JP6994531B2 - Call center equipment, processing methods and programs - Google Patents

Description

The present invention relates to a call center device, a processing method and a program.

At the call center, many operators receive various inquiries from customers and respond to them. For example, Patent Document 1 discloses a call center business system that reduces stress due to a shortage of operators who make inquiries to a call center and improves the efficiency of call center business.

Japanese Unexamined Patent Publication No. 2019-10139

Call centers often record the content of inquiries received from customers and the results of responses made by operators. For example, when an operator who accepts a report regarding a defect of a product receives an inquiry from a customer, the operator inputs the content of the defect declared by the customer on the screen of the call center system during or after the conversation. However, in general, the contents of the declaration received from the customer are often diverse, and the operator is required to have appropriate skills in order to accurately input the contents of the declaration while talking with the customer. In addition, the work of inputting the contents of the declaration each time the correspondence with the customer is completed is a considerable burden on the operator who receives the declaration from many customers in one day. Similar problems may occur not only to operators who accept repairs of product failures, but also to operators who accept inquiries at call centers in general.

Therefore, an object of the present invention is to provide a technique capable of reducing the burden on the operator who accepts a declaration from a customer.

The call center device according to one aspect of the present invention voice-recognizes a conversation between an operator and a customer who receives a declaration regarding a predetermined object from the start of the conversation to a predetermined timing. By inputting a character string and a voice recognition unit that converts it into a character string, the declaration predicted from the conversation made between the start time of the conversation and the predetermined timing among the multiple declaration contents related to the predetermined target. It has a prediction unit that outputs the contents, and a screen control unit that reflects the declaration contents output from the prediction unit on the screen that the operator inputs the declaration contents from the customer regarding the predetermined target.

According to the present invention, it is possible to provide a technique capable of reducing the burden on an operator who accepts a declaration from a customer.

It is a figure which shows the system configuration example of the call center system which concerns on 1st Embodiment. It is a figure which shows the hardware configuration example of a call center apparatus and a call center terminal. It is a figure which shows the functional block composition example of the call center apparatus which concerns on 1st Embodiment. It is a figure which shows an example of the processing procedure performed by the call center apparatus which concerns on 1st Embodiment. It is a figure for demonstrating the process of predicting the contents of a declaration. It is a figure which shows the network structure example of the prediction model. It is a figure which shows an example of the content of the trouble which occurred in the mobile phone. It is a figure which shows the screen display example in a call center terminal. It is a figure which shows the functional block composition example of the call center apparatus which concerns on 2nd Embodiment. It is a figure which shows an example of the solution DB. It is a figure which shows the screen display example in a call center terminal.

Each embodiment of the present invention will be described with reference to the accompanying drawings. In each figure, those with the same reference numerals have the same or similar configurations. In the following description, the call center system will be described mainly on the premise that it is used in a call center that accepts reports of defects related to mobile phones, but the present embodiment is not limited thereto. The call center system according to the present embodiment can be applied to all call centers that accept various declarations regarding various objects such as goods and services.

[First Embodiment]
<System configuration>
FIG. 1 is a diagram showing a system configuration example of the call center system 1 according to the first embodiment. The call center system 1 includes a call center device 10 and a call center terminal 20. The call center system 1 may include a plurality of call center terminals 20. The call center device 10 and the call center terminal 20 are connected to each other via a wireless or wired communication network N, and can communicate with each other.

The call center device 10 is a device that provides various functions necessary for call center operations. For example, a process of distributing a call from a customer to a call center terminal 20 used by an operator, a process of managing the availability of an operator, and each of them. Performs processing such as managing the time when the operator is talking with the customer. The call center device 10 may be configured by one or a plurality of information processing devices, may be configured by using a virtual server operating on a hypervisor, or may be configured by using a cloud server. It may be configured.

The call center terminal 20 is a terminal used by an operator who performs business at the call center. As the call center terminal 20, any terminal having a communication function, such as a smartphone, a tablet terminal, a mobile phone, a personal computer (PC), a notebook PC, and a personal digital assistant (PDA), can be used.

The call center terminal 20 displays various screens used by the operator to deal with customers. For example, on the screen displayed on the call center terminal 20, a customer's telephone number, an input field for inputting the contents of the declaration from the customer, and the like are displayed.

In the first embodiment, when the operator receives a report from the customer regarding a defect that has occurred in the mobile phone being used by the customer, the operator specifically specifies what kind of report the content is through conversation with the customer. In addition, the specified declaration content is input to the screen displayed on the call center terminal 20.

The call center device 10 predicts the content of the declaration received from the customer by analyzing the content of the conversation between the operator and the customer, and automatically inputs the predicted content of the declaration on the screen of the call center terminal 20. .. At this time, the call center device 10 predicts the content of the declaration based on the conversation up to a certain point in time and displays it on the call center terminal 20 without waiting for the end of the conversation between the operator and the customer. Further, the call center device 10 repeatedly performs a process of predicting the content of the declaration from the content of the conversations that have been performed so far at predetermined intervals. That is, on the screen of the call center terminal 20, the declaration content predicted from the content of the conversation up to a certain point in time is displayed, and the declaration content is updated to be more accurate as the conversation progresses.

<Hardware configuration>
FIG. 2 is a diagram showing a hardware configuration example of the call center device 10 and the call center terminal 20. The call center device 10 and the call center terminal 20 include a processor 11 such as a CPU (Central Processing Unit) and a GPU (Graphical processing unit), a storage device 12 such as a memory, an HDD (Hard Disk Drive) and / or an SSD (Solid State Drive), and the like. It has a communication IF (Interface) 13 that performs wired or wireless communication, an input device 14 that accepts input operations, and an output device 15 that outputs information. The input device 14 is, for example, a keyboard, a touch panel, a mouse and / or a microphone. The output device 15 is, for example, a display and / or a speaker.

