Artificial intelligence-based customer portrait method, device, equipment and medium

Abstract

The embodiment of the application discloses a customer service portrait method, device, equipment and medium based on artificial intelligence, wherein the method comprises: the voice customer service data is preprocessed to obtain first customer service data; the word use portrait is determined according to the word use standard index corresponding to the dialogue theme and the first customer service data, the response speed portrait is determined according to the time standard index corresponding to the dialogue theme and the first customer service data, and the reply specification portrait is determined according to the reply specification standard index corresponding to the dialogue theme and the first customer service data; the word use portrait, the response speed portrait and the reply specification portrait are fused to obtain a voice consultation service portrait; and the target customer service portrait is determined according to the voice consultation service portrait. Therefore, the word use, the response speed and the reply specification are comprehensively considered when the customer service portrait is performed, and the accuracy of the customer service portrait is improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a customer service profiling method, apparatus, device, and medium based on artificial intelligence. Background Technology

[0002] To improve conversion rates or after-sales service quality, customer service representatives typically provide consultation services, mostly through voice communication. However, creating a customer service profile by comparing sentences in the replies with standard sentences neglects other aspects of the consultation service, resulting in a one-dimensional approach and reduced accuracy. Summary of the Invention

[0003] Based on this, it is necessary to address the technical problem that the current method of comparing and scoring sentences in customer service replies with standard sentences to obtain customer service profiles does not consider other aspects of the consultation service, resulting in a single factor being considered when creating the profile and reducing the accuracy of the customer service profile. Therefore, an artificial intelligence-based customer service profiling method, device, equipment, and medium are proposed.

[0004] This application also proposes an artificial intelligence-based customer service profiling method, the method comprising:

[0005] Acquire voice customer service data and the corresponding dialogue topics, wherein the voice customer service data is the data of a target customer service representative providing voice consultation services to a target customer;

[0006] The voice customer service data is preprocessed to obtain the first customer service data;

[0007] A word profile is determined based on the word standard indicators corresponding to the dialogue topic and the first customer service data; a response speed profile is determined based on the time standard indicators corresponding to the dialogue topic and the first customer service data; and a response standard profile is determined based on the response standard indicators corresponding to the dialogue topic and the first customer service data.

[0008] The vocabulary profile, the response speed profile, and the response standard profile are fused together to obtain the voice consultation service profile.

[0009] Based on the voice consultation service profile, the target customer service profile is determined.

[0010] Further, the step of preprocessing the voice customer service data to obtain the first customer service data includes:

[0011] Collect raw dialogue voice and dialogue time data from the aforementioned voice customer service data;

[0012] The original dialogue speech is converted into Mandarin dialogue text using a speech conversion model corresponding to the language habit category of the target customer service representative.

[0013] A preset text correction model is used to perform text correction on the Mandarin dialogue text;

[0014] The Mandarin dialogue text is aligned using a preset entity alignment model.

[0015] The dialogue time data and the Mandarin dialogue text are used as the first customer service data.

[0016] Furthermore, the step of determining the word profile based on the word usage standard indicators corresponding to the dialogue topic and the first customer service data includes:

[0017] The Mandarin dialogue text in the first customer service data is segmented into words to obtain a phrase set;

[0018] The following methods are used to obtain a set of matching keywords: First, a set of matching keywords is obtained by searching for keywords in a preset keyword dictionary that match phrases in the phrase set. Second, a set of matching modal particles is obtained by searching for modal particles in a preset modal particle dictionary that match phrases in the phrase set. Third, a set of matching politeness particles is obtained by searching for politeness particles in a preset politeness particle dictionary that match phrases in the phrase set. Fourth, a set of matching care particles is obtained by searching for care particles in a preset care particle dictionary that match phrases in the phrase set. Finally, a set of matching professional particles is obtained by searching for professional particles in a professional particle dictionary corresponding to the dialogue topic that match phrases in the phrase set.

[0019] The frequency and percentage of keywords are calculated based on the hit keyword set and the phrase set; the frequency and percentage of modal particles are calculated based on the hit tone particle set and the phrase set; the frequency and percentage of polite words are calculated based on the hit polite word set and the phrase set; the frequency and percentage of caring words are calculated based on the hit caring word set and the phrase set; and the frequency and percentage of professional words are calculated based on the hit professional word set and the phrase set.

[0020] The word profile is determined based on the hit keyword set, the frequency of the keywords, the percentage of the keywords, the frequency of the interjections, the percentage of the interjections, the frequency of the polite words, the percentage of the polite words, the frequency of the caring words, the percentage of the caring words, the frequency of the professional words, the percentage of the professional words, and the word standard indicators corresponding to the dialogue topic.

[0021] The step of determining the response specification profile based on the response specification standard indicators corresponding to the dialogue topic and the first customer service data includes:

[0022] Using the standardized evaluation model corresponding to the dialogue topic, the customer service response text in the Mandarin dialogue text in the first customer service data is evaluated in a standardized manner to obtain the standardized evaluation result;

[0023] Using the professional evaluation model corresponding to the dialogue topic, the customer service response text in the Mandarin dialogue text in the first customer service data is professionally evaluated to obtain the professional evaluation result;

[0024] The response standard profile is determined based on the standardized evaluation results, the professional evaluation results, and the response standard indicators corresponding to the dialogue topic.

[0025] Furthermore, the step of determining the target customer service profile based on the voice consultation service profile includes:

[0026] Determine whether video customer service data exists, wherein the video customer service data is the data of the target customer service representative providing video consultation services to the target customer;

[0027] If it does not exist, the voice consultation service profile will be used as the target customer service profile.

[0028] If it exists, a profile is created based on the video customer service data to obtain a video consultation service profile. The voice consultation service profile and the video consultation service profile are then merged to obtain the target customer service profile.

