Intention recognition method based on large language model and customer portrait classification model
By combining large language model and customer portrait classification model, integrating natural language understanding and customer behavior analysis, the robustness and adaptability of existing intention recognition methods in complex scenarios are solved, and more accurate and personalized customer intentions are achieved. Identification improves service efficiency and accuracy.
Patent Information
- Application Number
- CN202410943941.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-07-15
AI Technical Summary
Existing intent recognition methods have problems of poor robustness and low adaptability when dealing with complex and varied customer intentions, especially in personalized service scenarios, which are difficult to accurately capture the true intentions of customers.
Combining the large language model and customer portrait classification model, by integrating natural language understanding and customer behavior analysis, comprehensive score vectors are obtained to indicate customer intentions, and the output of the fine-tuning large model and customer portrait classification model is used to identify intentions, and through the agent, the The prompt word project adjusts the weight to achieve more accurate intention understanding and personalized service.
It improves the accuracy and efficiency of intention recognition, can more accurately understand and predict customer intentions, provide more personalized and efficient customer service, and adapt to complex locale environments and changing customer needs.
Smart Images

Figure CN118940838A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and in particular to an intent recognition method based on a large language model and a customer portrait classification model. Background Art
[0002] In the field of natural language processing, intent recognition is a key link. With the continuous development of the Internet and artificial intelligence technology, intelligent dialogue systems such as intelligent customer service and intelligent assistants are becoming more and more popular in various application scenarios. However, traditional intent recognition methods usually rely on a large amount of pre-made manual rule knowledge and traditional NLP recognition methods such as conversation intent slot filling, which requires a lot of manual participation and has a low accuracy rate.
[0003] At present, the progress of deep learning technology has enabled large language models (such as GPT, LLAMA, etc.) to achieve remarkable results in natural language processing tasks. At the same time, customer portrait classification models are also increasingly valued because they can better understand customer needs and preferences. Customer portrait classification models can provide rich background information by mining customer attributes and behavior data. Therefore, combining large language models and customer portrait classification models for intent recognition has important research significance and application value.
[0004] The rapid development of big data and machine learning technologies has promoted the widespread application of intelligent customer service systems and intelligent personal assistants. Intent recognition, as the core technology of these applications, has become a research hotspot.
[0005] Current intent recognition methods mainly rely on predefined rules or traditional machine learning models, which have problems such as poor robustness and low adaptability. Large language models (such as GPT and LLAMA) perform well in natural language processing tasks due to their powerful language generation and understanding capabilities. However, current large language models still have certain limitations when dealing with personalized needs and complex scenarios. Generally, customer intent recognition implemented by large language models is mostly based on two routes: configuring customer conversation scenarios through prompt word engineering, and training a large amount of customer dialogue corpus through continuous pre-training or fine-tuning methods (such as SFT and LORA) to solve the understanding and recognition of customer intent.
[0006] With the development of Internet technology, enterprises and organizations need to process a large amount of customer interaction data to better understand customer needs and intentions. However, existing intent recognition systems usually rely on predefined rules, keyword matching, rule engines or simple statistical models, which have significant limitations in dealing with complex and changing customer intentions.
[0007] Although large language models (such as BERT, GPT, etc.) have shown potential in natural language processing. Existing technologies have certain contextual understanding capabilities in natural language understanding, but they are insufficient in personalized services. Customer profiling is a method of analyzing and describing customer characteristics and behaviors. It can help product departments and operations personnel better understand customer needs and preferences, thereby providing better quality and personalized products and services. Customer profiling technology also has considerable advantages in customer information understanding scenarios.
[0008] Although the customer portrait classification model can provide customer features, it still needs to be improved in real-time intent recognition. Existing technologies have not yet fully explored and utilized the advantages of the combination of the two. At present, some existing cases that can be seen are mainly to use the big model to extract richer and deeper features from the customer's original data such as conversation text, customer behavior information, and work order details, such as customer emotions, attitudes, intentions, preferences, values, etc. Because the conversation text may reflect the customer's real needs and personality, and it can increase the dimension and complexity of the data, thereby increasing the amount of information and differentiation of the data. Therefore, the natural language processing capabilities that the big language model is good at are used to extract the customer's semantics, pragmatics, tone, style, emotion, expression, sound, action and other levels of features from the customer's text, image, audio, video and other forms of data, so as to build a more comprehensive and detailed customer portrait. However, the combination of the above-mentioned big language model and customer portrait mainly uses the big language model as a feature extraction or data source for customer portrait, and only serves the customer portrait unilaterally, rather than combining the two in a two-way manner to meet the changes in special and varied rich business scenarios.
[0009] The particularly changing business scenario is facing highly personalized and context-changing customer interaction scenarios. Traditional intent recognition systems often find it difficult to accurately capture the true intentions of customers. Therefore, a more intelligent technology is needed that can deeply understand customer characteristics and effectively predict their needs to improve the accuracy and robustness of intent recognition. This new intent recognition technology should combine the advantages of large language models and customer portrait classification models to cope with complex language environments and changing customer needs, thereby achieving smarter customer intent recognition and more accurate subsequent intelligent behavior execution. Summary of the invention
[0010] The embodiment of the present application provides an intention recognition method based on a large language model and a customer portrait classification model, which combines deep learning technology and big data analysis, and integrates natural language understanding and customer behavior analysis to achieve high-precision recognition and response processing of customer intentions. The embodiment of the present application can improve the accuracy and efficiency of intention recognition, and can more accurately understand and predict customer intentions, thereby providing more personalized and efficient customer service.
[0011] In the first aspect, an embodiment of the present application proposes an intent recognition method based on a large language model and a customer portrait classification model, the method comprising: obtaining the customer's current relevant orders and background information based on the customer ID information; the customer portrait classification model obtains a first scoring vector based on the relevant orders and background information; the first scoring vector is used to indicate the current customer portrait classification score; the fine-tuning large model performs semantic recognition on the context dialogue of the input customer conversation to obtain a second scoring vector; the second scoring vector is used to indicate the customer's current emotional feedback score; the fine-tuning large model is obtained by fine-tuning the large language model; the first scoring vector and the second scoring vector are merged to obtain a comprehensive scoring vector; the comprehensive scoring vector is used to indicate different tendency weights in different business scenarios; the inference large model performs inference based on the comprehensive scoring vector and outputs the customer's intent; the corresponding business logic is driven to execute according to the customer's intent, and the corresponding business logic includes providing customized suggestions or triggering specific service processes.