<Functional block configuration>
FIG. 3 is a diagram showing an example of a functional block configuration of the call center device 10. The call center device 10 includes a storage unit 100, a voice acquisition unit 101, a voice recognition unit 102, a prediction unit 103, a screen control unit 104, and a prediction processing control unit 105. The storage unit 100 can be realized by using the storage device 12 included in the call center device 10. Further, in the voice acquisition unit 101, the voice recognition unit 102, the prediction unit 103, the screen control unit 104, and the prediction processing control unit 105, the processor 11 of the call center device 10 stores a program stored in the storage device 12. It can be realized by executing it. Further, the program can be stored in a storage medium. The storage medium in which the program is stored may be a non-transitory computer readable medium. The non-temporary storage medium is not particularly limited, but may be, for example, a storage medium such as a USB memory or a CD-ROM.

When the storage unit 100 inputs the content of the conversation between the operator and the customer, the storage unit 100 stores a prediction model for predicting the content of the declaration. The prediction model is stored for each type of declaration content to be predicted. Specific examples of the types of declaration contents will be described later. In the example of FIG. 3, n prediction models of prediction model 100M-1 to prediction model 100Mn (n is a positive integer) are shown. For example, the prediction model generates a large amount of teacher data that inputs the content of the conversation and outputs the declared content from the log data of the conversation between the operator and the customer in the past, and uses the teacher data to neuralize. It can be generated by training the network. In the following description, when the prediction model 100M-1 to the prediction model 100Mn are not distinguished, it is described as the prediction model 100M.

The voice acquisition unit 101 performs a process of acquiring voice data of a conversation between a customer and an operator.

The voice recognition unit 102 is the voice of the conversation held between the start time of the conversation and the predetermined timing for the conversation between the customer and the operator who receives the report of the failure or the like related to the mobile phone (predetermined target) from the customer. Performs processing to convert data into a character string by recognizing it by voice. The voice recognition unit 102 includes a voice recognition engine, recognizes the voice data input to the voice acquisition unit 101, and outputs a character string. The voice recognition unit 102 may output the character string divided into morphemes by performing morphological analysis on the converted character string.

Further, the voice recognition unit 102 may repeatedly perform voice recognition for the conversation data of the conversation performed between the start time of the conversation and the predetermined timing every time the predetermined time elapses. The predetermined time is arbitrary, but may be, for example, 30 seconds or 1 minute. Further, the voice recognition unit 102 may convert the conversation performed between the start time of the conversation and the end of the conversation into a character string.

The prediction unit 103 uses the character string output from the voice recognition unit 102 to predict the content of the declaration from the conversation performed between the start time of the conversation and the predetermined timing among the plurality of declaration contents regarding the predetermined target. Is output. More specifically, the prediction unit 103 inputs the character string output from the voice recognition unit 102 into each prediction model 100M stored in the storage unit 100, and the declaration content output from each prediction model 100M is displayed. Output as expected declaration content.

The content of the declaration predicted by the prediction unit 103 includes the content regarding the user of the predetermined target, the content of the defect that occurred in the predetermined target, the content regarding the timing at which the defect occurs in the predetermined target, and the content regarding the frequency of the defect in the predetermined target. At least one of them may be included.

Further, the prediction unit 103 may repeatedly output the contents of the declaration predicted from the conversation performed between the start time of the conversation and the predetermined timing every time the predetermined time elapses. For example, the prediction unit 103 outputs the predicted contents of the declaration based on the conversation made between the start time of the conversation and 30 seconds after the start of the conversation, and then, during the period from the start time of the conversation to 1 minute later. Outputs the predicted content of the declaration based on the conversation that took place, then outputs the predicted content of the declaration based on the conversation that took place between the start of the conversation and 1 minute and 30 seconds later, and then the next. In addition, the process of outputting the predicted contents of the declaration based on the conversation conducted between the start time of the conversation and the second minute may be repeated.

Further, the prediction unit 103 allows the operator to perform a predetermined procedure with the customer based on the character string of the conversation performed from the start time of the conversation converted by the voice recognition unit 102 to the end of the conversation. An index indicating whether or not the measures are taken may be output.

Further, the prediction unit 103 has a high probability that the prediction contents are predicted from the conversation performed between the start time of the conversation and the predetermined timing among the plurality of declaration contents regarding the predetermined target, and the predictions are in agreement. A predetermined number of declaration contents may be output in order.

The screen control unit 104 performs a process of generating various screens to be displayed on the call center terminal 20 and displaying them on the call center terminal 20. Further, the screen control unit 104 performs a process of reflecting the content of the declaration output from the prediction unit 103 on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target.

The screen control unit 104 has, for example, a Web server function, and transmits Web screen data (HTML data) to the call center terminal 20 when it is accessed from a Web browser installed in the call center terminal 20. It may be that. Alternatively, an application for drawing various screens is installed in the call center terminal 20, and the screen control unit 104 generates screen data in a format that can be recognized by the application when a request is received from the application. Then, it may be transmitted to the call center terminal 20.

Further, the screen control unit 104 may update the content of the declaration from the customer regarding the predetermined target, which is reflected on the screen of the call center terminal 20, every time the predetermined time elapses.

