Artificial intelligence customer service system based on natural language processing

Abstract

The invention discloses an artificial intelligence customer service system based on natural language processing, and particularly relates to the technical field of natural language processing, the artificial intelligence customer service system comprises an input processing module, an intention collaboration module, an intention optimization module and an intention dynamic decision module, the intent collaborative module is used for calculating a context entropy value based on historical N rounds of dialogue feature distribution and a time decay weight, the intent collaborative module is used for outputting an emotion intensity signal through a multi-mode emotion recognition model according to input information and generating an intent correction vector in combination with a preset intent-emotion mapping matrix, and the intent optimization module is used for constructing and optimizing a collaborative loss function so as to improve the intent correction vector. And the intention dynamic decision-making module is used for performing intention classification on the input of the user based on the optimized model parameters and performing dynamic adjustment according to context entropy and emotion analysis, and the intention dynamic decision-making module is used for adjusting and making a decision on the response strategy of the system based on the output of the input processing, intention collaboration and optimization module and according to the context of the current dialogue.

Description

Technical Field

[0001] The present invention relates to the technical field of natural language processing, and more specifically, to an artificial intelligence customer service system based on natural language processing. Background Art

[0002] With the rapid development of information technology, artificial intelligence (AI) is increasingly being applied across various fields, particularly in the customer service industry. The introduction of AI-powered customer service systems has become a key technology for improving service efficiency and customer experience. Traditional customer service methods typically rely on manual processing, which can lead to slow responses, high error rates, and excessive employee stress when faced with a large number of customer inquiries. This directly impacts customer satisfaction and business operational efficiency.

[0003] Natural language processing, a key branch of artificial intelligence, aims to enable computers to understand and generate natural language. By processing human language, natural language processing enables computers to interact with humans more naturally. Applied to AI customer service systems, natural language processing can significantly enhance the system's intelligence, enabling automated, efficient, and accurate customer service. Natural language processing technologies, including speech recognition, semantic understanding, sentiment analysis, and dialogue management, enable computers to understand customer needs, automatically generate appropriate responses, and provide precise assistance. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an artificial intelligence customer service system based on natural language processing to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solutions: an artificial intelligence customer service system based on natural language processing, comprising an input processing module, an intent collaboration module, an intent optimization module, and an intent dynamic decision module;

[0006] The input processing module receives and pre-processes the user's voice and text input, and calculates the context entropy value based on the feature distribution of N historical rounds of dialogue and the time decay weight;

[0007] The intention collaboration module outputs an emotion intensity signal through a multimodal emotion recognition model based on the input information, and generates an intention correction vector in combination with a preset intention-emotion mapping matrix;

[0008] The intent optimization module constructs and optimizes a collaborative loss function, including the original intent loss term, the context entropy weighted penalty term, and the sentiment distribution alignment term. Based on the optimized model parameters, it classifies the user's input intent and dynamically adjusts it based on the context entropy and sentiment analysis.

[0009] The intention dynamic decision module is based on the output of the input processing, intention collaboration and optimization modules, and adjusts and decides the system's response strategy in real time according to the context of the current conversation and user emotions.

[0010] Preferably, as a preferred solution of the artificial intelligence customer service system based on natural language processing described in the present invention, it includes the input processing module that accepts user voice and text input and performs preprocessing, calculates the context entropy value based on the feature distribution of historical N rounds of dialogue and time decay weight to quantify the complexity and context coherence of the dialogue, specifically including:

[0011] Receive voice and text information input by the user, convert the voice input into text through automatic speech recognition, and perform text preprocessing through word segmentation and stop word removal to obtain a processed text sequence;

[0012] Represent the state of each round of dialogue as a vector Based on the feature distribution of the historical N rounds of dialogue, the complexity of the current dialogue is calculated to obtain the context entropy of the current dialogue. The specific calculation formula of the context entropy is: , ,in, For historical dialogue rounds The probability distribution of represents context entropy, N is the number of historical conversations, t is the text representation of the current conversation, It is expressed as a time-decay weight, which is used to simulate the influence of time on context information.