[0012] In some implementations, a linear weighted algorithm is used to concatenate the first score vector and the second score vector to form a comprehensive score vector.
[0013] In some embodiments, fusing the first score vector and the second score vector to obtain a comprehensive score vector includes: using a multi-layer perceptron (MLP) to perform weighted fusion of the first score vector and the second score vector by learning a set of fixed weights to obtain the comprehensive score vector.
[0014] In some embodiments, the inference big model performs inference based on the comprehensive scoring vector and outputs the customer's intention classification, including: the number of customer intention classifications is M, and the Top-K intentions with the highest scores are selected according to the M intention classifications; or low-confidence intentions among the M intentions are filtered out by setting a threshold to obtain K intentions that meet the requirements; or the M intentions are sorted by priority; or the M intentions are executed in parallel.
[0015] In some embodiments, the first scoring vector and the second scoring vector are fused to obtain a comprehensive scoring vector, including: arranging the output weights of the first scoring vector and the second scoring vector through agent arrangement and prompt word engineering.
[0016] In some embodiments, the output weights of the first scoring vector and the second scoring vector are arranged through intelligent agent orchestration and prompt word engineering, including: determining the key nodes of the task process of the inference large model, the key nodes including: the customer's original intention input node, the inference call node of the customer portrait classification model, the inference call node of the fine-tuning large model, the weight merging decision node, the behavior execution node and the multi-module combination process node; converting the output of the customer portrait classification model inference call node into a descriptive statement based on the semantic understanding of the fine-tuning large model; configuring the prompt word engineering for the first scoring vector and the second scoring vector, and adjusting different tendency weights in different business scenarios.
[0017] In some embodiments, the execution of corresponding business logic is driven according to the customer intent classification, including: determining an agent behavior library, the agent behavior library pre-sets tools and / or atomic capability services for registering business operation behaviors; selecting a corresponding agent combination in the agent behavior library according to the customer intent classification, and the agent combination is used to orchestrate and execute behaviors.
[0018] In some embodiments, the method also includes: obtaining the results of the interactive behavior of customer feedback based on the contextual text of the customer interaction; using the results of the interactive behavior of customer feedback as training corpus for retraining the feature weights of the customer portrait classification model; the customer portrait classification model retrains the feature weights based on the training corpus; including data integration and feature expansion, feature quantization of feedback annotations, automatic feature engineering, AutoML automatic machine learning training, incremental learning and online updates.
[0019] In some embodiments, the method further includes: fine-tuning the large model to automatically generate new annotations based on the results of the interaction behavior fed back by the customer; and continuously fine-tuning the large model using the conversation interaction results and the new annotations as input-output pairs to align the conversation and intent recognition.
[0020] In the second aspect, the embodiment of the present application proposes an intention recognition system based on a large language model and a customer portrait classification model, based on the method provided in the first aspect, including: an information acquisition module, used to obtain the current relevant orders and background information of the customer according to the customer ID information; a customer portrait classification model, used to obtain a first scoring vector according to the relevant orders and background information; the first scoring vector is used to indicate the current customer portrait classification score; a fine-tuning large model, used to semantically recognize the context dialogue of the input customer dialogue to obtain a second scoring vector; a fusion module, used for the second scoring vector to indicate the current customer emotional feedback score; the fine-tuning large model is obtained by fine-tuning the large language model; the first scoring vector and the second scoring vector are fused to obtain a comprehensive scoring vector; the comprehensive scoring vector is used to indicate different tendency weights in different business scenarios; an inference large model, used to infer according to the comprehensive scoring vector and output the customer's intention; an intention driving module, used to drive the execution of corresponding business logic according to the customer's intention, and the corresponding business logic includes providing customized suggestions or triggering specific service processes. In the third aspect, the embodiment of the present application provides an electronic device, including: at least one memory, used to store programs; at least one processor, used to execute the program stored in the memory, when the program stored in the memory is executed, the processor is used to execute the method provided in the first aspect.
[0021] In a fourth aspect, an embodiment of the present application provides a computer storage medium, in which instructions are stored. When the instructions are executed on a computer, the computer executes the method provided in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] In order to more clearly illustrate the technical solutions of the multiple embodiments disclosed in this specification, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only the multiple embodiments disclosed in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0023] The following is a brief introduction to the drawings required for describing the embodiments or prior art.
[0024] Figure 1 A schematic diagram of a process engine framework for implementing agent orchestration provided in Example 1 of the present application;
[0025] Figure 2 A flow chart of the training phase of the customer portrait classification model provided in the embodiment of the present application;
[0026] Figure 3 A flow chart of the fine-tuning phase of a large language model provided in an embodiment of the present application;
[0027] Figure 4 A flowchart for building and training a large inference model provided in an embodiment of the present application;
[0028] Figure 5 This is a schematic diagram of the dual model fusion training process;
[0029] Figure 6 A schematic diagram of realizing the reasoning large model C by arranging prompt word engineering through the intelligent agent orchestration tool;
[0030] Figure 7 A schematic diagram for result feedback and continuous optimization;
[0031] Figure 8 A schematic diagram of a process engine framework for implementing agent orchestration provided in Example 1 of the present application. DETAILED DESCRIPTION
[0032] In the description of the embodiments of the present application, words such as "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a concrete way.
[0033] In the description of the embodiments of the present application, the term "and / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, B exists alone, and A and B exist at the same time. In addition, unless otherwise specified, the term "multiple" means two or more. For example, multiple systems refers to two or more systems, and multiple terminals refers to two or more terminals.
[0034] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprises", "has" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.
[0035] In the description of the embodiments of the present application, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it can be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.
