JP2019047170A - Call center system, call center device, interaction method, and program therefor, comprising customer complaint detection function - Google Patents

Abstract

To provide a call center system, call center device, interaction method, and program therefor, that prevent a complaint from being caused during an AI interaction.SOLUTION: A call center server includes: an AI interaction processing unit for performing a voice interaction with a customer using an AI; a call content analysis unit for analyzing a call content on the basis of the customer's call voice; a complaint detection processing unit for detecting a complaint state or a preliminary complaint state on the basis of a result of the analysis by the call content analysis unit; an interaction switching unit for performing switching from an AI interaction by the AI interaction processing unit to an operator interaction in which an operator at an operator terminal performs a voice interaction with the customer; and a control unit that controls the AI interaction processing unit, the call content analysis unit, and the complaint detection processing unit, performs switching control on the interaction switching unit in the case that a detection result of the complaint detection processing unit shows the complaint state, and performs the switching control only when information on the customer's switching desire is acquired from a customer terminal in the case that a detection result of the complaint detection processing unit shows the preliminary complaint state.SELECTED DRAWING: Figure 2

Description

  In an inbound call center system such as a financial institution, the present invention has a function of detecting a customer in a complaint state during a customer call by analyzing a conversation between a customer and a virtual operator or an operator and preventing a situation from deteriorating. It is related with the provided call center system.

  In order to respond to complaints in the conventional call center system, the supervisor is appropriate by automatically judging the degree of customer emotion and the situation in which the operator is confused about the response, and monitoring by a request from the operator or by a supervisor who is a senior operator. By performing the processing and switching the call to the supervisor, it leads to prevention of deterioration of the complaint state during the call (see, for example, Patent Document 1, Patent Document 2, Patent Document 3, and Patent Document 4).

JP 2004-37989 A JP 2007-4000 JP 2011-9902 A JP, 2006-119634, A

However, in the prior art, when a call center system side AI (Artificial Intelligence) virtual operator (hereinafter abbreviated as “AI”) is interacting with a customer (AI dialogue), it is clear from the contents of the call that the customer In situations where it is not possible to conclude that a complaint is being expressed, there is no criteria for switching to advanced operator dialogue, so dialogue switching cannot be performed at the request of AI, and complaints during AI dialogue can be prevented. There was a problem that it was not possible. Further, as a response in the situation described above, it is possible to judge dialogue switching by monitoring the senior operator, but this leads to an increase in the burden on the senior operator and also loses the significance of unmanned operation by AI.
Even if the initial dialogue is an operator dialogue, in the above situation, it is difficult to make a decision to switch to an advanced operator dialogue (supervisor dialogue). There was a problem that could not be prevented. Further, as a response in the above situation, the supervisor can determine whether to switch the dialogue by monitoring, but this has led to an increase in the burden on the supervisor.

  An object of the present invention is to solve such a problem, and an object thereof is to prevent complaints during an AI dialogue or an operator dialogue.

  The call center system according to the present invention is a call center system that includes an operator work terminal, a supervisor work terminal, and a call center server, and provides a call center service to a customer terminal, wherein the call center server performs AI conversation with a customer by AI. A processing unit, a call content analysis unit that analyzes a call content based on a customer's call voice, a complaint detection processing unit that detects a complaint state or a complaint reserve army state based on an analysis result of the call content analysis unit, and the AI Switching from the AI dialogue by the dialogue processing unit to an operator dialogue in which the operator of the operator work terminal performs voice dialogue with the customer, or from the operator dialogue to a supervisor dialogue in which the supervisor of the supervisor work terminal performs voice dialogue with the customer. Dialog switching unit for switching , Controlling the AI dialogue processing unit, the call content analysis unit, and the complaint detection processing unit, and for the dialogue switching unit, if the detection result of the complaint detection processing unit is a complaint state, A control unit that controls switching and controls the switching only when the customer's switching request information is acquired from the customer terminal if the detection result of the complaint detection processing unit is in the complaint reserve army state. Features.