[0029] Furthermore, the step of creating a profile based on the video customer service data to obtain a video consultation service profile includes:

[0030] The second customer service data is obtained by extracting customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data from the video customer service data.

[0031] The response time and speech rate profile is determined based on the standard indicators of response time and speech rate corresponding to the dialogue topic and the second customer service data.

[0032] A motion profile is determined based on the action standard indicators corresponding to the dialogue topic and the second customer service data;

[0033] A service attitude profile is determined based on the service attitude standard indicators corresponding to the dialogue topic and the second customer service data.

[0034] The response time and speech rate profile, the action profile, and the service attitude profile are fused together to obtain the video consultation service profile.

[0035] Furthermore, the step of extracting customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data from the video customer service data to obtain the second customer service data includes:

[0036] A preset tone and attitude evaluation model is used to evaluate the tone and attitude data of customer service videos in the video customer service data.

[0037] A preset customer service action evaluation model is used to evaluate the customer service action data in the video customer service data.

[0038] A preset enthusiasm assessment model is used to evaluate the enthusiasm level of the customer service videos in the video customer service data.

[0039] A preset response speed evaluation model is used to evaluate the response speed data of the customer service videos in the video customer service data.

[0040] A preset response time evaluation model is used to extract the response time data from the customer service videos in the video customer service data.

[0041] A preset communication motivation assessment model is used to evaluate the communication motivation data of the customer service videos in the video customer service data.

[0042] The customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data are used as the second customer service data.

[0043] Furthermore, the step of using the voice consultation service profile as the target customer service profile includes:

[0044] If the preset order database contains order data corresponding to the voice customer service data, then the voice consultation service profile is multiplied by a preset enhancement coefficient to obtain the target customer service profile; otherwise, the voice consultation service profile is used as the target customer service profile, wherein the enhancement coefficient is greater than 1.

[0045] The step of fusing the voice consultation service profile and the video consultation service profile to obtain the target customer service profile includes:

[0046] The voice consultation service profile and the video consultation service profile are weighted and summed to obtain a comprehensive profile;

[0047] If the order database contains order data corresponding to the voice customer service data, then the comprehensive profile is multiplied by the enhancement coefficient to obtain the target customer service profile; otherwise, the comprehensive profile is used as the target customer service profile.

[0048] This application also proposes an artificial intelligence-based customer service profiling device, the device comprising:

[0049] The data acquisition module is used to acquire voice customer service data and the dialogue topics corresponding to the voice customer service data, wherein the voice customer service data is the data of the target customer service providing voice consultation services to the target customer;

[0050] The first customer service data determination module is used to preprocess the voice customer service data to obtain the first customer service data;

[0051] The single-indicator profiling module is used to determine a word profiling based on the word standard indicator corresponding to the dialogue topic and the first customer service data, to determine a response speed profiling based on the time standard indicator corresponding to the dialogue topic and the first customer service data, and to determine a response standard profiling based on the response standard indicator corresponding to the dialogue topic and the first customer service data.

[0052] The voice consultation service profile determination module is used to fuse the word usage profile, the response speed profile, and the response standard profile to obtain the voice consultation service profile.

[0053] The target customer service profile determination module is used to determine the target customer service profile based on the voice consultation service profile.

[0054] A computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the following steps:

[0055] Acquire voice customer service data and the corresponding dialogue topics, wherein the voice customer service data is the data of a target customer service representative providing voice consultation services to a target customer;

[0056] The voice customer service data is preprocessed to obtain the first customer service data;

[0057] A word profile is determined based on the word standard indicators corresponding to the dialogue topic and the first customer service data; a response speed profile is determined based on the time standard indicators corresponding to the dialogue topic and the first customer service data; and a response standard profile is determined based on the response standard indicators corresponding to the dialogue topic and the first customer service data.

[0058] The vocabulary profile, the response speed profile, and the response standard profile are fused together to obtain the voice consultation service profile.

[0059] Based on the voice consultation service profile, the target customer service profile is determined.

[0060] A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the following steps:

[0061] Acquire voice customer service data and the corresponding dialogue topics, wherein the voice customer service data is the data of a target customer service representative providing voice consultation services to a target customer;

[0062] The voice customer service data is preprocessed to obtain the first customer service data;

[0063] A word profile is determined based on the word standard indicators corresponding to the dialogue topic and the first customer service data; a response speed profile is determined based on the time standard indicators corresponding to the dialogue topic and the first customer service data; and a response standard profile is determined based on the response standard indicators corresponding to the dialogue topic and the first customer service data.

[0064] The vocabulary profile, the response speed profile, and the response standard profile are fused together to obtain the voice consultation service profile.

[0065] Based on the voice consultation service profile, the target customer service profile is determined.

[0066] The AI-based customer service profiling method of this application determines a word profiling based on the standard indicators of word usage corresponding to the dialogue topic and the first customer service data; determines a response speed profiling based on the standard indicators of time usage corresponding to the dialogue topic and the first customer service data; and determines a response standard profiling based on the standard indicators of response format corresponding to the dialogue topic and the first customer service data. The word profiling, response speed profiling, and response standard profiling are then fused to obtain a voice consultation service profiling. This comprehensively considers word usage, response speed, and response format in customer service profiling, improving the accuracy of the profiling. Furthermore, by using standard indicators of word usage, time usage, and response format corresponding to the dialogue topic, the method employs indicators that align with the dialogue topic, further enhancing the accuracy of the customer service profiling. Attached Figure Description

[0067] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0068] in:

[0069] Figure 1 This is a flowchart illustrating an AI-based customer service profiling method in one embodiment.