[0036] In the description of the embodiments of the present application, the terms "first\second\third, etc." or module A, module B, module C, etc. are only used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that the specific order or sequence can be interchanged where permitted so that the embodiments of the present application described here can be implemented in an order other than that illustrated or described here.
[0037] In the description of the embodiments of the present application, the numbers representing the steps, such as S110, S120, etc., do not necessarily mean that the steps will be executed in this manner. If permitted, the order of the previous and next steps can be interchanged, or they can be executed simultaneously.
[0038] Relevant scientific terms used in this application:
[0039] Pre-training, first train a deep learning model on a large-scale dataset, such as using self-supervised learning or unsupervised learning.
[0040] Fine-tuning: fine-tuning the pre-trained model using the training set of the target task. Usually, only a part of the layers in the pre-trained model are fine-tuned, such as only fine-tuning the last few layers or some intermediate layers of the model. During the fine-tuning process, the model is optimized through the back-propagation algorithm to make the model perform better on the target task;
[0041] Evaluation: Use the test set of the target task to evaluate the fine-tuned model and obtain the performance indicators of the model on the target task.
[0042] LoRA is a method for fine-tuning large pre-trained language models by inserting low-rank matrices at specific layers to reduce the number of parameters and improve efficiency and performance.
[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.
[0044] The embodiment of the present application proposes a method based on a large language model and a customer portrait classification model, which fuses the output results of the customer portrait classification model with the fine-tuning model of the large language model based on historical customer behavior, so that the large language model can combine the characteristic data of the customer portrait with the current real-time customer conversation context as targeted reasoning, thereby completing a more accurate understanding and judgment of customer intentions. Respond to subsequent customer behaviors and call on relevant capabilities to meet customer intention needs in the form of intelligent agents. After obtaining positive or negative feedback from customers, the results of this round of interactive behavior are used as training corpus for further fine-tuning and alignment of the large model, and the feature weights in the original customer portrait classification model are adjusted.
[0045] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in conjunction with the accompanying drawings.
[0046] Figure 1 Flowchart of the intent recognition method based on the large language model and customer portrait classification model provided in the embodiment of the present application. Figure 1 As shown, the method comprises the following steps:
[0047] S1, obtain the customer's current related orders and background information based on the customer ID information.
[0048] S2, the customer portrait classification model obtains a first scoring vector according to relevant orders and background information; the first scoring vector is used to indicate the current customer business classification score.
[0049] S3, fine-tuning the large model to perform semantic recognition on the context of the input customer conversation to obtain a second scoring vector; the second scoring vector is used to indicate the customer's current emotional feedback score; the fine-tuning large model is obtained by fine-tuning the language large model.
[0050] S4, the reasoning model fuses the first scoring vector and the second scoring vector to obtain a comprehensive scoring vector; the comprehensive scoring vector is used to indicate different tendency weights in different business scenarios; reasoning is performed based on the comprehensive scoring vector to output the customer's intention.
[0051] S5, drives the execution of corresponding business logic based on the customer's intention, which includes providing customized suggestions or triggering specific service processes.
[0052] Based on the above application process, the technical solutions in the embodiments of this application are explained one by one through multiple aspects such as customer portrait classification model, fine-tuning large model, reasoning large model, intent-driven execution, etc.
[0053] First, we introduce the customer portrait classification model. Figure 2 This is a flow chart of the construction and training phases of the customer portrait classification model provided in the embodiment of the present application. Figure 2 As shown, the following steps are included:
[0054] S21, obtain customer data and pre-process the customer data; the customer data includes historical customer order records and / or work order records extracted from the database according to customer ID information, and the pre-processing includes data cleaning, data standardization, missing value processing and outlier processing.
[0055] S22, obtaining a sample set of customer portrait classification according to the customer data, and dividing the sample set into training set samples and test set samples of the customer portrait classification.
[0056] In some implementations, basic customer information features can be constructed based on customer data, customer tags can be generated, current resource information can be extracted, historical behavior features can be calculated, and a customer portrait classification sample set can be constructed based on basic customer information features, customer tags, current resource information, and historical behavior features.
[0057] Among them, basic customer information features include age, gender, region, etc.; customer tags include high value, potential churn, etc.; current resource information includes the type of package that has been ordered, contract period, status of activated resources, etc.; historical behavior features include number of complaints, diagnosis frequency, and probability of classification of daily subsequent operations, etc.
[0058] S23, construct a customer portrait classification model based on the existing machine learning classification model, and train the customer portrait classification model.
[0059] The customer portrait classification model can be built and trained by following the steps below:
[0060] S231, determine the algorithm of customer portrait classification model A.
[0061] The algorithm of the customer portrait classification model A can select a classification algorithm, where the classification algorithms include random forest, XGBoost, and AdaBoost.
[0062] S232, determining a training strategy, including using a grid search method to tune hyperparameters, and using a customer portrait classification training set sample to use a grid search method to tune hyperparameters to train a customer portrait classification model.
[0063] S24, evaluate and iterate the customer portrait classification model A.
[0064] Model A is evaluated using the test set of customer portrait classification to obtain the performance indicators of model A on the customer portrait classification task.
[0065] Indicators such as precision, recall, and F1 value can be used as performance indicators of customer portrait classification model A. According to the evaluation results of these performance indicators, erroneous samples can be identified, features in the samples can be corrected, or model parameters can be adjusted.
[0066] In some implementations, the system may also use feedback data from the context of customer sessions to periodically retrain the customer profile classification model A to adapt to new customer data.
[0067] In the actual customer conversation state, the customer's current related orders and background information are inferred through the trained customer profile classification model A to obtain a first prediction result, and the first prediction result is normalized and converted into a unified score scoring vector, which is recorded as the first scoring vector. The first scoring vector indicates the score of the current customer profile classification.
[0068] Exemplarily, the first scoring vector is such as [upgrade cloud computer package: 0.8, solve network problem: 0.7, ...].
[0069] Next, we will introduce the fine-tuning large model based on the large language model.