  A call center apparatus according to the present invention is a call center apparatus having a function of the call center server of a call center system that includes an operator work terminal and a call center server and provides a call center service to a customer terminal, and performs AI voice conversation with a customer through AI. A dialogue processing unit, a call content analysis unit for analyzing a call content based on a call voice of a customer, a complaint detection processing unit for detecting a complaint state or a complaint reserve army state based on an analysis result of the call content analysis unit, A dialogue switching unit for switching from an AI dialogue by the AI dialogue processing unit to an operator dialogue in which an operator of the operator work terminal performs a voice dialogue with a customer, the AI dialogue processing unit, the call content analysis unit, and the The complaint detection processing unit is controlled, and further, the complaint detection processing is performed for the dialogue switching unit. If the detection result of the part is a complaint state, the switching is controlled, and if the detection result of the complaint detection processing part is a complaint preliminary army state, the switching is performed only when the customer switching request information is acquired from the customer terminal. And a control unit that performs control.

  An interactive method of the present invention is an interactive method executed by a call center server in a call center system that includes an operator work terminal and a call center server and provides a call center service to a customer terminal. A processing step, a call content analysis step for analyzing a call content based on a customer's call voice, a complaint detection processing step for detecting a complaint state or a complaint reserve army state based on an analysis result of the call content analysis step, and the complaint If the detection result of the detection processing step is a complaint state, a step of switching from an AI dialogue in the AI dialogue processing step to an operator dialogue in which an operator of the operator work terminal performs a voice dialogue with a customer, and detection of the complaint detection processing step If the result is a complaint reserve Serial to the step of switching from said AI dialogue to the operator interaction only if the customer terminal has acquired the switching request information of the customer, characterized by having a.

  A program of the present invention is a program for an interactive method executed by the call center server in a call center system that includes an operator work terminal and a call center server and provides a call center service to a customer terminal, the computer of the call center server, AI dialogue processing step for performing voice dialogue with the customer by AI, call content analysis step for analyzing the call content based on the call voice of the customer, and complaint state or complaint reserve army state based on the analysis result of the call content analysis step If the detection result of the complaint detection processing step and the detection result of the complaint detection processing step is a complaint state, from the AI dialogue by the AI dialogue processing step to an operator dialogue in which the operator of the operator work terminal performs a voice dialogue with the customer Step to switch If the detection result of the complaint detection processing step is a complaint reserve army state, the step of switching from the AI dialogue to the operator dialogue is executed only when customer switch request information is acquired from the customer terminal. And

In the case of the AI dialogue, the present invention thus provided provides a criterion for switching to the operator dialogue even in a situation in which it is not possible to conclude from the content of the call that the customer expresses a complaint. Can prevent complaints inside. In addition, it is possible to eliminate the need for the operator to monitor as a response in the situation, reduce the burden on the operator, and achieve unmanned operation by AI.
Furthermore, even if the initial dialogue is an operator dialogue, it is possible to provide a criterion for switching to the supervisor dialogue in the above situation and prevent complaints during the operator dialogue. Further, it is possible to eliminate the necessity of monitoring by the supervisor as a countermeasure in the above situation, and to reduce the burden on the supervisor.

The block diagram which shows the whole structure of the call center system in a 1st Example. The block diagram which shows the functional structure of the call center server in a 1st Example. The flowchart which shows the algorithm of the complaint detection process in a 1st Example. The flowchart which shows the algorithm of the word space | interval calculation in a 1st Example. Explanatory drawing which shows the result of the speech recognition process in 1st Example, and the value of the flag with respect to it. Explanatory drawing which shows the value of the flag with respect to the result of the emotion analysis process in a 1st Example. Explanatory drawing which shows the result of the word space | interval calculation in a 1st Example, and the value of the flag with respect to it. Explanatory drawing which shows the customer's response of the process which asks the customer about dialog switching in a 1st Example, and the value of the flag with respect to it.

