A customer service quality management improvement system and method based on artificial intelligence

Abstract

The present application relates to the field of voice customer service quality management, specifically a customer service quality management improvement system and method based on artificial intelligence. First, the customer video information and voice signal are obtained through the information acquisition unit, and the customer service staff video information, voice signal and electromyographic signal are obtained; then, the signal processing unit receives the voice, video and bioelectric signals collected by the signal acquisition unit, filters and reduces noise on the above signals respectively, and then outputs them to the emotion recognition unit; then, the emotion recognition unit obtains and outputs the customer emotion information in time sequence, the customer service staff emotion information in time sequence, and the key information in the customer service staff video information and voice signal; then, the quality judgment unit scores the service quality of the customer service staff, determines whether the high-frequency keywords are valid keywords or invalid keywords according to the changes in customer emotions, and sends the scores and judgment results to the quality feedback unit; finally, the quality feedback unit counts the service quality scores of the customer service staff, and displays the high-frequency keywords sent by the quality judgment unit and the changes in customer emotions after the high-frequency keywords appear.

Description

Technical Field

[0001] The present application relates to the field of voice customer service quality management, and specifically to a customer service quality management improvement system and method based on artificial intelligence. Background Art

[0002] Insurance industry practitioners need to record audio and video when communicating with customers. In order to ensure the validity and availability of the audio and video data, the data needs to be quality checked. In the past quality inspection process, the video was uploaded to the quality inspection platform and mainly inspected manually. This method is not only inefficient, but also very wasteful of manpower. The strong subjectivity of manual quality inspection also leads to large deviations in the test results.

[0003] At the same time, due to factors such as major epidemics, online business has increased dramatically, and the number of insurance business promotions and contract signings conducted through remote video has gradually increased. The amount of data that needs to be recorded and videotaped for quality inspection has increased dramatically, greatly increasing the workload of quality inspectors and further reducing the efficiency and quality of quality inspection.

[0004] Based on this, a solution has emerged in which machines replace manpower for quality inspection. However, the existing technology often performs indiscriminate identification and quality inspection on all audio and video recordings, which wastes quality inspection resources and increases the computing power of artificial intelligence machines. At the same time, some customer service personnel who have been in the industry for a long time have gradually mastered the skills of controlling expressions and voices, and simply relying on traditional quality inspection methods can easily overlook some detailed information. Summary of the invention

[0005] In order to solve the above problems, a customer service quality management improvement system based on artificial intelligence is proposed, which includes:

[0006] Information collection unit, signal processing unit, emotion recognition unit, quality judgment unit and quality feedback unit;

[0007] The information collection unit includes a customer information collection module and a customer service staff information collection module, wherein the customer information collection module includes a first video collection device and a first voice signal collection device, and the customer service staff information collection module includes a second video collection device, a second voice signal collection device and an electromyographic signal collection device;

[0008] The customer service personnel information collection module may also include an electroencephalogram signal collection device;

[0009] The video collection device for collecting the customer video information and the customer service staff video information may be the same device;

[0010] The voice signal collection device for collecting the customer voice signal and the customer service staff voice signal may be the same device;

[0011] The signal processing unit receives the voice, video and bioelectric signals collected by the signal collection unit, performs filtering and noise reduction on the above signals respectively and then outputs them to the emotion recognition unit;

[0012] The bioelectric signal refers to the electromyographic signal collected by the electromyographic signal acquisition device and / or the electroencephalographic signal collected by the electroencephalographic signal acquisition device;

[0013] The electromyographic signal acquisition device may be a smart bracelet;

[0014] The EEG signal acquisition device may be a head-mounted EEG signal acquisition device;

[0015] When the video acquisition device for acquiring the customer video information and the customer service staff video information is the same device, the signal processing unit further segments and extracts the customer information and the customer service staff information in the video information;

[0016] When the voice signal collecting device for collecting the customer voice signal and the customer service staff voice signal is the same device, the signal processing unit further divides and extracts the customer information and the customer service staff information in the voice signal;

[0017] The emotion recognition unit includes a customer emotion recognition module and a customer service staff emotion recognition module;