Further, the screen control unit 104 may accept from the operator the selection of the declaration contents to be reflected on the screen of the call center terminal 20 from the predetermined number of declaration contents output from the prediction unit 103.

The prediction processing control unit 105 performs various processes necessary for the call center device 10 to perform the prediction processing. For example, the prediction processing control unit 105 processes the voice data of the conversation between the operator and the customer to be input to the voice recognition unit 102, and the character string data acquired from the voice recognition unit 102 is input to the prediction unit 103. Processing, processing to pass the declaration content output from the prediction unit 103 to the screen control unit, and the like are performed.

<Processing procedure>
(Explanation of processing procedure)
FIG. 4 is a diagram showing an example of a processing procedure performed by the call center device 10 according to the first embodiment.

First, the voice acquisition unit 101 acquires the voice data of the call made between the customer and the operator, and the conversation ID uniquely identifies the call made between the customer and the operator with respect to the acquired voice data. Is given.

Subsequently, the voice acquisition unit 101 transmits the voice data and the conversation ID to the voice recognition unit 102 (S10). The voice acquisition unit 101 continues to transmit voice data to the voice recognition unit 102 until the call between the customer and the operator ends. The voice recognition unit 102 stores the voice data received from the voice acquisition unit 101 in association with the conversation ID.

Subsequently, the voice acquisition unit 101 transmits a conversation start message indicating that the conversation has started between the customer and the operator to the screen control unit 104. The conversation start message includes a conversation ID (S11).

Subsequently, the screen control unit 104 determines the timing for executing the prediction processing of the declaration content (S12). Subsequently, the screen control unit 104 transmits the conversation ID to the prediction processing control unit 105 at the timing when it is determined that the prediction processing is to be executed (S13).

For example, the screen control unit 104 may determine the timing at which the operator instructs the execution of the prediction process by operating the screen as the timing for executing the prediction process of the declared content. Further, the screen control unit 104 may determine the timing at predetermined intervals from the time when the conversation starts as the timing to execute the prediction processing of the declaration content. For example, when the predetermined interval is set to 30 seconds, it is determined that the timing such as 30 seconds, 1 minute, 1 minute and a half, and 2 minutes after the conversation is started is the timing to execute the prediction processing of the declared contents. You may try to do it.

Subsequently, when the prediction processing control unit 105 receives the conversation ID from the screen control unit 104, the prediction processing control unit 105 transmits the received conversation ID to the voice recognition unit 102 (S14). The voice recognition unit 102 extracts voice data associated with the conversation ID received from the prediction processing control unit 105 from the stored voice data, and inputs the extracted voice data to the voice recognition engine. Performs voice recognition and conversion processing into a character string (S15). The timing at which the voice recognition unit 102 receives the conversation ID from the prediction processing control unit 105 is 30 seconds, 1 minute, 1 and a half minutes, 2 minutes, etc. from the start of the conversation. Therefore, the voice recognition unit 102 repeatedly performs voice recognition and character string conversion processing at a predetermined timing from the start of the conversation (timing such as 30 seconds, 1 minute, 1 minute and a half, 2 minutes after the start of the conversation).

Subsequently, the voice recognition unit 102 transmits the character string obtained by the conversion to the prediction processing control unit 105 (S16).

The prediction processing control unit 105 transmits the character string received from the voice recognition unit 102 to the prediction unit 103 (S17). The prediction unit 103 inputs the received character string into the prediction models 100M-1 to 100M-n, and acquires the declaration contents output from the prediction models 100M-1 to 100M-n, so that the declaration described by the customer is made. Predict the content (S18). Subsequently, the prediction unit 103 transmits the prediction result to the prediction processing control unit 105.

The prediction processing control unit 105 transmits the conversation ID and the prediction result to the screen control unit 104 (S20). The screen control unit 104 displays the prediction ID and the prediction result on the operator's screen. More specifically, the screen control unit 104 performs a process of reflecting the prediction ID and the prediction result on the screen in which the operator inputs the contents of the declaration from the customer regarding the mobile phone (S21).

When the conversation between the operator and the customer is completed, the voice acquisition unit 101 transmits a message indicating the end of the conversation to the screen control unit 104 (S22). After that, the processing procedure from step S13 to step S21 is repeatedly executed at intervals such as 30 seconds, 1 minute, and 1 and a half minutes after the start of the conversation, so that the prediction result displayed on the screen of the call center terminal 20 is displayed. Is updated.

(Forecast of declaration contents)
FIG. 5 is a diagram for explaining a process of predicting the content of the declaration. As shown in FIG. 5, the prediction unit 103 inputs the character string obtained by voice-converting the conversation into the prediction models 100M-1 to 100M-n, so that the prediction models 100M-1 to 100M-n Get the prediction result from each of. The prediction models 100M-1 to 100M-n are neural networks having different types of prediction contents. The type of the declaration content predicted by the prediction models 100M-1 to 100M-n may be any type, but for example, the content related to the customer, the content of the defect that occurred in the mobile phone, and the timing when the defect occurs in the mobile phone. The content may be related to the frequency of problems with the mobile phone. The contents related to the customer are, for example, information on whether or not the person who is talking to the operator is the contractor of the mobile phone, and where the person who is talking to the operator is calling from. Information indicating (for example, whether the call is made from a mobile phone shop or from another place) or the like may be used.

For example, the prediction model 100M-1 is a model that predicts the contents related to the customer, the prediction model 100M-2 is a prediction model that predicts the contents of a defect that has occurred in the mobile phone, and the prediction model 100M-3 is a mobile phone. It is a prediction model that predicts the content regarding the timing at which a problem occurs in the telephone, and the prediction model 100M-4 may be a prediction model that predicts the content regarding the frequency of failure in the mobile phone.