[0070] Figure 2 This is a schematic diagram of the entire process of an AI-based customer service profiling method in one embodiment;

[0071] Figure 3 This is a structural block diagram of an AI-based customer service profiling device in one embodiment;

[0072] Figure 4 This is a structural block diagram of a computer device in one embodiment. Detailed Implementation

[0073] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0074] like Figure 1 As shown, in one embodiment, an AI-based customer service profiling method is provided. This method can be applied to both terminals and servers; this embodiment uses terminal application as an example. The AI-based customer service profiling method specifically includes the following steps:

[0075] S1: Obtain voice customer service data and the corresponding dialogue topic, wherein the voice customer service data is the data of the target customer service representative providing voice consultation services to the target customer;

[0076] Specifically, it can obtain user-inputted voice customer service data and the corresponding dialogue topics, or it can obtain voice customer service data and the corresponding dialogue topics from a database, or it can obtain voice customer service data and the corresponding dialogue topics from a third-party application.

[0077] Voice customer service data refers to all data generated when a target customer service representative provides voice consultation services to a target customer.

[0078] The topic of the conversation is the purpose or desired outcome of the conversation. For example, the topic of the conversation might be consulting about product A.

[0079] S2: Preprocess the voice customer service data to obtain the first customer service data;

[0080] Specifically, the voice customer service data is standardized, and the standardized data is used as the first customer service data. The standardization includes, but is not limited to, text correction and entity alignment.

[0081] S3: Determine the word usage profile based on the word usage standard indicators corresponding to the dialogue topic and the first customer service data; determine the response speed profile based on the time standard indicators corresponding to the dialogue topic and the first customer service data; and determine the response standard profile based on the response standard indicators corresponding to the dialogue topic and the first customer service data.

[0082] Specifically, from the Mandarin dialogue text of the first customer service data, the required parameter values ​​for the standard vocabulary indicators corresponding to the dialogue topic are extracted. These extracted parameter values ​​are then substituted into the standard vocabulary indicators corresponding to the dialogue topic to obtain a vocabulary profile. From the dialogue time data of the first customer service data, the required parameter values ​​for the standard time indicators corresponding to the dialogue topic are extracted. These extracted parameter values ​​are then substituted into the standard time indicators corresponding to the dialogue topic to obtain a response speed profile. From the Mandarin dialogue text of the first customer service data, the required parameter values ​​for the standard response specifications indicators corresponding to the dialogue topic are extracted. These extracted parameter values ​​are then substituted into the standard response specifications indicators corresponding to the dialogue topic to obtain a response specification profile.

[0083] Optionally, the word usage profile, the response speed profile, and the response standardization profile are all values ​​from 0 to k; they can be 0, k, or values ​​between 0 and k. Optionally, k is set to 100.

[0084] S4: The word usage profile, the response speed profile, and the response standard profile are fused together to obtain the voice consultation service profile;

[0085] Specifically, the word usage profile, the response speed profile, and the response standard profile are weighted and summed, and the data obtained by weighted summation is used as the voice consultation service profile.

[0086] Optionally, the profile for real-time voice consultation service is a value from 0 to k.

[0087] S5: Determine the target customer service profile based on the voice consultation service profile.

[0088] Specifically, the voice consultation service profile is used as the target customer service profile.

[0089] This embodiment determines a word usage profile based on the standard indicators of word usage corresponding to the dialogue topic and the first customer service data; it determines a response speed profile based on the standard indicators of time corresponding to the dialogue topic and the first customer service data; and it determines a response standard profile based on the standard indicators of response specifications corresponding to the dialogue topic and the first customer service data. The word usage profile, response speed profile, and response standard profile are then fused to obtain a voice consultation service profile. This comprehensively considers word usage, response speed, and response specifications when creating a customer service profile, improving its accuracy. Furthermore, by using the standard indicators of word usage, time, and response specifications corresponding to the dialogue topic, the use of indicators consistent with the dialogue topic further enhances the accuracy of the customer service profile.

[0090] In one embodiment, the step of preprocessing the voice customer service data to obtain the first customer service data includes:

[0091] S21: Collect raw dialogue voice and dialogue time data from the voice customer service data;

[0092] Specifically, the dialogue voice in the voice customer service data is used as the original dialogue voice; the time record table in the voice customer service data is used as the dialogue time data.

[0093] The conversation time data includes: the conversation roles, the start time of the conversation, and the end time of the conversation. The conversation roles are either target customer or target customer service representative.

[0094] S22: Using the speech conversion model corresponding to the language habit category of the target customer service, the original dialogue speech is converted into Mandarin dialogue text;

[0095] Specifically, due to the different upbringing of customer service personnel, the original dialogue text may contain local expression habits. Since the basic information of the target customer service personnel is determined in advance, the language habit category corresponding to the target customer service personnel can be determined in advance. The speech conversion model corresponding to the language habit category of the target customer service personnel can be used to convert the original dialogue speech into Mandarin dialogue text.

[0096] The speech conversion model is based on ASR (Automatic Speech Recognition) technology.

[0097] S23: Use a preset text correction model to perform text correction on the Mandarin dialogue text;

[0098] Specifically, a preset text correction model is used to correct errors in the Mandarin dialogue text.

[0099] Text correction models are models trained based on neural networks. The network structure and training methods of text correction models can be selected from existing technologies, and will not be elaborated here.

[0100] S24: Using a preset entity alignment model, perform entity alignment on the Mandarin dialogue text;

[0101] Specifically, a preset entity alignment model is used to align entities in the Mandarin dialogue text in order to standardize entity representations in the Mandarin dialogue text.

[0102] Entity alignment models are based on neural network training. The network structure and training methods for entity alignment models can be selected from existing technologies, and will not be elaborated here.

[0103] S25: Use the dialogue time data and the Mandarin dialogue text as the first customer service data.