[0018] The customer emotion recognition module inputs the customer video information and voice signal collected by the first video acquisition device and the first voice signal acquisition device and sent by the signal processing unit into a customer emotion recognition model pre-established based on a convolutional neural network, and outputs the customer emotion information in time series;

[0019] The customer service staff emotion recognition module inputs the customer service staff video information, voice signal and electromyographic signal collected by the second video acquisition device and the second voice signal acquisition device and the electromyographic signal acquisition device sent by the signal processing unit into the customer service staff emotion recognition model pre-established based on the convolutional neural network, and outputs the customer service staff emotion information in time series;

[0020] At the same time, the customer service staff emotion recognition module also extracts and saves key information from the customer service staff video information and voice signals collected by the second video acquisition device and the second voice signal acquisition device and sent by the signal processing unit, wherein the key information includes but is not limited to inappropriate expressions of customer service staff such as dissatisfaction, anger, and annoyance, as well as rapid speech speed, excessive volume, and inappropriate or even illegal keywords;

[0021] In addition, the customer service staff emotion recognition module also performs voiceprint recognition based on the customer service staff's voice signals to determine and record the customer service staff's identity information.

[0022] The quality judgment unit extracts the first timestamp when the customer's emotions fluctuate violently, wherein the violent emotions fluctuate violently refers to a time point or time period when the signal amplitude in the customer's video information or voice signal is greater than a preset threshold after the signals are filtered and denoised;

[0023] The quality judgment unit extracts a second timestamp when the customer service staff's emotions fluctuate violently, wherein the customer service staff's emotions fluctuate violently refers to a time point or time period when the amplitude of the electromyographic signal is greater than a preset threshold after the signal is filtered and denoised;

[0024] Taking the time of the first timestamp or the second timestamp as the time reference, extend forward and backward by a certain length of time to intercept the emotional information of the customer and the customer service staff;

[0025] When the timestamp is a time period, the preset time length is extended toward the start point of the time period and toward the end point of the time period;

[0026] When the quality inspection system is used for real-time monitoring of customer service quality, it only extends a certain length of time before and after the time base to intercept the audio signal of the video signal of the customer service staff's emotional information;

[0027] The preset duration is selected from 300ms to 3min;

[0028] Determining changes in customer emotions through intercepted customer emotion information, wherein the customer emotion changes are divided into positive changes and negative changes;

[0029] Positive change refers to the customer's change from negative emotion to positive emotion or from positive emotion to further positive emotion, and negative change refers to the customer's change from positive emotion to negative emotion or from negative emotion to further negative emotion;

[0030] The positive emotions include but are not limited to happiness, pleasure, excitement, obvious interest, etc., and the negative emotions include but are not limited to sadness, worry, anxiety, anger, dissatisfaction, restlessness, boredom, etc.;

[0031] In addition, the change from negative emotions to calmness is also a positive change;

[0032] The scoring coefficient K is set according to changes in customer emotions, and the scoring coefficient K can be set flexibly.

[0033] Count the number of times N when customer service staff trigger the judgment rules during the quality inspection period and the high-frequency keywords;

[0034] Among them, high-frequency keywords refer to keywords that appear more than three times in the same quality inspection period;

[0035] The judging rules can be set flexibly, including but not limited to one or more of the following:

[0036] The customer service staff used prohibited keywords during the conversation;

[0037] The customer service staff and the customer remain silent for a period of time greater than the predetermined time threshold during the conversation;

[0038] The customer service staff's speaking speed during the conversation is higher than the preset speaking speed threshold;

[0039] The customer service staff showed a forbidden expression during the conversation;

[0040] The customer service staff made prohibited body movements during the conversation;

[0041] The customer service agent used a prohibitive tone during the conversation;

[0042] The volume of the customer service staff during the conversation exceeds the preset volume threshold, or is less than the preset volume threshold, for example, the volume is too loud, or too soft; or the volume changes suddenly.

[0043] The customer service staff forcibly interrupted the customer during the conversation.