[0070] For fine-tuning, you first need to use self-supervised learning or unsupervised training on a large dataset to train a large language model as a pre-trained large model, and then use the fine-tuning corpus on the target task to fine-tune the pre-trained large model. Usually, only a part of the layers in the pre-trained large model are fine-tuned, such as only fine-tuning the last few layers or some intermediate layers of the large model. During the fine-tuning process, the model is optimized through the back-propagation algorithm, so that the fine-tuned large model performs better on the target task.
[0071] If there is already a pre-trained and fine-tuned baseline large model for a specific vertical domain class, this baseline large model can be used directly as a pre-trained large model.
[0072] In some implementations, various open source or closed source customer service big models can be directly used as pre-trained big models. The customer service big model is used to reason about the customer interaction context and output a score for the customer intent classification or a score for the customer behavior prediction classification.
[0073] For example, Figure 3 This is a flow chart of the fine-tuning phase of the large language model provided in the embodiment of the present application. Figure 3 As shown, the following steps are included:
[0074] S31, obtain fine-tuning corpus on a large-scale dataset and determine input-output pairs, where the input is the context of the customer interaction session and the output is the corresponding reasoning description of the pre-trained large model.
[0075] In some implementations, the behavioral dialogues of the customer's historical conversations may be used as fine-tuning information to construct a large amount of fine-tuning corpus for reasoning training.
[0076] Exemplarily, daily conversations are input into a pre-trained large model, and the corresponding reasoning description is output.
[0077] Based on the input customer description, the inference output is performed through the generation task of the pre-trained large model, and the input customer description and the output inference response constitute the questions and answers of the training dialogue of the large model, thereby obtaining the fine-tuning corpus.
[0078] Exemplarily, the pre-trained large model generates JSONL fine-tuning data in the following format according to the input customer description:
[0079]
[0080] Construct input-output pairs based on the fine-tuning corpus: the input is the customer description, and the output is the corresponding reasoning description. The reasoning description includes the intent classification score or the behavior prediction classification score.
[0081] S32, determining a fine-tuning method.
[0082] The first fine-tuning method adopts supervised fine-tuning (SFT), SFT includes the following steps: S321, determine the pre-trained large model, the pre-trained large model can be Qwen or Llama; S322, determine the fine-tuning task, the fine-tuning task includes the intent classification task or the behavior prediction classification task; S323, use the prepared fine-tuning corpus to fine-tune all parameters of the pre-trained large model.
[0083] The second fine-tuning method uses a low-rank adapter (Low-Rank Adaptation). The fine-tuning steps of the low-rank adapter include: S324, selecting a basic large language model such as Qwen, LLaMA; S325, defining the rank of the adaptation layer, only training the newly added low-rank matrix, and keeping the basic pre-trained large model unchanged.
[0084] S33, uses the two fine-tuning methods provided by S32 to fine-tune the pre-trained large model respectively. During the fine-tuning process, the learning rate, batch size and other parameters are set respectively, and the appropriate optimizer such as AdamW is used to monitor the training loss and verification indicators to obtain the first fine-tuned large model and the second fine-tuned large model.
[0085] S34, evaluate the performance of the first fine-tuned large model and the second fine-tuned large model, select the fine-tuned large model with the best performance, and deploy the best fine-tuned large model to the production environment.
[0086] In some embodiments, a test set can be extracted based on historical customer context interaction results, and the performance of the first fine-tuned large model obtained under the SFT method and the second fine-tuned large model obtained under the LoRA method can be evaluated based on the test set. The effects of SFT and LoRA are compared, the best solution is selected, and the best fine-tuned large model is deployed to the production environment.
[0087] It should be understood that the fine-tuned large model obtained in S34 in the embodiment of the present application performs semantic recognition on the customer context dialogue, determines the customer's current emotional feedback, and feeds back a score.
[0088] In the production environment, the customer interaction context is inferred by fine-tuning the large model B to obtain a second prediction result, and the second prediction result is normalized and converted into a unified score rating vector, which is recorded as the second rating vector.
[0089] Exemplarily, the second scoring vector is such as [consultation package: 0.4, complaint: 0.5, ...].
[0090] Next, we introduce the inference model C.
[0091] For example, Figure 4 The flowchart of constructing and training the inference model C provided in the embodiment of the present application is as follows. Figure 4 As shown, the following steps are included:
[0092] S31, calling the second scoring vector output by the reasoning of the large model B fine-tuned in the production environment and the first scoring vector output by the reasoning of the customer portrait classification model A.
[0093] S32 can select step 3A dual-model fusion training or step 3B intelligent agent orchestration real-time reasoning as one of the two routes for the actual implementation of the reasoning large model C based on the customer's computing resources and actual development scenarios.
[0094] If the computing power is sufficient and the implementation time period is sufficient, step 3A can be selected to perform dual-model fusion training to determine the large inference model C.
[0095] For example, Figure 5 The schematic diagram of the process of dual-model fusion training and reasoning large model C provided in step 3A. Figure 5 As shown in the figure, the process of merging the dual model reasoning results is as follows:
[0096] Sa1, perform vector fusion on the first scoring vector and the second scoring vector, determine the weight of each scoring vector, and form a comprehensive scoring vector c.
[0097] In some implementations, a linear weighting algorithm may be used to concatenate the first rating vector a and the second rating vector b to form a comprehensive rating vector c, and the linear weighting algorithm is as follows: c = w1·a+w2·b. Where w1 and w2 are linear weighting weights, and the values of w1 and w2 may be preset or adjusted based on experience, wherein the first rating vector a comes from the current customer profile classification score inferred by the customer profile model; the second rating vector b comes from the current customer sentiment feedback score inferred by the fine-tuned large model based on dialogue reasoning.
[0098] Exemplarily, a linear weighted algorithm is used to merge the first scoring vector a and the second scoring vector b, which is implemented by the following steps:
[0099] Sa11, assuming a = [0.2, 0.4, 0.5], where the example vector a represents the customer portrait evaluation values of three scoring dimensions generated based on the customer portrait model, such as purchase intention score 0.2, budget margin score 0.4, and existing order replacement intention score 0.5;
[0100] b = [0.3, 0.2, 0.4], where the example vector b represents the real-time score values of the three customer rating dimensions generated by the fine-tuned large model based on the current context conversation, such as purchase intention score 0.3, budget margin score 0.2, and existing order replacement intention score 0.4.