[0013] Preferably, as a preferred solution of the artificial intelligence customer service system based on natural language processing described in the present invention, the intention collaboration module outputs an emotion intensity signal through a multimodal emotion recognition model based on the input information, and generates an intention correction vector in combination with a preset intention-emotion mapping matrix, specifically including:

[0014] Build a multimodal emotion recognition model to extract features from the input speech and text, including acoustic features and emotional features. Then, perform a weighted average fusion of the features from different modalities to obtain a global emotion feature vector. This fused feature vector is then used for classification to obtain the probability distribution of emotion categories.

[0015] The multimodal emotion recognition model is used to analyze the input information, which includes voice and text input, and extract the emotion intensity signal. The emotion intensity signal represents the intensity of the user's current emotion. Different intensity values ​​represent different emotion categories. The emotion intensity signal vector is represented as ,in, Indicates the The intensity value of the emotion class, is the number of emotion categories, within the scope;

[0016] Setting up an intent-emotion mapping matrix Each row of the matrix represents an emotion category, and each column represents an intention associated with it. Matrix multiplication is used to calculate the emotion intensity signal and the intention-emotion mapping matrix to generate the intention correction vector ,in, represents the influence of emotion u on intention v, The correction value of each intention is represented, and the user's intention is adjusted by the intention correction vector to achieve coordinated adjustment of emotion and intention.

[0017] Preferably, as a preferred solution of the artificial intelligence customer service system based on natural language processing described in the present invention, it includes the construction of a multimodal emotion recognition model, extracting features from the input speech and text, including acoustic features and emotional features, performing weighted averaging fusion on the features of different modalities to obtain a global emotion feature vector, and using the fused feature vector for classification to obtain the probability distribution of emotion categories, further including:

[0018] The acoustic feature vector extracted from speech is expressed as , the sentiment feature vector extracted from the text is expressed as , the fused emotion feature vector is ,in, is the fusion feature, and is the weight coefficient, which indicates the contribution of speech and text features to the emotion recognition results;

[0019] Use the softmax function to input the fused emotion feature vector into the emotion classifier and output an emotion category probability distribution: ,in, 、 and 、 are the weight and bias of the classifier, is the predicted probability of the kth emotion, and K is the total number of emotion categories.

[0020] Preferably, as a preferred solution of the artificial intelligence customer service system based on natural language processing described in the present invention, the intention optimization module constructs and optimizes the collaborative loss function, including the original intention loss term, the context entropy weighted penalty term, and the emotion distribution alignment term. Based on the optimized model parameters, the user's input is classified into intents, and dynamically adjusted according to the context entropy and emotion analysis, specifically including:

[0021] Construct the collaborative loss function, the specific formula is ,in, represents the loss term for the original intent classification, represents the context entropy weighted penalty term, represents the sentiment distribution alignment term, and It is a hyperparameter that controls the weight of each loss;

[0022] By optimizing the collaborative loss function , and get the optimal parameters of the model Based on the optimized model parameters, the user's input is classified into intents, and the intents are dynamically adjusted according to the results of context entropy and sentiment analysis. The specific formula is: ,in, is the probability of the intent category predicted by the model, represents contextual entropy, f is a function based on contextual entropy and sentiment analysis, is the forecast result after dynamic adjustment. It is the result of sentiment analysis.

[0023] Preferably, as a preferred solution of the artificial intelligence customer service system based on natural language processing described in the present invention, the original intention loss term uses cross entropy loss to measure the difference between the predicted intention and the true intention of the model, and the intention label input by the user is represented as , the probability distribution predicted by the model is expressed as , the original intention loss term is , where Q is the number of intent categories, is the true label of the q-th intent, is the probability of the qth type of intention predicted by the model;

[0024] The contextual entropy weighted penalty term is used to measure the entropy value of the model in different contexts, and the output distribution of the model is limited by the weighted entropy penalty term. The specific formula is: ,in, is the probability distribution of intention q in a given context, is the weighting coefficient, which indicates the influence of context entropy on the model;

[0025] The sentiment distribution alignment term is used to make the model's intent classification results consistent with the sentiment distribution, and the Kullback-Leibler divergence is used to measure the predicted sentiment distribution. and true sentiment distribution The specific formula is ,in, and are the true and predicted distributions of emotions, respectively, and K is the number of emotion categories.