[0104] This embodiment improves the accuracy of Mandarin dialogue text by using a speech conversion model corresponding to the language habit category of the target customer service representative. By performing text correction and entity alignment on the Mandarin dialogue text, the text standardization of the Mandarin dialogue text is improved, further enhancing the accuracy of the customer service profile.

[0105] In one embodiment, the step of determining the word profile based on the word usage standard indicators corresponding to the dialogue topic and the first customer service data includes:

[0106] S31: Segment the Mandarin dialogue text in the first customer service data to obtain a phrase set;

[0107] Specifically, the Mandarin dialogue text in the first customer service data is segmented into words, and the resulting phrases are used as a phrase set.

[0108] S32: Search for keywords that are the same as phrases in the phrase set from a preset keyword dictionary to obtain a set of matched keywords; search for modal particles that are the same as phrases in the phrase set from a preset modal particle dictionary to obtain a set of matched modal particles; search for polite words that are the same as phrases in the phrase set from a preset politeness word dictionary to obtain a set of matched politeness words; search for caring words that are the same as phrases in the phrase set from a preset care word dictionary to obtain a set of matched care words; search for professional words that are the same as phrases in the phrase set from a professional word dictionary corresponding to the dialogue topic to obtain a set of matched professional words.

[0109] Specifically, keywords are words related to the content of the consultation, such as product names. If the topic of the conversation is a consultation about cervical cancer-related assessments, then the range of keywords includes: HPV (human papillomavirus) and TCT (Thinker's liquid-based cytology test).

[0110] Modal particles are function words that express tone and are often used at the end of a sentence or at pauses in the middle of a sentence to express various tones.

[0111] Polite words, also known as polite expressions, refer to words used in verbal communication that convey respect and friendliness.

[0112] Words expressing care or concern are those that convey the meaning of helping, loving, or looking after someone.

[0113] The specialized terms in the dictionary corresponding to the dialogue topic can be industry-specific terms, etc.

[0114] S33: Calculate the frequency and percentage of keywords based on the hit keyword set and the phrase set; calculate the frequency and percentage of modal particles based on the hit tone particle set and the phrase set; calculate the frequency and percentage of polite words based on the hit polite word set and the phrase set; calculate the frequency and percentage of caring words based on the hit caring word set and the phrase set; calculate the frequency and percentage of professional words based on the hit professional word set and the phrase set.

[0115] Keyword count is the total number of times all keywords in the keyword set appear in the phrase set; keyword percentage is the keyword count divided by the number of phrases in the phrase set.

[0116] The frequency of modal particles is the total number of times all modal particles in the modal particle set appear in the phrase set; the modal particle percentage is the frequency of modal particles divided by the number of phrases in the phrase set.

[0117] The polite word count is the total number of times all polite words in the polite word set appear in the phrase set; the polite word percentage is the polite word count divided by the number of phrases in the phrase set.

[0118] The number of care words is the total number of times all care words in the care word set appear in the phrase set; the proportion of care words is the number of care words divided by the number of phrases in the phrase set.

[0119] The term count is the total number of times all professional terms in the term set appear in the phrase set; the term percentage is the term count divided by the number of phrases in the phrase set.

[0120] S34: Determine the word profile based on the hit keyword set, the keyword frequency, the keyword percentage, the frequency of interjections, the percentage of interjections, the frequency of polite words, the percentage of polite words, the frequency of caring words, the percentage of caring words, the frequency of professional words, the percentage of professional words, and the word standard indicators corresponding to the dialogue topic;

[0121] The terminology standards include: a set of essential keywords, essential keyword evaluation metrics, keyword evaluation metrics, evaluation metrics for modal particles, evaluation metrics for polite words, evaluation metrics for caring words, evaluation metrics for professional terms, and a weight set. The essential keyword set includes one or more essential keywords. The essential keyword evaluation metrics include: inclusion ratio and score. The keyword evaluation metrics include: frequency, percentage, and score. The modal particle evaluation metrics include: frequency, percentage, and score. The polite word evaluation metrics include: frequency, percentage, and score. The caring word evaluation metrics include: frequency, percentage, and score. The professional term evaluation metrics include: frequency, percentage, and score. The weight set includes: essential keyword weight, keyword weight, modal particle weight, polite word weight, caring word weight, and professional term weight.

[0122] Specifically, the first score is obtained by dividing the number of essential keywords in the hit keyword set by the total number of keywords in the hit keyword set; the second score is obtained by querying the keyword frequency and keyword proportion from the keyword evaluation index; the third score is obtained by querying the frequency and proportion of modal words from the modal word evaluation index; the fourth score is obtained by querying the frequency and proportion of polite words from the polite word evaluation index; the fifth score is obtained by querying the frequency and proportion of caring words from the caring word evaluation index; and the sixth score is obtained by querying the frequency and proportion of professional words from the professional word evaluation index. The first, second, third, fourth, fifth, and sixth scores are weighted and summed according to the weight set in the word usage standard index to obtain the word usage profile.

[0123] The step of determining the response specification profile based on the response specification standard indicators corresponding to the dialogue topic and the first customer service data includes:

[0124] S35: Using the standardized evaluation model corresponding to the dialogue topic, perform standardized evaluation on the customer service reply text in the Mandarin dialogue text in the first customer service data to obtain the standardized evaluation result;

[0125] Specifically, the customer service reply text in the Mandarin dialogue text in the first customer service data is input into the standardized evaluation model corresponding to the dialogue topic for standardized evaluation to obtain the standardized evaluation result.

[0126] Standardization refers to conforming to the business process corresponding to the topic of the dialogue, such as the inspection process.

[0127] A standardized evaluation model is used to compare and score the process corresponding to the Mandarin dialogue text with the business process corresponding to the dialogue topic. The standardized evaluation model is a multi-classification model trained on a neural network, with the classification label being the conformity score. The network structure and training method of the multi-classification model can be selected from existing technologies and will not be elaborated here.