[0101] Sa12, determine the weights, select the values of weights w1 and w2 according to preset experience, and the weight values satisfy w1+w2=1, ensuring that the combined score vector c is still in the same range. If normalization is not required, this condition may not be met.
[0102] Sa13, calculate the comprehensive score vector c: perform weighted summation on each corresponding element of vectors a and b to obtain the comprehensive score vector c.
[0103] The formula of the linear weighted algorithm is as follows: c = w1·a+w2·b; the specific operation is as follows: Assume w1 = 0.6 and w2 = 0.4, which means that the customer portrait weight is 60% and the customer dialogue fine-tuning model is 40%; then: ci = 0.6·ai+0.4·bi;
[0104] The specific calculation is as follows:
[0105] c1=0.6·0.2+0.4·0.3=0.12+0.12=0.24
[0106] c2=0.6·0.4+0.4·0.2=0.24+0.08=0.32
[0107] c3=0.6·0.5+0.4·0.4=0.30+0.16=0.46
[0108] Therefore, the merged vector c is: c = [0.24, 0.32, 0.46]; this process can be summarized as a weighted linear combination of two vectors, and each element ci of vector c is the weighted sum of the score vectors ai and bi generated by the two original models. Finally, the final vector c obtained after offline weighting is the real-time score value of the customer's three rating dimensions, such as purchase intention score 0.24, budget margin score 0.32, and existing order replacement intention score 0.46.
[0109] In some implementations, a linear weighted average may be performed to combine the first scoring vector a and the second scoring vector b to determine a comprehensive scoring vector c.
[0110] In some embodiments, a multi-layer perceptron (MLP) algorithm can be used to assign weights to each score vector. MLP is a feed-forward neural network that is typically used to process fixed-size input vectors. MLP learns the complex relationship between input features through a series of linear transformations and nonlinear activation functions. MLP performs weighted fusion by learning a set of fixed weights.
[0111] The specific steps of using MLP to assign weights to each rating vector are as follows:
[0112] Sa31, concatenates multiple input rating vectors into a larger rating vector.
[0113] Sa32, processes the concatenated score vector through a multi-layer perceptron to learn the complex relationship of the input features.
[0114] Sa33, outputs the fused comprehensive score vector c through the last layer of MLP.
[0115] The scoring vector dimension is calculated in the input layer of MLP; ReLU activation and Dropout regularization are performed in the hidden layer of MLP; Softmax classification is performed in the output layer of MLP to output the probability of each intent.
[0116] Sa4, train the inference large model C based on the comprehensive score vector.
[0117] The comprehensive rating vector is used as input, and the inference model performs inference based on the comprehensive rating vector and outputs the customer's intention classification. The real intention of the historical data is used as the label, and the cross entropy loss is minimized through iterative calculation to obtain the trained inference model C.
[0118] In some implementations, the inference model performs inference based on the comprehensive score vector, and the number of customer intent classifications outputted is M. An intent decision strategy may be formulated, and the intent decision strategy includes one or a combination of the following:
[0119] Select the top-K intents based on the M intent classifications output by the large inference model C; filter out the low-confidence intents in the top-K intents by setting a threshold (such as 0.6) to obtain multiple intents that meet the requirements; or process multiple intents in a set manner, such as sorting the M intents by priority or executing the M intents in parallel.
[0120] In step 3A, the reasoning capabilities of the customer portrait classification model A and the fine-tuned large model B are integrated to enhance the reasoning performance of customer intent through the reasoning and prediction capabilities of the dual models. Using step 3A requires additional model fusion training costs, but it can improve the reasoning efficiency of the large reasoning model C in actual production deployment.
[0121] When computing power is limited and rapid implementation is required, you can choose step 3B to understand the intention based on the dual model and determine the inference large model C through the agent orchestration prompt word engineering.
[0122] For example, Figure 6 A schematic diagram of implementing the inference model C by arranging the prompt word engineering through the agent orchestration tool provided for step 3B. Step 3B includes adjusting the output of the customer profile classification model A and the fine-tuning large model B through agent orchestration and prompt word engineering in the actual inference scenario, and arranging the output weights of the intent classification of the customer profile classification model A and the fine-tuning large model B. Among them, the prompt word engineering can be implemented through the agent orchestration tool or the large language model application development framework. Figure 6 As shown, step 3B includes the following steps:
[0123] Sb1, use BPMN to design a flowchart and define the key nodes of the task process of the reasoning model C.
[0124] In some implementations, an agent orchestration tool may be used to determine key nodes in the reasoning large model C process.
[0125] Exemplarily, the agent orchestration tool can be a self-developed process orchestration tool or a large model application development framework, including Langchain, semantic-kernel, etc.
[0126] First, determine the key nodes of the task flow of the inference large model C, including the customer's original intention input node, the inference call node of the customer portrait classification model A, the inference call node of the fine-tuning large model B, the weight merging decision node (based on the prompt word engineering configuration weight), the behavior execution node (intelligent agent tool call) and the multi-module combination process node.
[0127] It can be understood that the weighted merging decision node is also the prompt word engineering node. It describes the task requirements of the large inference model C for merging reasoning through natural language. The weighted merging decision node is determined by the weight configured based on the prompt word engineering.
[0128] The behavior execution node is a large-scale inference model agent node that can make external Assistant / Tool API calls. For example, "query the weather forecast" is a behavior execution node of an agent tool, because the large-scale inference model C itself cannot generate the current real-time Beijing weather. After the agent tool node accesses weatherdt.com to obtain weather information, the large-scale inference model C continues to reason; for the customer ordering scenario, "query customer business data" or "initiate the opening of a customer order" is also a call to an external API capability.
[0129] Sb2, for the subsequent weight merging decision points, converts the output results of the inference call node of the customer portrait classification model A into a descriptive statement that can be semantically understood by the fine-tuned large model B.
[0130] Through the prompt word engineering + intelligent agent node, the inference model C performs decision reasoning based on the prompt words and according to the user's text requirements, and executes subsequent operations.