[0026] Preferably, as a preferred solution of the artificial intelligence customer service system based on natural language processing described in the present invention, the intention dynamic decision module is based on the output of the input processing, intention collaboration and optimization modules, and adjusts and decides the system's response strategy in real time according to the context of the current conversation and user emotions, specifically including:

[0027] Based on the dynamically adjusted intent prediction results, combined with the context of the current conversation and the user's emotions, the response strategy is determined. The response strategy is the system's selection of the best response behavior based on the input intent, context, and emotions. The specific formula is: ,in, It is response options, is the match between intent categories, response options, and sentiment analysis, is the weight of the response option, and the selected response strategy A is the response that maximizes the matching degree;

[0028] Self-adjustment is performed based on user feedback satisfaction. When the feedback satisfaction is negative, the model is further optimized by adjusting the loss function and updating the weights to adapt to different conversation contexts and user emotions. This further includes:

[0029] Display feedback: Users click the "satisfied" or "unsatisfied" buttons to provide feedback ratings, and the user's feedback satisfaction is ,in, , Expressed complete satisfaction, Express neutrality, Totally dissatisfied, and are the minimum and maximum values ​​of the score respectively;

[0030] Implicit feedback: The user's emotional value in the current conversation is obtained based on the user's frequency of continued conversation, dwell time, interactive tone, and response attitude;

[0031] The user's feedback satisfaction is obtained through a comprehensive combination of explicit and implicit feedback, and self-adjustment is performed. When the user feedback is negative, the loss function is adjusted and optimized to adapt to different conversation contexts and user emotions.

[0032] On the other hand, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements a functional module of an artificial intelligence customer service system based on natural language processing as described above in the present invention.

[0033] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, an artificial intelligence customer service system based on natural language processing as described above in the present invention is implemented.

[0034] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0035] By accepting the user's voice and text input and preprocessing it, the context entropy value is calculated based on the feature distribution of historical N rounds of conversations and the time decay weight, the emotion intensity signal is output through the multimodal emotion recognition model according to the input information, and the intention correction vector is generated in combination with the preset intention-emotion mapping matrix, and a collaborative loss function is constructed and optimized, including the original intention loss term, the context entropy weighted penalty term and the emotion distribution alignment term. Based on the optimized model parameters, the user's input is classified into intents, and dynamically adjusted according to the context entropy and emotion analysis. According to the context of the current conversation and the user's emotions, the response strategy of the system is adjusted and decided in real time. The collaborative work of the input processing module, the intention collaboration module, the intention optimization module, and the intention dynamic decision-making module can significantly improve service efficiency, optimize customer experience, and reduce enterprise operating costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0037] Figure 1 This is a method flow chart of an artificial intelligence customer service system based on natural language processing in the present invention.

[0038] Table 1 is a data record table of a simulation experiment of an artificial intelligence customer service system based on natural language processing in the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0040] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the described features. In the description of this application, "plurality" means two or more, unless otherwise specifically specified.

[0041] In the description of this application, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art will recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in this application.

[0042] Example 1

[0043] This embodiment provides Figure 1 An artificial intelligence customer service system based on natural language processing is shown, which specifically includes an input processing module, an intent collaboration module, an intent optimization module, and an intent dynamic decision module;

[0044] The input processing module receives and pre-processes the user's voice and text input, and calculates the context entropy value based on the feature distribution of N historical rounds of dialogue and the time decay weight;

[0045] The intention collaboration module outputs an emotion intensity signal through a multimodal emotion recognition model based on the input information, and generates an intention correction vector in combination with a preset intention-emotion mapping matrix;

[0046] The intent optimization module constructs and optimizes a collaborative loss function, including the original intent loss term, the context entropy weighted penalty term, and the sentiment distribution alignment term. Based on the optimized model parameters, it classifies the user's input intent and dynamically adjusts it based on the context entropy and sentiment analysis.