Embodiments of a call center system having a customer complaint detection function according to the present invention will be described below with reference to the drawings. The present embodiment also includes a call center device having a call center server function, a switching method, and a program thereof.
In the present invention, a call is a telephone conversation that a customer makes to the call center through the Internet or a telephone line, and a conversation is a response that the call center performs, such as a response to the call or an announcement to the customer. The AI dialog is defined as a dialog performed by the AI, the operator dialog is defined as a dialog performed by the operator, and the supervisor dialog is defined as a dialog performed by the supervisor. Further, an utterance is defined as a series of customer calls between call center dialogs.
The text data, speech detection time, NG word, and number of NG words are all speech recognition data as a recognition result of speech recognition by speech recognition technology, which is a conventional technology, and text data is a utterance text. Define as data. The utterance detection time is defined as the duration of utterance. The NG word is defined as a predetermined word in the utterance indicating that the customer is in a complaint state described later. The number of NG words is defined as the number of NG words in the utterance.
The anger emotion presence / absence information is emotion analysis data as an analysis result obtained by analyzing an utterance by an emotion analysis technique, and is defined as information indicating whether or not there is an anger emotion in the utterance.
Furthermore, a complaint state is defined as a state in which a customer clearly expresses a complaint about the content of a call. The complaint reserve army state is defined as a state in which a customer cannot express a complaint clearly from the content of the call, but may represent a complaint. A state that is neither a complaint state nor a complaint reserve state is defined as a no complaint state. Complaint detection is defined as detecting a complaint state or a preliminary complaint state.

FIG. 1 is a block diagram showing the overall configuration of the call center system in the first embodiment.

  In FIG. 1, a call center system 1 includes a call center server (call center device) 11 for controlling exchanges between a customer and an operator (including AI), input voice data, text data, utterance detection time, and predetermined keywords (in the present invention). A speech recognition system 12 configured by a known technique (for example, Patent Document 4) that outputs not the keyword itself but the number of NG words for a predetermined NG word), inputs speech and identifies the emotion of the speech (in the present invention) Only an anger emotion presence / absence information for each utterance is output) Emotion analysis system 13 configured by a known technique (for example, Patent Document 3), an operator work terminal 14 having a monitor for displaying the contents of a data storage unit described later ( (Including supervisor work terminal), connected via internet or telephone line Made from the customer terminal 15.

  FIG. 2 is a block diagram showing a functional configuration of the call center server in the first embodiment.

In FIG. 2, the control unit 111 has a function of controlling each functional block of the call center server 11 by executing a program and performing data processing and arithmetic processing of the memory unit 112 described later by executing the program. This program may be provided by being stored in the memory unit 112 described later, or may be provided by being stored in a memory built in the control unit 111.
The memory unit 112 has a function for temporarily storing data obtained from each processing unit of the call center server 11 and for processing and arithmetic processing of data performed by the control unit 111.
The data storage unit 113 stores voice recognition data (text data, utterance detection time, number of NG words), emotion analysis data (whether or not anger emotion presence / absence information), word intervals and text characters described later in FIG. It has a function of accumulating the value of each flag.
The AI dialog processing unit 114 has a function of receiving a customer's utterance and responding from the AI. The AI dialog processing unit includes a function of determining a customer's response to an AI customer announcement.
The voice recognition input / output processing unit 115 has a function of passing a customer's speech voice to the voice recognition system 12 and receiving voice recognition data from the voice recognition system 12.
The emotion analysis input / output processing unit 116 has a function of passing the customer's utterance voice to the emotion analysis system 13 and receiving the presence / absence information of anger emotion as emotion analysis data from the emotion analysis system 13.
The complaint detection processing unit 117 has a function of detecting a complaint state or a preliminary complaint state based on information received from the speech recognition input / output processing unit 115, the emotion analysis input / output processing unit 116, and the word interval processing unit 118 described later. . The call content analysis unit has the functions of the voice recognition input / output processing unit 115, the emotion analysis input / output processing unit 116, and the word interval processing unit 118.
The word interval processing unit 118 has a function of calculating a word interval issued by the customer based on the information received from the voice recognition input / output processing unit 115 and passing the obtained value to the complaint detection processing unit 117.
The dialogue switching unit 119 has a function of switching from the AI dialogue to the operator dialogue based on the correspondence of the AI dialogue processing unit 114.

FIG. 3 is a flowchart showing an algorithm for a customer complaint detection process in the first embodiment. Description will be made with reference to FIGS. 1, 2, 5, 6, 7, and 8 according to the step represented by S in the flowchart.
The AI dialog processing unit 114, the call content analysis unit (voice recognition input / output processing unit 115, emotion analysis input / output processing unit 116 and word interval processing unit 118), and the complaint detection processing unit 117 are all controlled by the control unit 111. It is assumed that the procedure is controlled according to this flowchart, and the description of the control operation of the control unit 111 is omitted in the following description.