FIG. 6 is a diagram showing an example of a network structure of prediction models 100M-1 to 100M-n. The network structure shown in FIG. 6 is only an example, and it is possible to adopt a network structure different from that shown in FIG. The input data (character string) input to the prediction models 100M-1 to 100M-n is input to the Embedding layer. In the Embedding layer, the input character string is converted into vector data. At this time, words having similar meanings are converted into similar vector data. Words with similar meanings, such as "battery" and "battery", are converted to similar vector data.

The 1D Convolution layer extracts a characteristic part of the conversation between the operator and the customer by performing the convolution process a plurality of times on the character string represented by the vector data.

The Attention layer selects the part corresponding to the content of the declaration to be predicted from the parts such as greetings, symptom hearings, and troubleshooting (interviews) included in the input conversation between the operator and the customer.

The Dense layer outputs a probability value indicating a prediction result for each declaration content based on the output from the Attention layer.

FIG. 7 is a diagram showing an example of the content of a defect that has occurred in a mobile phone. As shown in FIG. 7, the content of the defect that has occurred in the mobile phone may be expressed in three stages of a large item, a medium item, and a small item. In the example of FIG. 7, a large item indicates a place where a defect has occurred, a medium item indicates the content of the defect, and a small item indicates the content of the defect that occurs in more detail.

The prediction model 100M-2, which predicts the content of the defect that has occurred in the mobile phone described above, has a probability that the prediction results match for each of the combination patterns of the large item, the medium item, and the small item shown in FIG. The probability value indicating is output from the Dense layer. For example, when there are 200 combinations of all patterns, the prediction model 100M-2 outputs a probability value indicating the probability that the prediction results match for all 200 patterns from the Dense layer.

The prediction unit 103 may output the pattern having the highest probability value among the probability values of all the patterns output by the prediction model 100M as a prediction result about the content of the defect that has occurred in the mobile phone. A predetermined number of patterns may be output in order from the pattern having the highest probability value as a prediction result having a high probability that the predictions match, which is the content of a defect that has occurred in the mobile phone.

Further, the content regarding the timing at which the mobile phone malfunction is predicted by the prediction unit 103 is, for example, when the power is turned on (the malfunction occurs when the mobile phone is turned on), or when a specific application is opened, the screen is displayed. When touched, when the volume switch is pressed, etc., but it is not limited to this.

In addition, the content predicted by the prediction unit 103 regarding the frequency of problems with the mobile phone is, for example, once a day, twice a day, once a week, twice a week, or once a week. The frequency is such as 3 times or only once, but the frequency is not limited to this.

(Display example)
FIG. 8 is a diagram showing an example of screen display in the call center terminal 20. The customer's telephone number is displayed in the display area A100. In the display area A200, a character string obtained by recognizing the customer's voice by voice and a character string obtained by recognizing the operator's voice by voice are displayed. In the display area A300, an input area for inputting the contents of the declaration heard from the customer by the operator is displayed. Each input area is a pull-down menu, and the operator can input the declaration contents by selecting the pattern corresponding to the declaration contents heard from the customer from the patterns of the declaration contents displayed in the pull-down menu. Can be done.

As the conversation between the customer and the operator progresses, the predicted content of the declaration based on the conversations that have been made so far is automatically entered in each input area. At this time, the declaration contents predicted by the prediction unit 103 to have the highest probability are displayed in each input area, and when the pull-down menu is pressed, a predetermined number of declaration contents are displayed in descending order of probability. good.

If there is an error in the prediction result of the prediction unit 103, the operator needs to correct the content of the declaration based on his / her own judgment. However, by narrowing down to a predetermined number of declaration contents in descending order of probability and displaying them in the pull-down menu, the operator does not have to search for the correct declaration contents from all the declaration contents patterns, so that the declaration contents can be quickly entered. It will be possible to fix it.

In FIG. 8, the caller information and the interlocutor (caller) correspond to the contents related to the customer. In addition, the content of the declaration (major item), the content of the declaration (medium item), and the content of the declaration (minor item) correspond to the content of the defect that occurred in the mobile phone. The occurrence status corresponds to the content related to the timing when a malfunction occurs in the mobile phone. The frequency of occurrence (frequency) and the frequency of occurrence (number of times) correspond to the contents related to the frequency of problems with mobile phones.

The input area A350 for inputting the interview results is an area where the operator selects the suspected part identified by troubleshooting based on the contents of the customer's declaration and the countermeasure (wait-and-see, parts replacement, replacement of each terminal, etc.). Is.

When the automatic input button B10 is pressed, the screen control unit 104 determines that the operator has instructed the execution of the prediction process as described in step S12 of FIG. 4, and transmits the conversation ID to the prediction process control unit 105. do. Upon receiving the conversation ID, the prediction processing control unit 105 executes the processing procedures from steps S14 to S21 in FIG. As a result, the prediction process of the declaration content operates when the automatic input button B10 is pressed, and the display area A300 shown in FIG. 8 is updated with the prediction result obtained as a result.

(Question for interview)
The prediction unit 103 determines the response method that the operator should perform with the customer based on the character string of the conversation that was performed from the start time of the conversation converted by the voice recognition unit 102 to the end of the conversation. Outputs an index indicating whether or not the response is performed according to the indicated manual (predetermined procedure). The indicators are, for example, "1. Not responding according to the manual", "2. Occasionally responding according to the manual", "3. Sufficiently responding according to the manual". It may be in three stages of. Further, the stage is not limited to three stages, and may be two stages or four or more stages.