[0047] The intention dynamic decision module is based on the output of the input processing, intention collaboration and optimization modules, and adjusts and decides the system's response strategy in real time according to the context of the current conversation and user emotions.

[0048] In this embodiment, the input processing module specifically needs to be explained. The input processing module receives and pre-processes the user's voice and text input, and calculates the context entropy value based on the feature distribution of N historical conversation rounds and the time decay weight to quantify the complexity and context coherence of the conversation. Specifically, it includes:

[0049] Receive voice and text information input by users, convert the voice input into text through automatic speech recognition, and perform text preprocessing through word segmentation and stop word removal to obtain a processed text sequence. It can quickly understand the user input and extract key information, including questions, emotions, and context, greatly accelerating customer service response time;

[0050] Represent the state of each round of dialogue as a vector Based on the feature distribution of the historical N rounds of dialogue, the complexity of the current dialogue is calculated to obtain the context entropy of the current dialogue. The specific calculation formula of the context entropy is: , ,in, For historical dialogue rounds The probability distribution of represents context entropy, N is the number of historical conversations, t is the text representation of the current conversation, It is expressed as a time-decay weight, which is used to simulate the influence of time on context information.

[0051] In this embodiment, the intention collaboration module specifically needs to be explained. The intention collaboration module outputs an emotion intensity signal based on the input information through a multimodal emotion recognition model, and generates an intention correction vector based on a preset intention-emotion mapping matrix. By integrating information from different sources, it can accurately identify user intent, specifically including:

[0052] Construct a multimodal emotion recognition model, extract features from the input speech and text, including acoustic features and emotional features, perform weighted averaging on the features of different modalities to obtain a global emotion feature vector, and use the fused feature vector for classification to obtain the probability distribution of emotion categories, further including:

[0053] The acoustic feature vector extracted from speech is expressed as , the sentiment feature vector extracted from the text is expressed as , the fused emotion feature vector is ,in, is the fusion feature, and is the weight coefficient, which indicates the contribution of speech and text features to the emotion recognition results;

[0054] Use the softmax function to input the fused emotion feature vector into the emotion classifier and output an emotion category probability distribution: ,in, 、 and 、 are the weight and bias of the classifier, is the predicted probability of the kth emotion, K is the total number of emotion categories;

[0055] The multimodal emotion recognition model is used to analyze the input information, which includes voice and text input, and extract the emotion intensity signal. The emotion intensity signal represents the intensity of the user's current emotion. Different intensity values ​​represent different emotion categories. The emotion intensity signal vector is represented as ,in, Indicates the The intensity value of the emotion class, is the number of emotion categories, within the scope;

[0056] Setting up an intent-emotion mapping matrix Each row of the matrix represents an emotion category, and each column represents an intention associated with it. Matrix multiplication is used to calculate the emotion intensity signal and the intention-emotion mapping matrix to generate the intention correction vector ,in, represents the influence of emotion u on intention v, The correction value of each intention is represented, and the user's intention is adjusted by the intention correction vector to achieve coordinated adjustment of emotion and intention.

[0057] In this embodiment, the intent optimization module specifically needs to be explained. The intent optimization module constructs and optimizes a collaborative loss function, including an original intent loss term, a context entropy weighted penalty term, and a sentiment distribution alignment term. Based on the optimized model parameters, the module classifies the user's input intent and dynamically adjusts it based on context entropy and sentiment analysis. This module can optimize responses based on the user's historical data, preferences, and current needs, avoiding mechanical answers and enhancing a personalized experience. Specifically, it includes:

[0058] The collaborative loss function is constructed as follows: ,in, represents the loss term for the original intent classification, represents the context entropy weighted penalty term, represents the sentiment distribution alignment term, and It is a hyperparameter that controls the weight of each loss;

[0059] The original intent loss term uses cross entropy loss to measure the difference between the model's predicted intent and the true intent, and the user input intent label is represented as , the probability distribution predicted by the model is expressed as , the original intention loss term is , where Q is the number of intent categories, is the true label of the q-th intent, is the probability of the qth type of intention predicted by the model;