    S100: The value of the identification number i indicating the number of customer utterances in the memory unit 112 is set to zero. The value of i is incremented by 1 every time the customer speaks in S101 described later.

    S101: Each of the five flags of the memory unit 112, that is, the complaint state detection flag compt, the complaint preliminary army state detection flag danger, the dialogue switching flag change, the same announcement continuation flag cont1, and the conversation continuation flag cont2 Number) is set to the initial value 0 each time the utterance is started. In this step, the identification number i is set to i + 1.

  S102: This step is a step of performing voice recognition processing. That is, voice data is input to the voice recognition system 12 for each utterance, and the text data received by the voice recognition input / output processing unit 115 from the voice recognition system 12 for each utterance, the utterance detection time, and the number of NG words are stored in the memory unit 112. . Further, these stored data are stored in the data storage unit 113 for each identification number i. Predetermined NG words are set in the speech recognition system 12 in advance, the speech recognition system 12 detects NG words based on the input speech data, and the number of NG words for each input speech data is speech recognition input / output It is assumed that the data is passed to the processing unit 115.

  S103: The complaint detection processing unit 117 receives the number of NG words stored in the memory unit 112, and determines whether or not it is in a complaint state according to the number of NG words. As an example of the criterion, when the number of NG words is 1 or more, it is determined as a complaint state. The complaint detection processing unit 117 sets the flag compt to 1 when it is determined that it is a complaint state, and sets the flag compt to 0 when it is determined that it is not a complaint state. Then, the value of “compt” is stored in the data storage unit 113 for each identification number i.

  FIG. 5 is an explanatory diagram showing the result of the speech recognition process and the flag value corresponding thereto. That is, it is an example of data stored in the data storage unit 113 for each identification number i in S102 and S103 (the utterance detection time is omitted here because it is not used for determination in S103). The example of FIG. 5 is a case where the word “idiot” is included in the predetermined NG word, and the text data “I don't hear it because it ’s an idiot” to “idiot” in the identification number 4. In this example, the number of NG words in the identification number 4 detected as NG word is 1, and the flag compt is 1 at this time.

  S104: If the flag compt is 1, the flag change is set to 1 (S112) and the process proceeds to S113. If the flag compt is 0, the process proceeds to S105.

S105: This step is a step of performing emotion analysis processing. That is, voice data is input to the emotion analysis system 13 for each utterance, and the emotion analysis input / output processing unit 116 stores the presence / absence information of the anger emotion received from the emotion analysis system 13 for each utterance in the memory unit 112. Further, this stored data is stored in the data storage unit 113 for each identification number i.
The emotion analysis system 13 detects the presence or absence of anger emotion from the frequency or the like based on the input voice data, and passes the presence or absence information of anger emotion to the emotion analysis input / output processing unit 116 for each input voice data. To do.

  S106: The complaint detection processing unit 117 receives the presence / absence information of the anger emotion in the memory unit 112 in S105, determines that there is anger emotion information, determines that it is a complaint state, and if there is no anger emotion information, it is not a complaint state to decide. The complaint detection processing unit 117 sets the flag compt to 1 when it is determined that it is a complaint state, and sets the flag compt to 0 when it is determined that it is not a complaint state. Then, the value of “compt” is stored in the data storage unit 113 for each identification number i.

  FIG. 6 is an explanatory diagram showing a result of the emotion analysis process and a flag value corresponding thereto. That is, it is an example of data stored in the data storage unit 113 for each identification number i in S105 and S106. The example of FIG. 6 is a case where the anger emotion is included in the identification number 4 when the anger emotion is included in the text data “That's why I can't hear you”. In this example, the flag compt is 1 at this time.

  S107: If the flag compt is 1, the flag change is set to 1 (S112) and the process proceeds to S113. If the flag compt is 0, the process proceeds to S108.

    S108: This step is a step of calculating a word interval. The word interval processing unit 118 calculates the word interval based on the speech recognition data stored in the memory unit 112, and the result (word interval SP (i), number of text characters (i) and value of danger described later) is stored in the memory unit. 112 and stored in the data storage unit 113 for each identification number i. This step will be described in detail with reference to FIG.