[0044] The customer service staff recommended keywords during the conversation.

[0045] The customer service agent remains silent during the conversation for a period greater than a predetermined time threshold.

[0046] The length of time that a customer service representative maintains a calm speaking pace during a conversation.

[0047] The customer service agent appeared calm or positive during the conversation.

[0048] The customer service agent showed calm or positive body language during the conversation.

[0049] The customer service agent maintained a calm or positive tone during the conversation.

[0050] How long the customer service agent maintains a calm voice during the conversation.

[0051] The service quality score of the customer service personnel during the quality inspection period is determined according to the number of times, the pre-set evaluation rule coefficient M and the scoring coefficient K.

[0052] The processing method for the duration is as follows: a duration threshold t is set, for example, 15s, and each time a duration threshold t is reached is counted as one time, and the time less than a duration threshold is not counted in the number of times.

[0053] Among them, the pre-set evaluation rule coefficient M refers to assigning a corresponding score to each evaluation rule triggered, wherein prohibited behaviors receive negative scores and recommended behaviors receive positive scores. The scores of different evaluation rules can be set flexibly.

[0054] Customer service staff's service quality score = K*(100+∑N i M i ),

[0055] Where Ni is the number of times the i-th evaluation rule is triggered; Mi is the score of the i-th evaluation rule.

[0056] The number of occurrences L of high-frequency keywords is counted, and the high-frequency keywords are sorted according to the number of occurrences and sent to the backend.

[0057] The time when each high-frequency keyword appears is used as the third timestamp, customer emotion information of a preset time period is intercepted after the third timestamp, and changes in customer emotion during the time period are analyzed. According to the changes in customer emotion, it is determined whether the high-frequency keyword is a valid keyword or an invalid keyword, and the judgment result is sent to the quality feedback unit.

[0058] The quality feedback unit is used to count the service quality scores of customer service personnel, display the high-frequency keywords sent by the quality judgment unit and the changes in customer emotions after the high-frequency keywords appear.

[0059] A method for improving customer service quality management based on artificial intelligence, comprising the following steps:

[0060] First, the information collection unit is started to obtain the customer's video information and voice signal, and the customer service staff's video information, voice signal and electromyographic signal;

[0061] Then, the signal processing unit receives the voice, video and bioelectric signals collected by the signal collection unit, performs filtering and noise reduction on the above signals respectively and then outputs them to the emotion recognition unit;

[0062] Afterwards, the emotion recognition unit acquires and outputs the customer emotion information in time series, the customer service staff emotion information in time series, and the key information in the customer service staff video information and voice signal;

[0063] Afterwards, the quality judgment unit scores the service quality of the customer service personnel, determines whether the high-frequency keywords are valid keywords or invalid keywords according to the changes in customer emotions, and sends the score and judgment results to the quality feedback unit;

[0064] Finally, the quality feedback unit counts the service quality scores of the customer service staff, displays the high-frequency keywords sent by the quality judgment unit and the changes in customer emotions after the high-frequency keywords appear.

[0065] When the video acquisition device for acquiring the customer video information and the customer service staff video information is the same device, the signal processing unit firstly segments and extracts the customer information and the customer service staff information in the video information.

[0066] When the voice signal collecting device for collecting the customer voice signal and the customer service staff voice signal is the same device, the signal processing unit first divides and extracts the customer information and the customer service staff information in the voice signal.

[0067] The beneficial effects of the present invention are:

[0068] (1) In the process of collecting emotional information of customer service personnel, bioelectric signals such as electromyographic signals or electroencephalographic signals are introduced as references to avoid information masking caused by customer service personnel controlling their expressions and voices;

[0069] (2) Quality checks are only performed at critical moments when customer service staff and customers’ emotions change dramatically, effectively reducing the amount of data calculations, saving calculation time and calculation costs;

[0070] (3) An effective method for scoring the quality of customer service personnel is provided. Based on this method, the shortcomings of customer service personnel in the process of voice communication with customers can be identified through voice recognition and voice analysis, so that the salesperson can make corrections later. The good performance can also be identified, which can be shared with colleagues later. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a schematic diagram of the customer service quality management improvement system based on artificial intelligence described in this application;