[0060] The contextual entropy weighted penalty term is used to measure the entropy value of the model in different contexts, and the output distribution of the model is limited by the weighted entropy penalty term. The specific formula is: ,in, is the probability distribution of intention q in a given context, is the weighting coefficient, which indicates the influence of context entropy on the model;

[0061] The sentiment distribution alignment term is used to make the model's intent classification results consistent with the sentiment distribution, and the Kullback-Leibler divergence is used to measure the predicted sentiment distribution. and true sentiment distribution The specific formula is ,in, and are the true distribution and predicted distribution of emotions, respectively, and K is the number of emotion categories;

[0062] By optimizing the collaborative loss function , and get the optimal parameters of the model Based on the optimized model parameters, the user's input is classified into intents, and the intents are dynamically adjusted according to the results of context entropy and sentiment analysis. The specific formula is: ,in, is the probability of the intent category predicted by the model, represents contextual entropy, f is a function based on contextual entropy and sentiment analysis, is the forecast result after dynamic adjustment. It is the result of sentiment analysis.

[0063] In this embodiment, the dynamic intent decision-making module is specifically described. This module is based on the output of the input processing, intent coordination, and optimization modules, and adjusts and decides the system's response strategy in real time according to the context of the current conversation and user emotions. By analyzing user input, it extracts valuable data, provides enterprises with user behavior analysis and business insights, and improves service quality and customer satisfaction. Specifically, it includes:

[0064] Based on the dynamically adjusted intent prediction results, combined with the context of the current conversation and the user's emotions, the response strategy is determined. The response strategy is the system's selection of the best response behavior based on the input intent, context, and emotions. The specific formula is: ,in, It is response options, is the match between intent categories, response options, and sentiment analysis, is the weight of the response option, and the selected response strategy A is the response that maximizes the matching degree;

[0065] Self-adjustment is performed based on user feedback satisfaction. When the feedback satisfaction is negative, the model is further optimized by adjusting the loss function and updating the weights to adapt to different conversation contexts and user emotions. This further includes:

[0066] Display feedback: Users click the "satisfied" or "unsatisfied" buttons to provide feedback ratings, and the user's feedback satisfaction is ,in, , Expressed complete satisfaction, Express neutrality, Totally dissatisfied, and are the minimum and maximum values ​​of the score respectively;

[0067] Implicit feedback: The user's emotional value in the current conversation is obtained based on the user's frequency of continued conversation, dwell time, interactive tone, and response attitude;

[0068] The user's feedback satisfaction is obtained through a comprehensive combination of explicit and implicit feedback, and self-adjustment is performed. When the user feedback is negative, the loss function is adjusted and optimized to adapt to different conversation contexts and user emotions.

[0069] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0070] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the functional module of an artificial intelligence customer service system based on natural language processing as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0071] Example 2

[0072] The following is another embodiment of the present invention, which provides an artificial intelligence customer service system based on natural language processing. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0073] This experiment aims to verify the effectiveness of an artificial intelligence customer service system based on natural language processing in real-time dynamic response and understanding of user emotions. Through multimodal input, emotion recognition, intent correction and optimization technologies, the system's understanding of user needs and the accuracy of its response are improved.

[0074] The simulation experiment steps are implemented according to the content of an artificial intelligence customer service based on natural language processing provided in Example 1, and the specific steps include:

[0075] Set the conversation collection frequency to 1000Hz and the collection time to 30 minutes;

[0076] Set up a synchronization trigger line and collect conversation data from 100 users, including 20 users with different emotional states and 80 ordinary users;

[0077] Based on historical conversation data, virtual voice and text input data is synthesized for label classification and fitted with an intent classification model.

[0078] Divide the collected conversation data into short time windows of 1 minute each;

[0079] Calculate the maximum context entropy value within a short time window and use this value to normalize the weight of the conversation turn;

[0080] Extract pre-processed text features, combine them with emotion intensity signals, and make corrections through the intention collaboration module;

[0081] The optimized collaborative loss function is used to classify the intent in each window and output the classification results.