[0131] For example, the customer's purchase intention Score (rating value) is 7 (0 is the lowest and 10 is the highest); the customer's current experience feedback Score is 4 (0 is very dissatisfied and 10 is very satisfied).
[0132] The multiple prompt word engineering nodes of the process orchestration constitute a process of weighted merging of the first scoring vector output by the customer portrait classification model A and the second scoring vector output by the fine-tuned large model B.
[0133] Sb3, determine the weight merging decision node and configure the prompt word project.
[0134] In order to handle complex and implicit intentions, prompt word engineering is configured for the output of customer portrait classification model A and the output of fine-tuned large model B, and different tendency weights are adjusted in different business scenarios.
[0135] In some implementations, advanced prompt word engineering techniques such as thought chain can be used to define, including: configuring a first prompt word based on the inference call node of the customer portrait classification model, the first prompt word is used to guide the system to call the customer portrait classification model to obtain a first scoring vector; the first scoring vector includes a score result for the customer portrait classification prediction;
[0136] A second prompt word is configured based on the inference call node of the fine-tuned large model, and the second prompt word is used to guide the system to call the fine-tuned large model to obtain a second scoring vector, and the second scoring vector includes a score result of sentiment reasoning on the customer context;
[0137] The third prompt word is configured based on the weight merging decision node, and the third prompt word is used to guide the system to arrange the output weights of the customer portrait classification model and the intention classification of the fine-tuning large model for the current different customer status information according to the first scoring vector and the second scoring vector, so as to judge the customer's intention tendency;
[0138] A fourth prompt word is configured based on the behavior execution node, and the fourth prompt word is used to guide the system to output a selectable operation function;
[0139] The fifth prompt word is configured based on the multi-module combination process node. The fifth prompt word is used to guide the system to combine the first prompt word, the second prompt word, the third prompt word and the fourth prompt word together to arrange the output weight of the customer portrait classification model and fine-tune the intent classification of the large model. For example, a prompt word engineering reference part example is as follows:
[0140]
[0141] Sb4, determine the agent behavior library, which is used to preset tools and atomic capability services for registering business operation behaviors.
[0142] In some embodiments, the business operation behavior tools pre-registered in the agent behavior library include one or a combination of the following types:
[0143] Ordering, including calling the service activation API to create and change orders;
[0144] Business consulting, including intelligent agents in RAG scenarios, to search, enhance and generate business knowledge;
[0145] In some implementations, the atomic capability services pre-registered in the agent behavior library include one or a combination of the following capabilities:
[0146] Graph recognition question and answer; and / or recommendation services, including calling external models of collaborative filtering algorithms (customer-based collaborative filtering) to perform customer preference recommendation tasks.
[0147] Sb5, based on the agent behavior library, selects the corresponding agent combination according to the intention, arranges the behavior weights and executes the behavior.
[0148] Exemplarily, customer portrait model A concludes that the customer is a person who is prone to complain, and the SCORE of "dissatisfaction" is high; fine-tuning large model B concludes that the probability SCORE of the customer complaining about "slow network performance" is high; then based on the weighted results of step 3A or the prompt word engineering combination results of step 3B, the conclusion is drawn: [slow network 0.8, dissatisfied 0.9]; for customer portrait classification and emotion "slow network + dissatisfied", the corresponding agent combination can be selected to solve the above results after customer evaluation. The agent combination includes network diagnosis agent and soothing agent. The network diagnosis agent is used for dedicated line configuration test calls, routing detection calls, ARP detection interface calls, and link status detection calls; the soothing agent is used to generate analysis results + generate apology texts + generate compensation suggestions.
[0149] Compared with the linear weighted or MLP neural network calculation used in step 3A, step 3B is relatively flexible. As long as the user adjusts the natural language text describing the requirements in the prompt word node of the process, it is easier to understand and operate for the end user.
[0150] Optionally, the large reasoning model C can also use integrated learning to vote on the results of decision reasoning of multiple reasoning chains and select the result with the most votes as the correct intention.
[0151] If the COT-SC (Self-Consistency with Chain of Thought) method is used in the prompt word engineering arrangement, it is equivalent to having multiple reasoning chains for decision reasoning results, and voting for the most correct results. This method can effectively improve the accuracy of reasoning of the large reasoning model C.
[0152] For example, there are 5 large model reasoning routes, 3 of which lead to the conclusion that "the customer will buy Package A", and the other 2 are "buy Package B" and "do not buy"; then the voting result is "the customer will buy Package A", and subsequent measures are taken.
[0153] The method provided in the embodiment of the present application also obtains customer feedback based on the behavioral results of the inference large model C, and continuously optimizes the customer portrait classification model A and fine-tunes the large model B.
[0154] Figure 7 Schematic diagram of result feedback and continuous optimization in the method provided in the embodiment of the present application. Figure 7 As shown, the following steps are included:
[0155] S41, based on the context text of the customer interaction, obtain the results of the interactive behavior of the customer feedback, and use the results of the interactive behavior of the customer feedback as the feature weight retraining of the customer portrait classification model A in steps S23-S24 and the training corpus for fine-tuning and aligning the large model B in steps S31-S34, including feedback alignment corpus generation, customized fine-tuning or preference training corpus generation.
[0156] The customer portrait classification model dynamically adjusts feature weights based on the training corpus to enhance or weaken the influence of the corresponding features.
[0157] Among them, customer feedback includes positive feedback and negative feedback. The results of customer feedback interaction behavior include: clear operation choices made on the front-end interface, such as clicking "satisfied" or "unsatisfied"; emotion recognition of customer feedback response sentences through fine-tuning the large model B, and context-guided feedback during the conversation, such as positive words ("very good") vs negative words ("no"); feedback results obtained through front-end interaction mode monitoring, such as rapid and continuous replies: "may indicate impatience", such as long silence: "may indicate confusion or dissatisfaction".