The prediction unit 103 may obtain the index by inputting the character string of the conversation performed from the start time of the conversation to the end of the conversation into the trained model that outputs the index. .. Alternatively, the prediction unit 103 may use a specific keyword (for example, "remote diagnosis", "troubleshooting", etc., described in the manual in the character string of the conversation performed from the start of the conversation to the end of the conversation. ) May be determined based on the determination conditions such as whether or not the index exists.

[Second Embodiment]
Subsequently, the call center system 1 according to the second embodiment will be described. In the description of the second embodiment, the points not particularly mentioned are the same as those of the first embodiment.

<Functional block configuration>
FIG. 9 is a diagram showing an example of a functional block configuration of the call center device 10 according to the second embodiment.

The storage unit 100 stores the information indicating one or more solutions for improving the defect and the resolution rate when the solution is executed in association with each other for each pattern of the defect that occurs in the predetermined target. Measure DB100N is stored. The solution rate may be omitted from the solution DB 100N. Further, when the storage unit 100 inputs the content of the conversation between the operator and the customer, the storage unit 100 determines whether or not the operator has proposed to the customer to execute a solution for improving the problem that has occurred in the predetermined target. The prediction model M to be predicted is stored. For example, the prediction model M inputs the content of the conversation from the log data of the conversation between the operator and the customer in the past, and outputs a flag indicating whether or not the solution is proposed to the customer. It can be generated by generating a large number of teacher data to be used and training a neural network using the teacher data.

When the operator performs a predetermined input operation, the prediction unit 103 is based on the character string of the conversation performed between the start time of the conversation between the operator and the customer and the time when the predetermined input operation is performed. Then, it predicts whether or not the operator has proposed to the customer to implement a solution for improving the defect that has occurred in the predetermined target. More specifically, the prediction unit 103 inputs the character string output from the voice recognition unit 102 into the prediction model 100M, and when the probability output from the prediction model 100M is equal to or higher than a predetermined threshold value, the operator determines. It is also possible to output the prediction result that the execution of the solution for improving the defect is proposed.

When the screen control unit 104 predicts in the prediction unit 103 that the operator has not proposed to the customer the execution of a solution for improving the defect that has occurred in the predetermined target, the screen control unit 104 improves the defect that has occurred in the predetermined target. Display information on the screen that prompts the customer to propose the solution to the solution.

Further, the screen control unit 104 may acquire one or more solutions corresponding to the contents of the declaration regarding the defect that occurred in the predetermined target output from the prediction unit 103 from the solution DB 100N and display them on the screen. good. Further, the screen control unit 104 acquires from the solution DB 100N one or more solutions corresponding to the contents of the declaration regarding the defect occurring in the predetermined target output from the prediction unit 103, and the solution rate corresponding to the solution. It may be displayed on the screen.

The input unit 106 receives from the operator the input of information indicating whether or not the problem that has occurred in the predetermined target has been solved by the customer executing one or more solutions displayed on the screen.

The update processing unit 107 updates the resolution rate stored in the solution DB 100N based on the information received by the input unit 106 indicating whether or not the problem that has occurred in the predetermined target has been resolved.

FIG. 10 is a diagram showing an example of the solution DB 100N. In the "major item", "medium item", and "minor item", the content of the defect that occurs in the predetermined target is stored. The "major item", "medium item" and "minor item" in FIG. 10 correspond to the large item, the medium item and the minor item shown in FIG. 7. "Solution 1" and "Solution rate / Time required", "Solution 2" and "Solution rate / Time required", and "Solution 3" and "Solution rate / Time required" are "Major items" and "Medium". Stores one or more solutions for improving the problems represented by the "item" and "sub-item" items, the solution rate when the solution is executed, and the estimated time required to execute the solution. Will be done.

In the solution DB 100N shown in FIG. 10, three solutions 1 to 3 are shown, but four or more solutions may be stored, or two or less solutions may be stored. good.

<Processing procedure>
Subsequently, the processing procedure performed by the call center device 10 according to the second embodiment will be described with reference to FIG. In the second embodiment, the screen control unit 104 obtains one or more solutions, a solution rate, and a required time corresponding to the prediction result sent from the prediction processing control unit 105 in the processing procedure of step S20 from the solution DB 100N. Acquired, and the acquired one or more solutions, the resolution rate, and the required time are displayed on the screen (S21).

(Display example)
FIG. 11 is a diagram showing an example of screen display in the call center terminal 20. Of the screen display examples shown in FIG. 11, the parts where the same display as that of FIG. 7 is performed are designated by the same reference numerals as those of FIG. 7, and the description thereof will be omitted.

The screen displayed on the call center terminal 20 includes a display area A500 that displays one or more solutions corresponding to the contents of the declaration regarding the defect (large item, medium item, and small item). In the display area A500, a display area A501 displaying the solution 1, a display area A502 displaying the solution 2, and a display area A503 displaying the solution 3 are displayed side by side in a predetermined order. The predetermined order may be, for example, the order in which the resolution rate is high, or the order in which the time required when the solution is executed is short. In the example of FIG. 11, the solutions are displayed side by side in descending order of the resolution rate.

If the prediction result by the prediction unit 103 is changed according to the progress of the conversation between the operator and the customer and the declaration contents displayed in the display areas A320 to A322 are updated, the display areas A501 to A503 are displayed. The displayed solutions 1 to 3 are also updated. Further, if there is an error in the prediction result of the prediction unit 103 and the operator corrects the declaration content based on his / her own judgment, the solutions 1 to 3 displayed in the display areas A501 to A503 are also the declaration content corrected by the operator. Will be changed to the corresponding solution.