  FIG. 7 is an explanatory diagram showing the result of word interval calculation and the flag value corresponding thereto. That is, it is an example of data stored in the data storage unit 113 for each identification number i in S108 (speech detection time is in S102). The example of FIG. 7 is the case where the word interval PH (3) of the text data “I'm sorry, I can't hear you well” with the identification number 3 is less than PH (1), and the flag danger at this time is This is an example of 1.

  S109: If the flag “danger” set in S108 is 1, it is determined that it is in the complaint reserve army state, and the process proceeds to a determination process (S110) as to whether or not the customer desires to switch the dialogue. If the flag “danger” is 0, there is no complaint The process proceeds to the process (S116) to continue the dialogue as it is.

  S110: This step is a process for inquiring the customer about dialog switching, that is, a process for obtaining customer switching request information from the customer terminal 15. As a specific example, the AI dialogue processing unit 114 announces to the customer “Do you want to switch to operator dialogue” (this is only an example of this announcement), and the AI dialogue processing unit 114 determines and obtains the response. To do. When the switching request information is acquired, the AI dialogue processing unit sets the flag change to 1 and proceeds to S111. The announcement method may be a speech synthesis method or a recorded standard sentence (synthetic speech or human speech) may be played (the same applies hereinafter). If the switching request information is not acquired, the process proceeds to S111 with the flag change set to 0 (initial value). As a method for acquiring the customer switching request information, even if the AI dialogue processing unit 114 obtains the customer's reply from the AI dialogue processing unit 114 by voice recognition, the announcement of the AI dialogue processing unit 114 is the customer. Alternatively, the push button may be pressed and the bush button signal may be received and acquired by the AI dialogue processing unit 114.

  FIG. 8 is an explanatory diagram showing the customer's response to the process for asking the customer to switch the dialog and the flag value for the response, and is an example in which the customer's response is determined by voice recognition. When the patterns 1, 2, and 3 are desired to be switched and each change is 1, the patterns 4, 5, and 6 are not desired to be switched, and each change is 0.

  S111: When the flag change is 1, the process proceeds to the dialogue switching process (S113). When the flag change is 0, the process proceeds to the next announcement process (S114).

  S113: When the flag change is 1, the dialogue switching unit 119 switches from the AI dialogue by the AI dialogue processing unit 114 to the operator dialogue by the operator work terminal 14. The dialogue switching unit 119 transfers the accumulated data (such as text data during the AI dialogue) in the data accumulation unit 113 to the operator work terminal 14 and ends the customer complaint detection process.

  S114: This step is a process for asking the customer to continue the AI dialogue. In other words, it is a process of obtaining information on the continuation of the customer's AI dialogue from the customer terminal 15. As a specific example, the AI dialogue processing unit 114 announces to the customer, “Do you want to continue the AI dialogue in the future?” (This is only an example of this announcement), and the AI dialogue processing unit 114 gives the response. Determine and get. When the AI dialog processing unit 114 acquires the AI dialog continuation request information, the flag cont1 is set to 1. If the AI conversation continuation request information is not acquired, cont1 remains 0 (initial value). As a method for acquiring the customer's AI dialogue switching request information, even if the AI dialogue processing unit 114 obtains the customer's response from the AI dialogue processing unit 114 by voice recognition, A method may be used in which the announcement prompts the customer to push the push button and the AI dialogue processing unit 114 receives and acquires the bush button signal.

  S115: If the flag cont1 is 1, the process proceeds to S116. If the flag cont1 is 0, the customer complaint detection process is terminated.

  S116: The AI dialog processing unit 114 determines whether the customer's call is continued. If it is determined that the process continues, the AI dialog processing unit sets the flag cont2 to 1.

  S117: If the flag cont2 is 1, the process proceeds to S101. If the flag cont2 is 0, the customer complaint detection process is terminated. In addition, if the telephone call with the customer continues after finishing the complaint detection process, the above complaint detection process may be started again.

FIG. 4 is a flowchart showing a word interval calculation algorithm in the first embodiment. Explanation will be given with reference to FIG. 1 and FIG. 2 according to the step represented by S in the flowchart S201: The word interval processing unit 118 and the text data SP (i) corresponding to the identification number i stored in the memory unit 112 and the speech The detection time T (i) is received.