[0072] Figure 2 This is a schematic diagram of the information collection unit described in this application;

[0073] Figure 3 The process of calculating the service quality score of the customer service personnel described in this application;

[0074] Figure 4 The implementation process of the customer service quality management improvement method based on artificial intelligence described in this application; DETAILED DESCRIPTION

[0075] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0076] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0077] In the present application, the terms "upper", "lower", "left", "right", "front", "back", "top", "bottom", "inner", "outer", "middle", "vertical", "horizontal", "lateral", "longitudinal" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings. These terms are mainly used to better describe the present invention and its embodiments, and are not used to limit the indicated devices, elements or components to have a specific orientation, or to be constructed and operated in a specific orientation.

[0078] In addition, some of the above terms may be used to express other meanings in addition to indicating orientation or positional relationship. For example, the term "on" may also be used to express a certain dependency or connection relationship in some cases. For those skilled in the art, the specific meanings of these terms in the present invention can be understood according to specific circumstances.

[0079] In addition, the terms "installed", "set", "provided with", "connected", "connected", and "socketed" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be an internal connection between two devices, elements, or components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0080] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0081] A customer service quality management and improvement system based on artificial intelligence, comprising:

[0082] Information collection unit, signal processing unit, emotion recognition unit, quality judgment unit and quality feedback unit;

[0083] The information collection unit includes a customer information collection module and a customer service staff information collection module, wherein the customer information collection module includes a first video collection device and a first voice signal collection device, and the customer service staff information collection module includes a second video collection device, a second voice signal collection device and an electromyographic signal collection device;

[0084] The customer service personnel information collection module may also include an electroencephalogram signal collection device;

[0085] The video collection device for collecting the customer video information and the customer service staff video information may be the same device;

[0086] The voice signal collection device for collecting the customer voice signal and the customer service staff voice signal may be the same device;

[0087] The signal processing unit receives the voice, video and bioelectric signals collected by the signal collection unit, performs filtering and noise reduction on the above signals respectively and then outputs them to the emotion recognition unit;

[0088] The bioelectric signal refers to the electromyographic signal collected by the electromyographic signal acquisition device and / or the electroencephalographic signal collected by the electroencephalographic signal acquisition device;

[0089] The electromyographic signal acquisition device may be a smart bracelet;

[0090] The EEG signal acquisition device may be a head-mounted EEG signal acquisition device;

[0091] When the video acquisition device for acquiring the customer video information and the customer service staff video information is the same device, the signal processing unit further segments and extracts the customer information and the customer service staff information in the video information;

[0092] When the voice signal collecting device for collecting the customer voice signal and the customer service staff voice signal is the same device, the signal processing unit further divides and extracts the customer information and the customer service staff information in the voice signal;

[0093] The emotion recognition unit includes a customer emotion recognition module and a customer service staff emotion recognition module;

[0094] The customer emotion recognition module inputs the customer video information and voice signal collected by the first video acquisition device and the first voice signal acquisition device and sent by the signal processing unit into a customer emotion recognition model pre-established based on a convolutional neural network, and outputs the customer emotion information in time series;

[0095] The customer service staff emotion recognition module inputs the customer service staff video information, voice signal and electromyographic signal collected by the second video acquisition device and the second voice signal acquisition device and the electromyographic signal acquisition device sent by the signal processing unit into the customer service staff emotion recognition model pre-established based on the convolutional neural network, and outputs the customer service staff emotion information in time series;

[0096] At the same time, the customer service staff emotion recognition module also extracts and saves key information from the customer service staff video information and voice signals collected by the second video acquisition device and the second voice signal acquisition device and sent by the signal processing unit, wherein the key information includes but is not limited to inappropriate expressions of customer service staff such as dissatisfaction, anger, and annoyance, as well as rapid speech speed, excessive volume, and inappropriate or even illegal keywords;

[0097] In addition, the customer service staff emotion recognition module also performs voiceprint recognition based on the customer service staff's voice signals to determine and record the customer service staff's identity information.