[0128] S36: Using the professional evaluation model corresponding to the dialogue topic, perform a professional evaluation on the customer service reply text in the Mandarin dialogue text in the first customer service data to obtain a professional evaluation result;

[0129] Specifically, the customer service reply text in the Mandarin dialogue text in the first customer service data is input into the professional evaluation model corresponding to the dialogue topic for professional evaluation to obtain the professional evaluation result.

[0130] A professional evaluation model is used to compare and score customer service responses in the Mandarin dialogue text with standard sentences. The professional evaluation model is a multi-classification model trained on a neural network, with the classification label being the score.

[0131] S37: Determine the response standard profile based on the standardized evaluation results, the professional evaluation results, and the response standard indicators corresponding to the dialogue topic.

[0132] Specifically, based on the standardized evaluation results and the professional evaluation results, the response standardization profile is obtained by weighted summation according to the response standardization indicators corresponding to the dialogue topic.

[0133] This embodiment determines the word usage profile based on the hit keyword set, the frequency of keywords, the percentage of keywords, the frequency of interjections, the percentage of interjections, the frequency of polite words, the percentage of polite words, the frequency of caring words, the frequency of professional words, the percentage of professional words, and the word usage standard indicators corresponding to the dialogue topic. It determines the response standard profile based on the standardization evaluation results, the professionalization evaluation results, and the response standard indicators corresponding to the dialogue topic. By comprehensively considering keywords, interjections, polite words, caring words, professional words, standardization, and professionalization, the accuracy of customer service profiles is improved.

[0134] In one embodiment, the step of determining the target customer service profile based on the voice consultation service profile includes:

[0135] S51: Determine whether video customer service data exists, wherein the video customer service data is the data of the target customer service representative providing video consultation services to the target customer;

[0136] Specifically, based on the personnel identifier corresponding to the target customer service representative and the personnel identifier corresponding to the target customer, it is determined whether video customer service data exists in the database.

[0137] S52: If it does not exist, then the voice consultation service profile shall be used as the target customer service profile;

[0138] Specifically, if it does not exist, that is, if there is no video customer service data, it means that the target customer service representative did not provide video consultation services to the target customer. Therefore, the voice consultation service profile is used as the target customer service profile.

[0139] S53: If it exists, then a profile is created based on the video customer service data to obtain a video consultation service profile. The voice consultation service profile and the video consultation service profile are then merged to obtain the target customer service profile.

[0140] Specifically, if video customer service data exists, it means that the target customer service representative has provided video consultation services to the target customer. Therefore, a preset video customer service profiling method is used to create a profile based on the video customer service data to obtain a video consultation service profile. The voice consultation service profile and the video consultation service profile are then weighted and summed to obtain the target customer service profile.

[0141] In this embodiment, when video customer service data is unavailable, the voice consultation service profile is used as the target customer service profile. When video customer service data is available, the voice consultation service profile and the video consultation service profile are fused together to form the target customer service profile, thereby further improving the accuracy of the customer service profile.

[0142] In one embodiment, the step of creating a profile based on the video customer service data to obtain a video consultation service profile includes:

[0143] S5311: Extract customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data from the video customer service data to obtain second customer service data;

[0144] Specifically, customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data are extracted from the video customer service data, and the extracted customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data are used as second customer service data.

[0145] S5312: Determine the response time and speech rate profile based on the standard indicators of response time and speech rate corresponding to the dialogue topic and the second customer service data;

[0146] Specifically, the required parameter values ​​for the standard indicators of response time and speech rate corresponding to the dialogue topic are extracted from the response speech rate data and response time data of the second customer service data. The extracted parameter values ​​are then substituted into the standard indicators of response time and speech rate corresponding to the dialogue topic to obtain the response time and speech rate profile.

[0147] S5313: Determine the action profile based on the action standard indicators corresponding to the dialogue topic and the second customer service data;

[0148] Specifically, the required parameter values ​​for the action standard indicators corresponding to the dialogue topic are extracted from the customer service action data of the second customer service data. The extracted parameter values ​​are then substituted into the action standard indicators corresponding to the dialogue topic to obtain the action profile.

[0149] S5314: Determine a service attitude profile based on the service attitude standard indicators corresponding to the dialogue topic and the second customer service data;

[0150] Specifically, the required parameter values ​​for the service attitude standard indicators corresponding to the dialogue topic are extracted from the customer service tone and attitude data, customer service enthusiasm data, and communication initiative data of the second customer service data. The extracted parameter values ​​are then substituted into the service attitude standard indicators corresponding to the dialogue topic to obtain a service attitude profile.

[0151] S5315: The response time and speech rate profile, the action profile, and the service attitude profile are fused together to obtain the video consultation service profile.

[0152] Specifically, the video consultation service profile is obtained by weighting and summing the response time and speech rate profile, the action profile, and the service attitude profile.

[0153] This embodiment comprehensively considers customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data when creating customer service profiles, thereby improving the accuracy of customer service profiles.

[0154] In one embodiment, the step of extracting customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data from the video customer service data to obtain second customer service data includes:

[0155] S53111: Using a preset tone and attitude evaluation model, the tone and attitude data of the customer service videos in the video customer service data are evaluated.

[0156] Specifically, the customer service videos in the video customer service data are segmented according to a preset time interval. Each segmented sub-video is input into a preset tone and attitude evaluation model for tone classification prediction and attitude classification prediction. The tone results and attitude results obtained from the classification prediction are used as the customer service tone and attitude data.

[0157] The tone and attitude assessment model is a multi-classification model based on neural network training. The classification labels can be tone level or attitude level.