  S202: The word interval processing unit 118 calculates the number of text characters N (i) from SP (i) to SP (i), and calculates the word interval PH (i) = T (i) / N (i).

  S203: The word interval processing unit 118 calculates PH (1) -PH (i), and determines that it is not a claim reserve army when it exceeds 0, and is not a claim reserve army when it is 0 or less.

  S204: If the result of S203 is the claim reserve army, the flag danger is set to 1.

  S205: If the result of S203 is not the claim reserve army, the flag danger is set to 0.

    According to the first embodiment, in the case of the AI dialogue, even if it is not possible to conclude from the contents of the call that the customer clearly expresses a complaint, a criterion for switching to the operator dialogue is provided, and the AI dialogue is provided. Can prevent complaints inside. In addition, it is possible to eliminate the need for the operator to monitor as a response in the situation, reduce the burden on the operator, and achieve unmanned operation by AI.

The contents of the first embodiment can be applied to a call center system having a customer complaint detection function when switching from an operator dialogue to a supervisor dialogue. In the second embodiment, the operator work terminal 14 in FIG. 1 includes a supervisor work terminal. The initial dialogue is not an AI dialogue by the AI dialogue processing unit 114 but an operator dialogue by the operator work terminal 14.
Only differences from the first embodiment will be described below.
<About the dialog switching unit 119>
The dialogue switching unit 119 has a function of switching from the operator dialogue to the supervisor dialogue based on the correspondence of the AI dialogue processing unit 114 or based on the correspondence of the operator.
<About S110 and S113 to S117 in the description of FIG. 3>
S110: This step is a process for inquiring the customer about dialog switching, that is, a process for obtaining customer switching request information from the customer terminal 15. As a specific example, the AI dialogue processing unit 114 or the operator makes an announcement to the customer, “Do you want to switch to supervisor dialogue?” (This is only an example of this announcement), and the AI dialogue processing unit 114 or the operator sends the response. Determine and get. When the switching request information is acquired, the AI dialogue processing unit sets the flag change to 1 and proceeds to S111. The method of announcement by the AI dialogue processing unit 114 may be a method using speech synthesis or a method of playing a recorded fixed sentence (synthetic speech or human speech) (the same applies hereinafter). If the flag change is 0, the process proceeds to S112. As a method for the AI dialog processing unit 114 to acquire the customer switching request information, even if the AI dialog processing unit 114 acquires the customer's response by voice recognition, the announcement of the AI dialog processing unit 114 is the customer. Alternatively, the push button may be pressed and the bush button signal may be received and acquired by the AI dialogue processing unit 114.

  S113: When the flag change is 1, the dialogue switching unit 119 switches from the operator dialogue by the operator work terminal 14 to the supervisor dialogue by the supervisor work terminal. The dialogue switching unit 119 transfers the accumulated data (such as text data during the operator dialogue) from the data accumulation unit 113 to the supervisor work terminal 14 and ends the customer complaint detection process.

  S114 to S117: In the second embodiment, these processes are not performed. If danger = 0 in S109 and change = 0 in S111, both proceed to S101.

    According to the second embodiment, even when the initial dialogue is an operator dialogue, in a situation where it is not possible to conclude from the contents of the call that the customer clearly represents a complaint, a criterion for switching to the supervisor dialogue is provided. Complaints during operator interaction can be prevented. Further, it is possible to eliminate the necessity of monitoring by the supervisor as a countermeasure in the above situation, and to reduce the burden on the supervisor.

DESCRIPTION OF SYMBOLS 1 Call center system 11 Call center server 12 Voice recognition system 13 Emotion analysis system 14 Operator work terminal 15 Customer terminal 111 Control part 112 Memory part 113 Data storage part 114 AI dialog processing part 115 Voice recognition input / output processing part 116 Emotion analysis input / output processing part 117 Complaint detection processing unit 118 Word interval processing unit 119 Dialog switching unit

Claims (6)