[0098] The quality judgment unit extracts the first timestamp when the customer's emotions fluctuate violently, wherein the violent emotions fluctuate violently refers to a time point or time period when the signal amplitude in the customer's video information or voice signal is greater than a preset threshold after the signals are filtered and denoised;

[0099] The quality judgment unit extracts a second timestamp when the customer service staff's emotions fluctuate violently, wherein the customer service staff's emotions fluctuate violently refers to a time point or time period when the amplitude of the electromyographic signal is greater than a preset threshold after the signal is filtered and denoised;

[0100] Taking the time of the first timestamp or the second timestamp as the time reference, extend forward and backward by a certain length of time to intercept the emotional information of the customer and the customer service staff;

[0101] When the timestamp is a time period, the preset time length is extended toward the start point of the time period and toward the end point of the time period;

[0102] When the quality inspection system is used for real-time monitoring of customer service quality, it only extends a certain length of time before and after the time base to intercept the audio signal of the video signal of the customer service staff's emotional information;

[0103] The preset duration is selected from 300ms-3min;

[0104] Determine the customer's emotional changes through intercepted customer emotional information, and the customer's emotional changes are divided into positive changes and negative changes; positive changes refer to the customer's change from negative emotions to positive emotions or from positive emotions to further positive emotions, and negative changes refer to the customer's change from positive emotions to negative emotions or from negative emotions to further negative emotions;

[0105] The positive emotions include but are not limited to happiness, pleasure, excitement, obvious interest, etc., and the negative emotions include but are not limited to sadness, worry, anxiety, anger, dissatisfaction, restlessness, boredom, etc.;

[0106] In addition, the change from negative emotions to calmness is also a positive change;

[0107] Set the rating coefficient K according to changes in customer sentiment, for example:

[0108] Customer sentiment changes Rating coefficient K Change from negative emotions to positive emotions 1.2 From positive emotions to further positive emotions 1.1 From positive emotions to negative emotions 0.9 From negative emotions to further negative emotions 0.8

[0109] Among them, the scoring coefficient K can be set flexibly.

[0110] Count the number of times N when customer service staff trigger the judgment rules during the quality inspection period and the high-frequency keywords;

[0111] Among them, high-frequency keywords refer to keywords that appear more than three times in the same quality inspection period;

[0112] The judging rules can be set flexibly, including but not limited to one or more of the following:

[0113] The customer service staff used prohibited keywords during the conversation;

[0114] The customer service staff and the customer remain silent for a period of time greater than the predetermined time threshold during the conversation;

[0115] The customer service staff's speaking speed during the conversation is higher than the preset speaking speed threshold;

[0116] The customer service staff showed a forbidden expression during the conversation;

[0117] The customer service staff made prohibited body movements during the conversation;

[0118] The customer service agent used a prohibitive tone during the conversation;

[0119] The volume of the customer service staff during the conversation exceeds the preset volume threshold, or is less than the preset volume threshold, for example, the volume is too loud, or too soft; or the volume changes suddenly.

[0120] The customer service staff forcibly interrupted the customer during the conversation.

[0121] The customer service staff recommended keywords during the conversation.

[0122] The customer service agent remains silent during the conversation for a period greater than a predetermined time threshold.

[0123] The length of time that a customer service representative maintains a calm speaking pace during a conversation.

[0124] The customer service agent appeared calm or positive during the conversation.

[0125] The customer service agent showed calm or positive body language during the conversation.

[0126] The customer service agent maintained a calm or positive tone during the conversation.

[0127] How long the customer service agent maintains a calm voice during the conversation.

[0128] The service quality score of the customer service personnel during the quality inspection period is determined according to the number of times, the pre-set evaluation rule coefficient M and the scoring coefficient K.

[0129] The processing method for the duration is as follows: a duration threshold t is set, for example, 15s, and each time a duration threshold t is reached is counted as one time, and the time less than a duration threshold is not counted in the number of times.