The display area A501 has a display area A501-1 that displays a character string indicating a solution, a resolution rate A501-2 that indicates the probability that the problem will be solved when the solution is executed, and a case where the solution is executed. The required time A501-3 is included. Further, in the display area A501, an input area A501-in which information indicating whether or not the problem has been solved as a result of the customer executing the solution (for example, either "solved" or "not resolved") is input. 4 is included. The information input to the input area A501-4 by the operator is accepted by the input unit 106. Further, the update processing unit 107 updates the solution DB 100N based on the input information.

For example, when "solved" or "not resolved" is input in the input area A501-4, the update processing unit 107 in the solution DB 100N, the large item, the medium item, and the small item are "power supply" and "power supply", respectively. Update the resolution rate corresponding to the restart solution in the records that are "power down" and "power off suddenly during use". If "solved" is entered, the resolution rate will increase, and if "not resolved" is entered, the resolution rate will decrease. The resolution rate (%) is calculated by the formula "(the number of times" solved "was entered) ÷ (the number of times" solved "was entered + the number of times" not solved "was entered) x 100". can do.

Further, the input unit 106 may accept the input of the time required for the customer to execute the solution. For example, in the input unit 106, the customer solves the time from the timing when the operator clicks or touches the display area A501-1 until the information indicating whether or not the problem has been solved is input to the input area A501-4. It may be accepted as the time required to execute the measure. The update processing unit 107 may update the required time stored in the solution DB 100N by using the time required by the customer to execute the solution received by the input unit 106. The calculation of the required time for updating may be performed, and for example, it may be the average value of the required time in the same solution received by the input unit 106 in the past.

At this time, if the time required for the customer to execute the solution received by the input unit 106 is shorter than a predetermined period (for example, 15 seconds, 30 seconds, etc.), the update processing unit 107 operates the operator for the time. It may be regarded as an abnormal value due to an error or the like, and may not be used for updating the required time stored in the solution DB 100N.

Further, in the display area A300, there is an input button B20 for inputting that the customer has been proposed to execute a solution for improving the defect. When the input button B20 is pressed, the voice recognition unit 102 outputs a character string of the conversation that took place between the start of the conversation and the time when the input button B20 is pressed. Further, the prediction unit 103 uses the character string output from the voice recognition unit 102 to predict whether or not the operator has proposed to the customer to execute a solution for improving the problem that has occurred in the predetermined target. .. If it is predicted that the customer is not proposed to execute the solution, the screen control unit 104 prompts the operator to propose the execution of the solution to the customer (for example, "solution". Please suggest to the customer ”, such as a character string or pop-up) is displayed on the screen. As a result, even though there is a possibility that the problem can be solved by having the customer execute the solution, hardware replacement etc. will be wasted without having the customer execute the solution. Can be suppressed.

<Summary>
According to each of the above-described embodiments, the call center device 10 automatically predicts the content of the declaration received from the customer and automatically inputs the content of the declaration based on the content of the conversation between the customer and the operator. input. As a result, the operator does not have to specify the content of the declaration by himself / herself and input it on the screen, so that it is possible to reduce the burden on the operator who receives the declaration from the customer. In addition, since the operator can focus on hearing the contents of the declaration and the malfunction event from the customer, the contents of the declaration can be specified more accurately.

Further, the call center device 10 outputs a predetermined number of declared contents in descending order of the predicted contents of the declaration and the probability that the predictions match. As a result, the operator only has to select the correct declaration content from the predetermined declaration contents, so that the operator's burden is increased as compared with the case of selecting the correct declaration content from all the declaration contents patterns. It will be reduced.

Further, the call center device 10 outputs an index indicating whether or not the operator is dealing with the customer according to the manual. In call center operations, manuals stipulate hearing procedures that should be performed in order to objectively and accurately grasp the details of defects and the location of failures. According to each embodiment, it becomes possible for the operator to grasp that various procedures are performed according to the manual.

Further, the call center device 10 checks whether or not the operator has urged the customer to execute a solution for improving the problem that has occurred in the predetermined target, and if it is determined that the operator has not urged the customer, alarm information. Was output. This makes it possible to encourage the operator to have the customer implement the solution.

Further, the call center device 10 displays the resolution rate and the required time in addition to the solution corresponding to the content of the declaration. This allows the operator to select and have the customer execute a solution with a high resolution rate and a short time required.

Each of the embodiments described above is for facilitating the understanding of the present invention, and is not for limiting the interpretation of the present invention. The flowcharts, sequences, elements included in the embodiments, arrangements, materials, conditions, shapes, sizes, and the like described in the embodiments are not limited to those exemplified, and can be appropriately changed. Further, it is possible to partially replace or combine the configurations shown in different embodiments.