[0082] Based on the optimized model output, the system response strategy is dynamically adjusted to generate the optimal response behavior;

[0083] Evaluate the accuracy of response behavior and make system self-adjustments based on user feedback satisfaction.

[0084] The specific data of the above simulation experiment are:

[0085] Time / minute User emotional state Input features (text / speech) Contextual entropy Emotional intensity Intent Classification Response Strategy User satisfaction Warning threshold Response result (yes / no) 0-5 neutral "Hello, can you help me?" 0.12 0.05 Search Products Provide information 0.85 0.80 no 5-10 Happy "Thanks for the help!" 0.08 0.90 feedback Praise response 0.90 0.80 no 10-15 anxiety "I need help, what's going on?" 0.14 0.70 Urgent Issues Solve the problem 0.78 0.75 yes 15-20 dissatisfied "Why are you so slow?" 0.18 0.85 complaint Apology response 0.60 0.75 yes 20-25 calm "OK, thanks" 0.10 0.02 greeting Simple response 0.95 0.80 no

[0086] Table 1

[0087] Experimental analysis:

[0088] By dynamically adjusting the optimization model for intent classification, the system can capture subtle changes in user emotions in real time and more accurately determine response strategies. When users are anxious, the system prioritizes problem-solving strategies, improving the timeliness and relevance of responses.

[0089] The system's warning threshold can effectively determine whether users are dissatisfied through historical analysis of multiple rounds of conversations and emotional intensity assessment, allowing for proactive response adjustments.

[0090] The experimental results show that when the user's emotional intensity is high, the system can respond in a timely manner and mobilize appropriate strategies to deal with it, effectively improving user satisfaction.

[0091] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

Claims

1. An artificial intelligence customer service system based on natural language processing, characterized by: Includes input processing module, intent collaboration module, intent optimization module, and intent dynamic decision module; The input processing module receives and pre-processes the user's voice and text input, and calculates the context entropy value based on the feature distribution of N historical rounds of dialogue and the time decay weight; The intention collaboration module outputs an emotion intensity signal through a multimodal emotion recognition model based on the input information, and generates an intention correction vector in combination with a preset intention-emotion mapping matrix; The intent optimization module constructs and optimizes a collaborative loss function, including the original intent loss term, the context entropy weighted penalty term, and the sentiment distribution alignment term. Based on the optimized model parameters, it classifies the user's input intent and dynamically adjusts it based on the context entropy and sentiment analysis. The intention dynamic decision module is based on the output of the input processing, intention collaboration and optimization modules, and adjusts and decides the system's response strategy in real time according to the context of the current conversation and user emotions.

2. The artificial intelligence customer service system based on natural language processing according to claim 1, characterized in that: The input processing module receives and pre-processes the user's voice and text input, and calculates the context entropy value based on the feature distribution of N historical rounds of dialogue and the time decay weight, specifically including: Receive voice and text information input by the user, convert the voice input into text through automatic speech recognition, and perform text preprocessing through word segmentation and stop word removal to obtain a processed text sequence; Represent the state of each round of dialogue as a vector Based on the feature distribution of the historical N rounds of dialogue, the complexity of the current dialogue is calculated to obtain the context entropy of the current dialogue. The specific calculation formula of the context entropy is: , ,in, For historical dialogue rounds The probability distribution of represents context entropy, N is the number of historical conversations, t is the text representation of the current conversation, Expressed as time-decay weight.

3. The artificial intelligence customer service system based on natural language processing according to claim 1, characterized in that: The intention collaboration module outputs an emotion intensity signal through a multimodal emotion recognition model based on the input information, and generates an intention correction vector in combination with a preset intention-emotion mapping matrix, specifically including: Build a multimodal emotion recognition model to extract features from the input speech and text, including acoustic features and emotional features. Then, perform a weighted average fusion of the features from different modalities to obtain a global emotion feature vector. This fused feature vector is then used for classification to obtain the probability distribution of emotion categories. The input information is analyzed by the multimodal emotion recognition model to extract the emotion intensity signal, which represents the intensity of the user's current emotion. The emotion intensity signal vector is represented as ,in, Indicates the The intensity value of the emotion class, is the number of emotion categories; Setting up an intent-emotion mapping matrix , use matrix multiplication to calculate the emotion intensity signal and the intention-emotion mapping matrix to generate the intention correction vector ,in, represents the influence of emotion u on intention v, Indicates the modifier value for each intention.