[0130] Among them, the pre-set evaluation rule coefficient M refers to assigning a corresponding score to each evaluation rule triggered, wherein prohibited behaviors receive negative scores and recommended behaviors receive positive scores. The scores of different evaluation rules can be set flexibly.

[0131] Customer service staff's service quality score = K*(100+∑N i M i ),

[0132] Where Ni is the number of times the i-th evaluation rule is triggered; Mi is the score of the i-th evaluation rule.

[0133] The number of occurrences L of high-frequency keywords is counted, and the high-frequency keywords are sorted according to the number of occurrences and sent to the backend.

[0134] The time when each high-frequency keyword appears is used as the third timestamp, customer emotion information of a preset time period is intercepted after the third timestamp, and changes in customer emotion during the time period are analyzed. According to the changes in customer emotion, it is determined whether the high-frequency keyword is a valid keyword or an invalid keyword, and the judgment result is sent to the quality feedback unit.

[0135] The quality feedback unit is used to count the service quality scores of customer service personnel, display the high-frequency keywords sent by the quality judgment unit and the changes in customer emotions after the high-frequency keywords appear.

[0136] A method for improving customer service quality management based on artificial intelligence, comprising the following steps:

[0137] First, the information collection unit is started to obtain the customer's video information and voice signal, and the customer service staff's video information, voice signal and electromyographic signal;

[0138] Then, the signal processing unit receives the voice, video and bioelectric signals collected by the signal collection unit, performs filtering and noise reduction on the above signals respectively and then outputs them to the emotion recognition unit;

[0139] Afterwards, the emotion recognition unit acquires and outputs the customer emotion information in time series, the customer service staff emotion information in time series, and the key information in the customer service staff video information and voice signal;

[0140] Afterwards, the quality judgment unit scores the service quality of the customer service personnel, determines whether the high-frequency keywords are valid keywords or invalid keywords according to the changes in customer emotions, and sends the score and judgment results to the quality feedback unit;

[0141] Finally, the quality feedback unit counts the service quality scores of the customer service staff, displays the high-frequency keywords sent by the quality judgment unit and the changes in customer emotions after the high-frequency keywords appear.

[0142] When the video acquisition device for acquiring the customer video information and the customer service staff video information is the same device, the signal processing unit firstly segments and extracts the customer information and the customer service staff information in the video information.

[0143] When the voice signal collecting device for collecting the customer voice signal and the customer service staff voice signal is the same device, the signal processing unit first divides and extracts the customer information and the customer service staff information in the voice signal.