A call center system comprising an operator work terminal, a supervisor work terminal, and a call center server, and providing a call center service to a customer terminal,
The call center server
An AI dialogue processing unit for performing voice dialogue with customers by AI;
A call content analysis unit that analyzes the call content based on the customer's call voice,
A complaint detection processing unit that detects a complaint state or a preliminary complaint state based on the analysis result of the call content analysis unit,
Switching from an AI dialogue by the AI dialogue processing unit to an operator dialogue in which an operator of the operator work terminal performs a voice dialogue with a customer, or a supervisor in which a supervisor of the supervisor work terminal performs a voice dialogue with a customer from the operator dialogue A dialogue switching unit for switching to dialogue,
The AI dialogue processing unit, the call content analysis unit, and the complaint detection processing unit are controlled. Further, for the dialogue switching unit, if the detection result of the complaint detection processing unit is a complaint state, the switching is performed. A control unit that controls the switching only when the customer's switching request information is acquired from the customer terminal if the detection result of the complaint detection processing unit is a complaint reserve army state,
A call center system comprising: In the call center system according to claim 1,
The dialogue switching unit is a dialogue switching unit that switches from the AI dialogue to the operator dialogue,
In the case where the detection result of the complaint detection process is in the complaint reserve army state and the switching request information cannot be obtained, the control unit further acquires the customer's AI conversation continuation request information from the customer terminal. A call center system for controlling a conversation processing unit to continue an AI conversation. In the call center system according to claim 1 or 2,
The dialogue switching unit is a dialogue switching unit that switches from the AI dialogue to the operator dialogue,
Whether the switching request information or the continuation request information is requested by the AI dialog processing unit to the customer on the customer terminal to switch the information in the switching request information or to continue the AI dialog in the continuation request information. The AI dialogue processing unit recognizes the voice of the customer's response to the announcement and the AI dialogue processing unit receives the push button signal generated by the customer's customer terminal operation for the announcement. Call center system, characterized by A call center device having a function of the call center server of a call center system comprising an operator work terminal and a call center server and providing a call center service to a customer terminal,
An AI dialogue processing unit for performing voice dialogue with customers by AI;
A call content analysis unit that analyzes the call content based on the customer's call voice,
A complaint detection processing unit that detects a complaint state or a preliminary complaint state based on the analysis result of the call content analysis unit,
A dialogue switching unit for switching from an AI dialogue by the AI dialogue processing unit to an operator dialogue in which an operator of the operator work terminal performs a voice dialogue with a customer;
The AI dialogue processing unit, the call content analysis unit, and the complaint detection processing unit are controlled. Further, for the dialogue switching unit, if the detection result of the complaint detection processing unit is a complaint state, the switching is performed. A control unit that controls the switching only when the customer's switching request information is acquired from the customer terminal if the detection result of the complaint detection processing unit is a complaint reserve army state,
A call center device characterized by comprising: An interaction method executed by a call center server in a call center system that includes an operator work terminal and a call center server and provides a call center service to a customer terminal,
AI dialogue processing step for performing voice dialogue with the customer by AI;
Call content analysis step for analyzing call content based on customer call voice,
Complaint detection processing step for detecting a complaint state or a complaint reserve army state based on the analysis result of the call content analysis step;
If the detection result of the complaint detection processing step is a complaint state, switching from the AI dialogue by the AI dialogue processing step to an operator dialogue in which an operator of the operator work terminal performs a voice dialogue with a customer;
If the detection result of the complaint detection processing step is a complaint reserve army state, the step of switching from the AI dialogue to the operator dialogue only when obtaining customer switching request information from the customer terminal;
A dialogue method characterized by comprising: A program for an interaction method executed by the call center server in a call center system comprising an operator work terminal and a call center server and providing a call center service to a customer terminal,
In the computer of the call center server,
AI dialogue processing step for performing voice dialogue with the customer by AI;
Call content analysis step for analyzing call content based on customer call voice,
Complaint detection processing step for detecting a complaint state or a complaint reserve army state based on the analysis result of the call content analysis step;
If the detection result of the complaint detection processing step is a complaint state, switching from the AI dialogue by the AI dialogue processing step to an operator dialogue in which an operator of the operator work terminal performs a voice dialogue with a customer;
If the detection result of the complaint detection processing step is a complaint reserve army state, the step of switching from the AI dialogue to the operator dialogue only when obtaining customer switching request information from the customer terminal;
A program characterized by having executed.

Images