[0158] S42, fine-tuning the large model B to automatically generate new annotations based on the results of the interactive behavior feedback from customers. This is achieved through the following steps:
[0159] S421, tracing back multiple turns of contextual conversations;
[0160] S422, using the SpanBERT algorithm of the fine-tuned large model B reasoning or NLP model to identify pronoun references and complete reference resolution;
[0161] S423, uses the ToD-BERT algorithm to track the conversation state. The ToD-BERT algorithm is a general Bert model for understanding user intent;
[0162] S424, causing the fine-tuned large model B to automatically generate new annotations;
[0163] Exemplarily, GPT4 or QWEN110B is used to capture the transition from "unsatisfied" -> "problem solved" -> "satisfied".
[0164] S425, fine-tune the large model B, mark the feedback results that cannot be clearly marked as "fuzzy", and record the feedback results.
[0165] Judgment records,For a small number of interaction result records marked as “fuzzy”, a manual intervention process is used to annotate feedback to improve the quality of feedback.
[0166] S43, retraining customer portrait classification model A, including data integration and feature expansion, feature quantification of feedback annotation, automatic feature engineering, AutoML automatic machine learning training, incremental learning and online update.
[0167] S44, fine-tuning the pre-trained large model with the conversation interaction results and the new annotations to obtain an updated fine-tuned large model B, such as a language large model or a customer service interaction large model, including: task-oriented fine-tuning and re-fine-tuning. The task-oriented fine-tuning includes the customer feedback dialogue after the feedback annotation of this round.
[0168] S45, retrain the large inference model C to adapt to the updated customer portrait classification model A and fine-tune the large model B to improve the accuracy of inference.
[0169] The intent recognition method based on a large language model and a customer portrait classification model provided in the embodiment of the present application can be applied to a customer dialogue system based on a cloud platform, receive customer input, select an inference implementation route to recognize customer intent based on the computing power and production environment conditions provided by the customer, and execute subsequent tasks; call the large language model for semantic analysis, and at the same time obtain and update the customer portrait from the database, calculate the final intent recognition result through a fusion algorithm, and then drive the execution of corresponding business logic, such as providing customized suggestions or triggering specific service processes.
[0170] Example 1
[0171] Example 1 is the application of an intent recognition method based on a large language model and a customer portrait classification model in an operator's cloud computer ordering customer service system.
[0172] When customers are chatting with AI customer service, the system pre-processes the customer's sentences into the fine-tuned large model B based on the large language model for preliminary intent recognition, and at the same time builds a customer profile based on the user's historical purchase records, browsing habits and other information.
[0173] The inference model C integrates these two parts of information to make the final intention judgment. The customer service system can accurately identify the user's intention to inquire about the product, feedback problems or provide after-sales service, so as to provide accurate responses.
[0174] Figure 8 This is a schematic diagram of the process engine framework for implementing agent orchestration provided in Example 1 of the present application. Figure 8 As shown, through a process engine framework that implements agent orchestration, the customer intention classification model A and the fine-tuned large model B perform real-time dual-model reasoning on the customer's conversation, obtain the customer's complete intention behavior, and then call related subsequent business operations.
[0175] S81, the customer inputs interaction information.
[0176] S82, query the customer's related orders and background information according to the customer ID, including whether the customer has expired, new orders, new accounts, customer types, etc.
[0177] S83, based on the customer's order status, account status and customer type, determine whether it is a new customer, a new order or an order expired, etc.; perform customer portrait reasoning (label-based reasoning), perform customer portrait classification prediction based on the queried customer-related information, and perform ordering customer portrait classification prediction.
[0178] S84, call the customer portrait classification prediction API to obtain the customer portrait classification score.
[0179] In some implementations, a customer portrait classification model may be called to generate a classification score (Score) of the customer's ordering intention based on the customer's existing packages and customer background information.
[0180] Exemplarily, the Adaboost classification model is called to classify customer portraits, and a classification score (Score) of the customer's ordering intention is generated based on the customer's existing packages and customer background information.
[0181] S85, reasoning about customer context interaction text.
[0182] You can use a large model fine-tuned based on Qwen14B for text reasoning services.
[0183] Here, we use the Qwen14B large model for fine-tuning. The Qwen14B large model is a large model with fewer parameters (14B = 14 billion parameters). After evaluation, it is sufficient to complete the efficient fine-tuning training task and achieve the availability of intent classification with relatively less computing power requirements. We use the Qwen14B fine-tuning large model to perform semantic recognition on customer contextual dialogues, determine the customer's current emotional feedback, and output the current emotional feedback score.
[0184] S86, implement intent classification based on the prompt word engineering application deployed on the inference large model C (Qwen70B model).
[0185] Construct prompt words based on the output of the two models and assign weights: Combine the customer portrait classification model to generate a classification score (Score) for the customer's ordering intention and the current sentiment feedback score output by the fine-tuned large language model, compile prompt words and assign weights.
[0186] It should be noted that the reasoning large model C chooses the Qwen70B (70 billion parameters) large model with more parameters because the large model with larger parameters has better prompt word understanding and logical reasoning tasks, can understand the thinking chain definition of the prompt engineering output by the dual models, and perform high-quality reasoning.
[0187] In Example 1 of the present application, the prompt word engineering helps the reasoning large model C to understand the judgment results generated by the previous dual models through a prompt description of a thought chain.
[0188] The large inference model C performs intent inference. Based on the intent classification results of the prompt word project and the customer portrait classification model, it calls the corresponding tool capability API operation to implement the subsequent task behaviors of the classification results as follows:
[0189] Intent classification result A: Customer service recommendation, such as customer service related API.
[0190] Intent classification result B: purchase recommendation, such as purchase recommendation API.
[0191] Intent classification result C: Cloud computer business subscription, such as reasoning based on the Owen70B large model.
[0192] Intent classification result D: other business recommendations, such as other recommended APIs.
[0193] S87, based on the prompt word engineering, provides feedback and interaction to customers based on the operation results of different execution classification results of S86.
[0194] Based on the large model reasoning service C (based on Qwen70B), it processes customer interaction feedback.
[0195] S88, record customer feedback for evaluation.
[0196] The large model can be used for automatic evaluation, and some results can be manually evaluated and recorded.