[0144] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with this profession can make some changes or modifications to equivalent embodiments of equivalent changes using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A customer service quality management improvement system based on artificial intelligence, comprising: Information collection unit, signal processing unit, emotion recognition unit, quality judgment unit and quality feedback unit; The information collection unit includes a customer information collection module and a customer service staff information collection module, wherein the customer information collection module includes a first video collection device and a first voice signal collection device, and the customer service staff information collection module includes a second video collection device, a second voice signal collection device and an electromyographic signal collection device; The signal processing unit receives the voice, video and bioelectric signals collected by the signal collection unit, performs filtering and noise reduction on the above signals respectively and then outputs them to the emotion recognition unit; The bioelectric signal refers to the electromyographic signal collected by the electromyographic signal acquisition device; The emotion recognition unit includes a customer emotion recognition module and a customer service staff emotion recognition module; The customer emotion recognition module inputs the customer video information and voice signal collected by the first video acquisition device and the first voice signal acquisition device and sent by the signal processing unit into a customer emotion recognition model pre-established based on a convolutional neural network, and outputs the customer emotion information in time series; The customer service staff emotion recognition module inputs the customer service staff video information, voice signal and electromyographic signal collected by the second video acquisition device, the second voice signal acquisition device and the electromyographic signal acquisition device sent by the signal processing unit into the customer service staff emotion recognition model pre-established based on the convolutional neural network, and outputs the customer service staff emotion information in time series; The quality judgment unit extracts the first timestamp when the customer's emotions fluctuate violently, wherein the violent emotion fluctuation of the customer refers to the time point or time period when the signal amplitude in the customer's video information or voice signal is greater than the preset threshold after the signals are filtered and denoised; the quality judgment unit extracts the second timestamp when the customer service staff's emotions fluctuate violently, wherein the violent emotion fluctuation of the customer service staff refers to the time point or time period when the amplitude in the electromyographic signal is greater than the preset threshold after the signals are filtered and denoised; the time when the first timestamp or the second timestamp is located is used as the time reference to extend forward and backward by a certain length of time to intercept the emotional information of the customer and the customer service staff; when the timestamp is a time period, the preset time length is expanded before the starting point of the time period and the preset time length is expanded toward the end point of the time period; the customer's emotional changes are judged by the intercepted customer emotional information, and the The customer emotion changes are divided into positive changes and negative changes; the scoring coefficient K is set according to the customer emotion changes; the number of times N and high-frequency keywords triggered by the customer service staff during the quality inspection period are counted; the service quality score of the customer service staff during the quality inspection period is determined according to the number of times the judgment rule is triggered, the pre-set judgment rule coefficient M and the scoring coefficient K; the number of occurrences L of high-frequency keywords are counted, and the high-frequency keywords are sorted according to the number of occurrences and sent to the backend; the time when each high-frequency keyword appears is used as the third timestamp, and the customer emotion information of a preset time length is intercepted after the third timestamp, and the customer emotion changes during the time period are analyzed, and the high-frequency keywords are determined to be valid keywords or invalid keywords according to the customer emotion changes, and the judgment results are sent to the quality feedback unit; The quality feedback unit is used to count the service quality scores of customer service personnel, display the high-frequency keywords sent by the quality judgment unit and the changes in customer emotions after the high-frequency keywords appear; The customer service personnel information collection module also includes an electroencephalogram signal collection device, and the electroencephalogram signal collection device is a head-mounted electroencephalogram signal collection device; The bioelectric signal also includes an electroencephalogram signal from an electroencephalogram signal acquisition device; When the quality inspection system is used for real-time monitoring of customer service quality, it only extends a certain length of time before and after the time base to intercept the audio signal of the video signal of the customer service staff's emotional information; Positive change refers to the customer's change from negative emotion to positive emotion or from positive emotion to further positive emotion, and negative change refers to the customer's change from positive emotion to negative emotion or from negative emotion to further negative emotion; The customer service staff emotion recognition module also extracts and saves key information from the customer service staff video information and voice signals collected by the second video acquisition device and the second voice signal acquisition device and sent by the signal processing unit, wherein the key information includes expressions of dissatisfaction, anger, and annoyance of the customer service staff; The customer service staff emotion recognition module also performs voiceprint recognition based on the customer service staff's voice signals to determine and record the customer service staff's identity information; The voice signal collection device for collecting the customer voice signal and the customer service staff voice signal is the same device, and the signal processing unit also divides and extracts the customer information and the customer service staff information in the voice signal; By introducing bioelectric signals as a reference in the process of collecting emotional information from customer service personnel, information concealment caused by customer service personnel controlling their expressions and voices can be avoided.

2. A method for improving customer service quality management based on artificial intelligence, which uses the customer service quality management improvement system based on artificial intelligence as described in claim 1, and comprises the following steps: first, starting the information acquisition unit to obtain customer video information and voice signals, and obtaining customer service staff video information, voice signals and electromyographic signals; then, the signal processing unit receives the voice, video and bioelectric signals collected by the signal acquisition unit, filters and reduces noise on the above signals respectively, and then outputs them to the emotion recognition unit; thereafter, the emotion recognition unit obtains and outputs the customer emotion information in time sequence, the customer service staff emotion information in time sequence, and the key information in the customer service staff video information and voice signals; then, the quality judgment unit scores the service quality of the customer service staff, determines whether the high-frequency keywords are valid keywords or invalid keywords according to the changes in customer emotions, and sends the scores and judgment results to the quality feedback unit; finally, the quality feedback unit counts the service quality scores of the customer service staff, and displays the high-frequency keywords sent by the quality judgment unit and the changes in customer emotions after the high-frequency keywords appear.

Images