[0158] Tone refers to one of the important techniques of customer service broadcasting language expression, which is the vocal form of specific sentences under the control of certain specific thoughts and feelings.

[0159] Attitude refers to the stable psychological inclination that customer service representatives hold towards customers.

[0160] S53112: Using a preset customer service action evaluation model, evaluate the customer service action data of the customer service video in the video customer service data;

[0161] Specifically, the customer service videos in the video customer service data are segmented according to a preset time interval. Each segmented sub-video is input into a preset customer service action evaluation model to classify and predict customer service actions. The classification results are used as the customer service action data.

[0162] The customer service action evaluation model is a multi-classification model trained on a neural network. The classification labels can be either arm action names or head action names.

[0163] S53113: Using a preset enthusiasm assessment model, evaluate the enthusiasm level of the customer service videos in the video customer service data;

[0164] Specifically, the customer service videos in the video customer service data are segmented according to a preset time interval. Each segmented sub-video is input into a preset enthusiasm level assessment model to classify and predict the enthusiasm level. The classification results are used as the customer service enthusiasm level data.

[0165] The enthusiasm assessment model is a multi-classification model trained on a neural network, and the classification label is the enthusiasm level.

[0166] S53114: Using a preset response speed evaluation model, the response speed data of the customer service video in the video customer service data is evaluated;

[0167] Specifically, the customer service videos in the video customer service data are segmented according to a preset time interval. Each segmented sub-video is input into a preset response speech rate evaluation model to classify and predict the response speech rate. The classification results are used as the response speech rate data.

[0168] The response speech rate assessment model is a multi-classification model trained on a neural network, with the classification label being the speech rate level.

[0169] S53115: Using a preset response time evaluation model, extract the response time data from the customer service video in the video customer service data;

[0170] Specifically, the customer service videos in the video customer service data are segmented according to a preset time interval. Each segmented sub-video is input into a preset response time evaluation model to classify and predict the response time. The classification results are used as the response time data.

[0171] The response time assessment model is a multi-classification model trained on a neural network, and the classification label is the response speed level.

[0172] S53116: Using a preset communication positivity assessment model, evaluate the communication positivity data of the customer service videos in the video customer service data;

[0173] Specifically, the customer service videos in the video customer service data are segmented according to a preset time interval. Each segmented sub-video is input into a preset communication enthusiasm assessment model to classify and predict the communication enthusiasm. The classification results are used as the communication enthusiasm data.

[0174] The communication motivation assessment model is a multi-classification model based on neural network training, and the classification label is the motivation level.

[0175] S53117: The customer service tone and attitude data, the customer service action data, the customer service enthusiasm data, the response speed data, the response time data, and the communication enthusiasm data are used as the second customer service data.

[0176] This embodiment uses a model trained on a neural network to extract customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data from the video customer service data, providing a foundation for creating customer service profiles based on these data.

[0177] In one embodiment, the step of using the voice consultation service profile as the target customer service profile includes:

[0178] S521: If there is order data corresponding to the voice customer service data in the preset order database, then the voice consultation service profile is multiplied by a preset enhancement coefficient to obtain the target customer service profile; otherwise, the voice consultation service profile is used as the target customer service profile, wherein the enhancement coefficient is greater than 1.

[0179] Specifically, if there is order data in the preset order database that corresponds to the voice customer service data, meaning the consultation has been completed and the target customer is relatively satisfied, then the voice consultation service profile is multiplied by a preset enhancement coefficient to obtain the target customer service profile; if there is no order data in the preset order database that corresponds to the voice customer service data, meaning the consultation has not been completed and the target customer's satisfaction cannot be assessed, then the voice consultation service profile is used as the target customer service profile.

[0180] The step of fusing the voice consultation service profile and the video consultation service profile to obtain the target customer service profile includes:

[0181] S5321: Perform a weighted summation of the voice consultation service profile and the video consultation service profile to obtain a comprehensive profile;

[0182] S5322: If the order database contains order data corresponding to the voice customer service data, then the comprehensive profile is multiplied by the enhancement coefficient to obtain the target customer service profile; otherwise, the comprehensive profile is used as the target customer service profile.

[0183] Specifically, if there is order data in the preset order database that corresponds to the voice customer service data, meaning the consultation has been completed, it means the target customer is relatively satisfied. Therefore, the comprehensive profile is multiplied by the enhancement coefficient to obtain the target customer service profile. If there is no order data in the preset order database that corresponds to the voice customer service data, meaning the consultation has not been completed, it means the target customer's satisfaction cannot be assessed. Therefore, the comprehensive profile is used as the target customer service profile.

[0184] In this embodiment, when order data corresponding to the voice customer service data exists, the voice consultation service profile enhanced by the enhancement coefficient is used as the target customer service profile, or the comprehensive profile enhanced by the enhancement coefficient is used as the target customer service profile, thereby improving the accuracy of the determined target customer service profile.

[0185] like Figure 2 As shown, in one embodiment, an artificial intelligence-based customer service profiling device is proposed, the device comprising:

[0186] The data acquisition module 801 is used to acquire voice customer service data and the dialogue topic corresponding to the voice customer service data, wherein the voice customer service data is the data of the target customer service providing voice consultation services to the target customer;

[0187] The first customer service data determination module 802 is used to preprocess the voice customer service data to obtain the first customer service data;

[0188] The single-indicator profiling module 803 is used to determine a word profiling based on the word standard indicator corresponding to the dialogue topic and the first customer service data, to determine a response speed profiling based on the time standard indicator corresponding to the dialogue topic and the first customer service data, and to determine a response standard profiling based on the response standard indicator corresponding to the dialogue topic and the first customer service data.

[0189] The voice consultation service profile determination module 804 is used to fuse the word usage profile, the response speed profile, and the response standard profile to obtain a voice consultation service profile.