1 ... Call center system, 10 ... Call center device, 11 ... Processor, 12 ... Storage device, 13 ... Communication IF, 14 ... Input device, 15 ... Output device, 20 ... Call center terminal, 100 ... Storage unit, 100M ... Prediction model, 100N ... Solution DB, 101 ... Voice acquisition unit, 102 ... Voice recognition unit, 103 ... Prediction unit, 104 ... Screen control unit, 105 ... Prediction processing control unit, 106 ... Input unit, 107 ... Update processing unit

Claims (16)

A voice recognition unit that converts a conversation between an operator who receives a declaration regarding a predetermined target from a customer and the customer into a character string by recognizing the conversation between the start of the conversation and the predetermined timing. When,
By inputting the character string, a prediction unit that outputs the declaration content predicted from the conversation performed between the start time of the conversation and the predetermined timing among the plurality of declaration contents related to the predetermined target, and the prediction unit.
A screen control unit that reflects the content of the declaration output from the prediction unit on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target.
Have,
The voice recognition unit repeatedly performs voice recognition every time a predetermined time elapses for a conversation held between the start time of the conversation and the predetermined timing.
The prediction unit repeatedly outputs the content of the declaration predicted from the conversation performed between the start time of the conversation and the predetermined timing every time the predetermined time elapses.
The screen control unit updates the content of the declaration from the customer regarding the predetermined target, which is reflected on the screen, every time the predetermined time elapses.
Call center device. A voice recognition unit that converts a conversation between an operator who receives a declaration regarding a predetermined target from a customer and the customer into a character string by recognizing the conversation between the start of the conversation and the predetermined timing. When,
By inputting the character string, a prediction unit that outputs the declaration content predicted from the conversation performed between the start time of the conversation and the predetermined timing among the plurality of declaration contents related to the predetermined target, and the prediction unit.
A screen control unit that reflects the content of the declaration output from the prediction unit on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target.
Have,
Among the plurality of declaration contents related to the predetermined object, the prediction unit has a probability that the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing and the predictions match. Output the specified number of declaration contents in descending order,
The screen control unit receives from the operator the selection of the declaration contents to be reflected on the screen from the predetermined number of declaration contents output from the prediction unit.
Call center device. A voice recognition unit that converts a conversation between an operator who receives a declaration regarding a predetermined target from a customer and the customer into a character string by recognizing the conversation between the start of the conversation and the predetermined timing. When,
By inputting the character string, a prediction unit that outputs the declaration content predicted from the conversation performed between the start time of the conversation and the predetermined timing among the plurality of declaration contents related to the predetermined target, and the prediction unit.
A screen control unit that reflects the content of the declaration output from the prediction unit on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target.
Have,
The voice recognition unit converts the conversation that took place between the start of the conversation and the end of the conversation into a character string.
In the prediction unit, the operator determines with the customer based on the character string of the conversation performed from the start time of the conversation to the end of the conversation converted by the voice recognition unit. Outputs an index indicating whether or not the response is in line with the procedure.
Call center device. A voice recognition unit that converts a conversation between an operator who receives a declaration regarding a predetermined target from a customer and the customer into a character string by recognizing the conversation between the start of the conversation and the predetermined timing. When,
By inputting the character string, a prediction unit that outputs the declaration content predicted from the conversation performed between the start time of the conversation and the predetermined timing among the plurality of declaration contents related to the predetermined target, and the prediction unit.
A screen control unit that reflects the content of the declaration output from the prediction unit on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target.
Have,
When the operator performs a predetermined input operation, the prediction unit determines that the operator may perform a predetermined input operation based on a character string of the conversation performed between the start time of the conversation and the time when the predetermined input operation is performed. Predict whether or not the customer has been proposed to implement a solution for improving the problem that has occurred in the predetermined target.
When the predictor predicts that the operator has not proposed to the customer the execution of a solution for improving the defect caused in the predetermined target, the screen control unit occurs in the predetermined target. Display information on the screen prompting the customer to propose the execution of a solution for improving the problem.
Call center device. The content of the declaration includes at least one of the content relating to the customer, the content of the defect occurring in the predetermined target, the content relating to the timing at which the defect occurs in the predetermined target, and the content relating to the frequency of the defect occurring in the predetermined target. Includes one
The call center device according to any one of claims 1 to 4. Further, it has a storage unit for storing information indicating one or more solutions for improving the defect for each defect pattern that occurs in the predetermined object.
The screen control unit acquires from the storage unit one or more solutions corresponding to the contents of the declaration regarding the defect that has occurred in the predetermined object output from the prediction unit and displays it on the screen.
The call center device according to any one of claims 1 to 5 . In the storage unit, information indicating one or more solutions for improving a defect and a resolution rate when the solution is executed are stored in association with each other for each defect pattern occurring in a predetermined target. ,
The screen control unit acquires from the storage unit one or more solutions corresponding to the contents of the declaration regarding the defect occurring in the predetermined object output from the prediction unit and the solution rate corresponding to the solution. And display it on the screen.
The call center device according to claim 6 . An input unit that accepts input from the operator of information indicating whether or not the problem that has occurred in the predetermined target has been solved by the customer executing the one or more solutions displayed on the screen.
Further includes an update processing unit that updates the resolution rate stored in the storage unit based on the information received by the input unit that indicates whether or not the problem that has occurred in the predetermined object has been resolved.
The call center device according to claim 7 . It is a processing method performed by the call center device.
A step of converting a conversation between an operator who receives a declaration regarding a predetermined object from a customer and the customer into a character string by voice-recognizing the conversation conducted between the start time of the conversation and the predetermined timing.
By inputting the character string, among a plurality of declaration contents related to the predetermined target, a step of outputting the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing, and
A step of reflecting the output content of the declaration on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target, and
Including
In the conversion step, the conversation performed between the start time of the conversation and the predetermined timing is repeatedly voice-recognized every time a predetermined time elapses.
In the output step, the content of the declaration predicted from the conversation performed between the start time of the conversation and the predetermined timing is repeatedly output every time the predetermined time elapses.
