherein the additional learning process is performed using a first providing step of searching for first information based on input information entered into the chat window and displaying the search results in the information window during the first mode;
a second providing step of searching for second information based on input information input to the chat window and displaying at least a portion of the search results in the information window during the second mode, the second providing step being executed by the computer ;
A customer support service providing method, wherein either or both of the first providing step and the second providing step classify search results into a plurality of groups based on the input information and display them in different tabs within the information window. a first providing step of searching for first information based on input information entered into the chat window and displaying the search results in the information window during the first mode;
a second providing step of searching for second information based on input information input to the chat window and displaying at least a part of the search results in the information window during a second mode;
A customer support service providing program in which either or both of the first providing step and the second providing step classify search results into a plurality of groups based on the input information and display them in different tabs in the information window.

Images