[0190] The target customer service profile determination module 805 is used to determine the target customer service profile based on the voice consultation service profile.

[0191] This embodiment determines a word usage profile based on the standard indicators of word usage corresponding to the dialogue topic and the first customer service data; it determines a response speed profile based on the standard indicators of time corresponding to the dialogue topic and the first customer service data; and it determines a response standard profile based on the standard indicators of response specifications corresponding to the dialogue topic and the first customer service data. The word usage profile, response speed profile, and response standard profile are then fused to obtain a voice consultation service profile. This comprehensively considers word usage, response speed, and response specifications when creating a customer service profile, improving its accuracy. Furthermore, by using the standard indicators of word usage, time, and response specifications corresponding to the dialogue topic, the use of indicators consistent with the dialogue topic further enhances the accuracy of the customer service profile.

[0192] Figure 4 An internal structural diagram of a computer device in one embodiment is shown. This computer device can specifically be a terminal or a server. Figure 4As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and may also store a computer program. When executed by the processor, this computer program enables the processor to implement an AI-based customer service profiling method. The internal memory may also store a computer program, which, when executed by the processor, enables the processor to implement an AI-based customer service profiling method. Those skilled in the art will understand that… Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0193] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the following steps:

[0194] Acquire voice customer service data and the corresponding dialogue topics, wherein the voice customer service data is the data of a target customer service representative providing voice consultation services to a target customer;

[0195] The voice customer service data is preprocessed to obtain the first customer service data;

[0196] A word profile is determined based on the word standard indicators corresponding to the dialogue topic and the first customer service data; a response speed profile is determined based on the time standard indicators corresponding to the dialogue topic and the first customer service data; and a response standard profile is determined based on the response standard indicators corresponding to the dialogue topic and the first customer service data.

[0197] The vocabulary profile, the response speed profile, and the response standard profile are fused together to obtain the voice consultation service profile.

[0198] Based on the voice consultation service profile, the target customer service profile is determined.

[0199] This embodiment determines a word usage profile based on the standard indicators of word usage corresponding to the dialogue topic and the first customer service data; it determines a response speed profile based on the standard indicators of time corresponding to the dialogue topic and the first customer service data; and it determines a response standard profile based on the standard indicators of response specifications corresponding to the dialogue topic and the first customer service data. The word usage profile, response speed profile, and response standard profile are then fused to obtain a voice consultation service profile. This comprehensively considers word usage, response speed, and response specifications when creating a customer service profile, improving its accuracy. Furthermore, by using the standard indicators of word usage, time, and response specifications corresponding to the dialogue topic, the use of indicators consistent with the dialogue topic further enhances the accuracy of the customer service profile.

[0200] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, causes the processor to perform the following steps:

[0201] Acquire voice customer service data and the corresponding dialogue topics, wherein the voice customer service data is the data of a target customer service representative providing voice consultation services to a target customer;

[0202] The voice customer service data is preprocessed to obtain the first customer service data;

[0203] A word profile is determined based on the word standard indicators corresponding to the dialogue topic and the first customer service data; a response speed profile is determined based on the time standard indicators corresponding to the dialogue topic and the first customer service data; and a response standard profile is determined based on the response standard indicators corresponding to the dialogue topic and the first customer service data.

[0204] The vocabulary profile, the response speed profile, and the response standard profile are fused together to obtain the voice consultation service profile.

[0205] Based on the voice consultation service profile, the target customer service profile is determined.

[0206] This embodiment determines a word usage profile based on the standard indicators of word usage corresponding to the dialogue topic and the first customer service data; it determines a response speed profile based on the standard indicators of time corresponding to the dialogue topic and the first customer service data; and it determines a response standard profile based on the standard indicators of response specifications corresponding to the dialogue topic and the first customer service data. The word usage profile, response speed profile, and response standard profile are then fused to obtain a voice consultation service profile. This comprehensively considers word usage, response speed, and response specifications when creating a customer service profile, improving its accuracy. Furthermore, by using the standard indicators of word usage, time, and response specifications corresponding to the dialogue topic, the use of indicators consistent with the dialogue topic further enhances the accuracy of the customer service profile.

[0207] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0208] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0209] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A customer service profiling method based on artificial intelligence, the method comprising: Acquire voice customer service data and the corresponding dialogue topics, wherein the voice customer service data is the data of a target customer service representative providing voice consultation services to a target customer; The voice customer service data is preprocessed to obtain the first customer service data; A word profile is determined based on the word standard indicators corresponding to the dialogue topic and the first customer service data; a response speed profile is determined based on the time standard indicators corresponding to the dialogue topic and the first customer service data; and a response standard profile is determined based on the response standard indicators corresponding to the dialogue topic and the first customer service data. The vocabulary profile, the response speed profile, and the response standard profile are fused together to obtain the voice consultation service profile. Based on the voice consultation service profile, determine the target customer service profile; The step of determining the target customer service profile based on the voice consultation service profile includes: Determine whether video customer service data exists, wherein the video customer service data is the data of the target customer service representative providing video consultation services to the target customer; If it does not exist, the voice consultation service profile will be used as the target customer service profile. If it exists, a profile is created based on the video customer service data to obtain a video consultation service profile. The voice consultation service profile and the video consultation service profile are then merged to obtain the target customer service profile. The step of using the voice consultation service profile as the target customer service profile includes: If the preset order database contains order data corresponding to the voice customer service data, then the voice consultation service profile is multiplied by a preset enhancement coefficient to obtain the target customer service profile; otherwise, the voice consultation service profile is used as the target customer service profile, wherein the enhancement coefficient is greater than 1. The step of fusing the voice consultation service profile and the video consultation service profile to obtain the target customer service profile includes: The voice consultation service profile and the video consultation service profile are weighted and summed to obtain a comprehensive profile; If the order database contains order data corresponding to the voice customer service data, then the comprehensive profile is multiplied by the enhancement coefficient to obtain the target customer service profile; otherwise, the comprehensive profile is used as the target customer service profile.