The step to be reflected on the screen is to update the content of the declaration from the customer regarding the predetermined target to be reflected on the screen every time the predetermined time elapses.
Processing method. It is a processing method performed by the call center device.
A step of converting a conversation between an operator who receives a declaration regarding a predetermined object from a customer and the customer into a character string by voice-recognizing the conversation conducted between the start time of the conversation and the predetermined timing.
By inputting the character string, among a plurality of declaration contents related to the predetermined target, a step of outputting the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing, and
A step of reflecting the output content of the declaration on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target, and
Including
The output step is the probability that the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing among the plurality of declaration contents related to the predetermined object, and the predictions match. Outputs the specified number of declaration contents in descending order of
The step to be reflected on the screen accepts from the operator the selection of the declaration contents to be reflected on the screen from the predetermined number of declaration contents output from the output step.
Processing method. It is a processing method performed by the call center device.
A step of converting a conversation between an operator who receives a declaration regarding a predetermined object from a customer and the customer into a character string by voice-recognizing the conversation conducted between the start time of the conversation and the predetermined timing.
By inputting the character string, among a plurality of declaration contents related to the predetermined target, a step of outputting the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing, and
A step of reflecting the output content of the declaration on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target, and
Including
The conversion step converts a conversation performed between the start of the conversation and the end of the conversation into a character string.
The output step is based on the character string of the conversation performed from the start time of the conversation converted in the conversion step to the end of the conversation, and the operator communicates with the customer. Outputs an index indicating whether or not the response is in accordance with the prescribed procedure.
Processing method. It is a processing method performed by the call center device.
A step of converting a conversation between an operator who receives a declaration regarding a predetermined object from a customer and the customer into a character string by voice-recognizing the conversation conducted between the start time of the conversation and the predetermined timing.
By inputting the character string, among a plurality of declaration contents related to the predetermined target, a step of outputting the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing, and
A step of reflecting the output content of the declaration on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target, and
Including
When the operator performs a predetermined input operation, the output step is based on the character string of the conversation performed between the start time of the conversation and the time when the predetermined input operation is performed by the operator. , Predict whether or not the customer has been proposed to implement a solution for improving the problem that has occurred in the predetermined target.
The step to be reflected on the screen is the predetermined step when it is predicted that the operator has not proposed to the customer the execution of the solution for improving the defect caused in the predetermined target in the output step. Displaying information on the screen prompting the customer to propose the execution of a solution for improving the problem that has occurred in the target.
Processing method. On the computer
A step of converting a conversation between an operator who receives a declaration regarding a predetermined object from a customer and the customer into a character string by voice-recognizing the conversation conducted between the start time of the conversation and the predetermined timing.
By inputting the character string, among a plurality of declaration contents related to the predetermined target, a step of outputting the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing, and
A step of reflecting the output content of the declaration on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target, and
To execute ,
In the conversion step, the conversation performed between the start time of the conversation and the predetermined timing is repeatedly voice-recognized every time a predetermined time elapses.
In the output step, the content of the declaration predicted from the conversation performed between the start time of the conversation and the predetermined timing is repeatedly output every time the predetermined time elapses.
The step to be reflected on the screen is to update the content of the declaration from the customer regarding the predetermined target to be reflected on the screen every time the predetermined time elapses.
program. On the computer
A step of converting a conversation between an operator who receives a declaration regarding a predetermined object from a customer and the customer into a character string by voice-recognizing the conversation conducted between the start time of the conversation and the predetermined timing.
By inputting the character string, among a plurality of declaration contents related to the predetermined target, a step of outputting the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing, and
A step of reflecting the output content of the declaration on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target, and
To execute,
The output step is the probability that the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing among the plurality of declaration contents related to the predetermined object, and the predictions match. Outputs the specified number of declaration contents in descending order of
The step to be reflected on the screen accepts from the operator the selection of the declaration contents to be reflected on the screen from the predetermined number of declaration contents output from the output step.
program. On the computer
A step of converting a conversation between an operator who receives a declaration regarding a predetermined object from a customer and the customer into a character string by voice-recognizing the conversation conducted between the start time of the conversation and the predetermined timing.
By inputting the character string, among a plurality of declaration contents related to the predetermined target, a step of outputting the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing, and
A step of reflecting the output content of the declaration on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target, and
To execute,
The conversion step converts a conversation performed between the start of the conversation and the end of the conversation into a character string.
The output step is based on the character string of the conversation performed from the start time of the conversation converted in the conversion step to the end of the conversation, and the operator communicates with the customer. Outputs an index indicating whether or not the response is in accordance with the prescribed procedure.
program. On the computer
A step of converting a conversation between an operator who receives a declaration regarding a predetermined object from a customer and the customer into a character string by voice-recognizing the conversation conducted between the start time of the conversation and the predetermined timing.
By inputting the character string, among a plurality of declaration contents related to the predetermined target, a step of outputting the declaration contents predicted from the conversation performed between the start time of the conversation and the predetermined timing, and
A step of reflecting the output content of the declaration on the screen input by the operator regarding the content of the declaration from the customer regarding the predetermined target, and
To execute,
When the operator performs a predetermined input operation, the output step is based on the character string of the conversation performed between the start time of the conversation and the time when the predetermined input operation is performed by the operator. , Predict whether or not the customer has been proposed to implement a solution for improving the problem that has occurred in the predetermined target.
The step to be reflected on the screen is the predetermined step when it is predicted that the operator has not proposed to the customer the execution of the solution for improving the defect caused in the predetermined target in the output step. Displaying information on the screen prompting the customer to propose the execution of a solution for improving the problem that has occurred in the target.
program.

Images