4. The artificial intelligence customer service system based on natural language processing according to claim 3, characterized in that: The multimodal emotion recognition model is constructed to extract features from the input speech and text, including acoustic features and emotional features, perform weighted averaging on the features of different modalities to obtain a global emotion feature vector, and use the fused feature vector for classification to obtain the probability distribution of emotion categories, further comprising: The acoustic feature vector extracted from speech is expressed as , the sentiment feature vector extracted from the text is expressed as , the fused emotion feature vector is ,in, is the fusion feature, and is the weight coefficient; Use the softmax function to input the fused emotion feature vector into the emotion classifier and output an emotion category probability distribution: ,in, 、 and 、 are the weight and bias of the classifier, is the predicted probability of the kth emotion, and K is the total number of emotion categories.

5. The artificial intelligence customer service system based on natural language processing according to claim 1, characterized in that: The intent optimization module constructs and optimizes a collaborative loss function, including the original intent loss term, the context entropy weighted penalty term, and the sentiment distribution alignment term. Based on the optimized model parameters, it classifies the user's input intent and dynamically adjusts it based on context entropy and sentiment analysis. Specifically, it includes: Construct the collaborative loss function, the specific formula is ,in, represents the loss term for the original intent classification, represents the context entropy weighted penalty term, represents the sentiment distribution alignment term, and It is a hyperparameter that controls the weight of each loss; By optimizing the collaborative loss function , and get the optimal parameters of the model Based on the optimized model parameters, the user's input is classified into intents, and the intents are dynamically adjusted according to the results of context entropy and sentiment analysis. The specific formula is: ,in, is the probability of the intent category predicted by the model, represents contextual entropy, f is a function based on contextual entropy and sentiment analysis, is the forecast result after dynamic adjustment. It is the result of sentiment analysis.

6. The artificial intelligence customer service system based on natural language processing according to claim 5, characterized in that: The original intent loss term uses cross entropy loss to measure the difference between the model's predicted intent and the true intent, and the user input intent label is represented as , the probability distribution predicted by the model is expressed as , the original intention loss term is , where Q is the number of intent categories, is the true label of the q-th intent, is the probability of the qth type of intention predicted by the model; The contextual entropy weighted penalty term is used to measure the entropy value of the model in different contexts, and the output distribution of the model is limited by the weighted entropy penalty term. The specific formula is: ,in, is the probability distribution of intention q in a given context, is the weighting coefficient; The sentiment distribution alignment term is used to make the model's intent classification results consistent with the sentiment distribution, and the Kullback-Leibler divergence is used to measure the predicted sentiment distribution. and true sentiment distribution The specific formula is ,in, and are the true and predicted distributions of emotions, respectively, and K is the number of emotion categories.

7. The artificial intelligence customer service system based on natural language processing according to claim 1, characterized in that: The intent dynamic decision module is based on the output of the input processing, intent coordination and optimization modules, and adjusts and decides the system's response strategy in real time according to the context of the current conversation and user emotions. Specifically, it includes: Based on the dynamically adjusted intent prediction results, combined with the current conversation context and user emotions, the response strategy is determined. The specific formula is: ,in, It is response options, is the match between intent categories, response options, and sentiment analysis, is the weight of the response option, and the selected response strategy A is the response that maximizes the matching degree; It adjusts itself based on the user's feedback satisfaction. When the feedback satisfaction is negative, it further optimizes the model by adjusting the loss function and updating the weights to adapt to different conversation contexts and user emotions.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, it implements the functional module of the artificial intelligence customer service system based on natural language processing as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the functional module of the artificial intelligence customer service system based on natural language processing described in any one of claims 1 to 7 is implemented.

Images