2. The customer service profiling method based on artificial intelligence according to claim 1, characterized in that, The step of preprocessing the voice customer service data to obtain the first customer service data includes: Collect raw dialogue voice and dialogue time data from the aforementioned voice customer service data; The original dialogue speech is converted into Mandarin dialogue text using a speech conversion model corresponding to the language habit category of the target customer service representative. A preset text correction model is used to perform text correction on the Mandarin dialogue text; The Mandarin dialogue text is aligned using a preset entity alignment model. The dialogue time data and the Mandarin dialogue text are used as the first customer service data.

3. The customer service profiling method based on artificial intelligence according to claim 1, characterized in that, The step of determining the word profile based on the word usage standard indicators corresponding to the dialogue topic and the first customer service data includes: The Mandarin dialogue text in the first customer service data is segmented into words to obtain a phrase set; The following methods are used to obtain a set of matching keywords: First, a set of matching keywords is obtained by searching for keywords in a preset keyword dictionary that match phrases in the phrase set. Second, a set of matching modal particles is obtained by searching for modal particles in a preset modal particle dictionary that match phrases in the phrase set. Third, a set of matching politeness particles is obtained by searching for politeness particles in a preset politeness particle dictionary that match phrases in the phrase set. Fourth, a set of matching care particles is obtained by searching for care particles in a preset care particle dictionary that match phrases in the phrase set. Finally, a set of matching professional particles is obtained by searching for professional particles in a professional particle dictionary corresponding to the dialogue topic that match phrases in the phrase set. The frequency and percentage of keywords are calculated based on the hit keyword set and the phrase set; the frequency and percentage of modal particles are calculated based on the hit tone particle set and the phrase set; the frequency and percentage of polite words are calculated based on the hit polite word set and the phrase set; the frequency and percentage of caring words are calculated based on the hit caring word set and the phrase set; and the frequency and percentage of professional words are calculated based on the hit professional word set and the phrase set. The word profile is determined based on the hit keyword set, the frequency of the keywords, the percentage of the keywords, the frequency of the interjections, the percentage of the interjections, the frequency of the polite words, the percentage of the polite words, the frequency of the caring words, the percentage of the caring words, the frequency of the professional words, the percentage of the professional words, and the word standard indicators corresponding to the dialogue topic. The step of determining the response specification profile based on the response specification standard indicators corresponding to the dialogue topic and the first customer service data includes: Using the standardized evaluation model corresponding to the dialogue topic, the customer service response text in the Mandarin dialogue text in the first customer service data is evaluated in a standardized manner to obtain the standardized evaluation result; Using the professional evaluation model corresponding to the dialogue topic, the customer service response text in the Mandarin dialogue text in the first customer service data is professionally evaluated to obtain the professional evaluation result; The response standard profile is determined based on the standardized evaluation results, the professional evaluation results, and the response standard indicators corresponding to the dialogue topic.

4. The customer service profiling method based on artificial intelligence according to claim 1, characterized in that, The step of creating a profile based on the video customer service data to obtain a video consultation service profile includes: The second customer service data is obtained by extracting customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data from the video customer service data. The response time and speech rate profile is determined based on the standard indicators of response time and speech rate corresponding to the dialogue topic and the second customer service data. A motion profile is determined based on the action standard indicators corresponding to the dialogue topic and the second customer service data; A service attitude profile is determined based on the service attitude standard indicators corresponding to the dialogue topic and the second customer service data. The response time and speech rate profile, the action profile, and the service attitude profile are fused together to obtain the video consultation service profile.

5. The customer service profiling method based on artificial intelligence according to claim 4, characterized in that, The step of extracting customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data from the video customer service data to obtain the second customer service data includes: A preset tone and attitude evaluation model is used to evaluate the tone and attitude data of customer service videos in the video customer service data. A preset customer service action evaluation model is used to evaluate the customer service action data in the video customer service data. A preset enthusiasm assessment model is used to evaluate the enthusiasm level of the customer service videos in the video customer service data. A preset response speed evaluation model is used to evaluate the response speed data of the customer service videos in the video customer service data. A preset response time evaluation model is used to extract the response time data from the customer service videos in the video customer service data. A preset communication motivation assessment model is used to evaluate the communication motivation data of the customer service videos in the video customer service data. The customer service tone and attitude data, customer service action data, customer service enthusiasm data, response speed data, response time data, and communication initiative data are used as the second customer service data.

6. A customer service profiling device based on artificial intelligence, characterized in that, The apparatus for applying the AI-based customer service profiling method according to any one of claims 1 to 5, the apparatus comprising: The data acquisition module is used to acquire voice customer service data and the dialogue topics corresponding to the voice customer service data, wherein the voice customer service data is the data of the target customer service providing voice consultation services to the target customer; The first customer service data determination module is used to preprocess the voice customer service data to obtain the first customer service data; The single-indicator profiling module is used to determine a word profiling based on the word standard indicator corresponding to the dialogue topic and the first customer service data, to determine a response speed profiling based on the time standard indicator corresponding to the dialogue topic and the first customer service data, and to determine a response standard profiling based on the response standard indicator corresponding to the dialogue topic and the first customer service data. The voice consultation service profile determination module is used to fuse the word usage profile, the response speed profile, and the response standard profile to obtain the voice consultation service profile. The target customer service profile determination module is used to determine the target customer service profile based on the voice consultation service profile.

7. A computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to perform the steps of the method as claimed in any one of claims 1 to 5.

8. A computer device comprising a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as claimed in any one of claims 1 to 5.

Images