[0197] The intent recognition method based on a large language model and a customer portrait classification model provided in the embodiment of the present application combines the multi-dimensional information processing capabilities of fine-tuning the large model and the customer portrait classification model to realize real-time personalized intent recognition algorithms and model intent recognition reasoning, has a self-learning and optimization feedback mechanism, provides optional dual-route reasoning implementation, and can be based on dual-model fusion training reasoning or real-time reasoning based on a process orchestration engine according to the customer's actual computing resources and production links.
[0198] The intent recognition method based on the large language model and the customer portrait classification model provided in the embodiment of the present application can greatly improve the accuracy and efficiency of intent recognition. By combining the two models for intent recognition, the fine-tuned large language model can better capture the sentiment and intent information of the semantic text in the customer context, thereby significantly improving the accuracy of intent recognition and reducing misjudgments and missed judgments. Through the pre-trained customer portrait classification model, the prediction and classification information based on the clear feature information in the customer context is improved. The combination of the two models improves personalized services and enhances customer experience, can better understand customer needs and preferences, realize personalized services, and thus improve customer experience.
[0199] The intent recognition method based on a large language model and a customer portrait classification model provided in the embodiment of the present application can provide more personalized services or recommendations that meet customer expectations, thereby enhancing customer satisfaction and loyalty.
[0200] The embodiment of the present application proposes an intention recognition system based on a large language model and a customer portrait classification model, based on the method provided in the above embodiment, including: an information acquisition module, used to obtain the customer's current relevant orders and background information according to the customer ID information; a customer portrait classification model, used to obtain a first scoring vector according to the relevant orders and background information; the first scoring vector is used to indicate the current customer portrait classification score; a fine-tuning large model, used to perform semantic recognition on the context dialogue of the input customer dialogue to obtain a second scoring vector; a fusion module, used for the second scoring vector to indicate the customer's current emotional feedback score; the fine-tuning large model is obtained by fine-tuning the large language model; the first scoring vector and the second scoring vector are fused to obtain a comprehensive scoring vector; the comprehensive scoring vector is used to indicate different tendency weights in different business scenarios; an inference large model, used to perform inference based on the comprehensive scoring vector and output the customer's intention; an intention-driven module, used to drive the execution of corresponding business logic according to the customer's intention, and the corresponding business logic includes providing customized suggestions or triggering specific service processes.
[0201] The system's self-learning and optimization capabilities: The system has the ability to self-learn and optimize, and can adapt to changing customer needs, ensuring that the system can develop as customer behavior changes and market trends.
[0202] Multi-dimensional feedback capture fully understands the customer's true feelings through implicit signals such as conversation features, emotional vocabulary, and interaction patterns. In this way, even if the customer is unwilling to express it explicitly, the system can capture subtle emotional changes.
[0203] Dynamic interaction adjustment: Based on the real-time interpretation of customer emotions and preferences, the system can dynamically adjust the way of inquiry during the interaction, such as being concise and to the point for impatient customers and patiently guiding confused customers. This adaptability greatly improves the customer experience.
[0204] The continuous evolution of the dual model is based on a continuous closed loop of semi-supervised learning. Through continuous feedback collection and automatic labeling of large language models, most feedback labeling is completed. For ambiguous samples that are difficult to judge, manual intervention is used to improve the labeling quality and continuously expand the training set.
[0205] The customer portrait classification model dynamically adjusts feature weights based on each interaction. Through technologies such as gradient boosting trees and SHAP values, the system can accurately enhance or weaken the influence of certain features, making the portrait closer to the customer's actual status.
[0206] The continuous evolution of the fine-tuned large model is not just a one-time fine-tuning, but a continuous optimization based on the conversation interaction results and feedback annotations that occur continuously in production, so that the fine-tuned large model is not only optimized at the technical level, but also aligned with professional staff in terms of behavioral standards.
[0207] An embodiment of the present application provides an electronic device, comprising: at least one memory for storing programs; and at least one processor for executing the programs stored in the memory. When the programs stored in the memory are executed, the processor is used to execute the above-mentioned method for predicting ecological indicators.
[0208] An embodiment of the present application provides a computer storage medium, in which instructions are stored. When the instructions are executed on a computer, the computer executes the above-mentioned method for predicting ecological indicators.
[0209] It is understood that the processor in the embodiments of the present application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.
[0210] The method steps in the embodiments of the present application can be implemented by hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC.
[0211] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted by a computer-readable storage medium. The computer instructions can be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital customer line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integration. The available medium can be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state drive (SSD)), etc.
[0212] It should be understood that the various numerical numbers involved in the embodiments of the present application are only used for the convenience of description and are not used to limit the scope of the embodiments of the present application.
Claims
1. An intention recognition method based on a large language model and a customer portrait classification model, characterized in that: The method comprises: Obtain the customer's current related orders and background information based on the customer ID information; The customer portrait classification model obtains a first scoring vector according to the relevant orders and background information; the first scoring vector is used to indicate the current customer portrait classification score; The fine-tuned large model performs semantic recognition on the context dialogue of the input customer dialogue to obtain a second scoring vector; the second scoring vector is used to indicate the customer's current emotional feedback score; the fine-tuned large model is obtained by fine-tuning the large language model; The first scoring vector and the second scoring vector are merged to obtain a comprehensive scoring vector; the comprehensive scoring vector is used to indicate different tendency weights in different business scenarios; The inference model performs inference based on the comprehensive scoring vector and outputs the customer's intention; The corresponding business logic is executed according to the customer's intention, and the corresponding business logic includes providing customized suggestions or triggering specific service processes.
2. The method according to claim 1, characterized in that The step of fusing the first scoring vector and the second scoring vector to obtain a comprehensive scoring vector includes: The first scoring vector and the second scoring vector are concatenated using a linear weighted algorithm to form the comprehensive scoring vector.
3. The method according to claim 1, characterized in that The step of fusing the first scoring vector and the second scoring vector to obtain a comprehensive scoring vector includes: The comprehensive scoring vector is obtained by weighted fusion of the first scoring vector and the second scoring vector by learning a set of fixed weights using a multilayer perceptron.
4. The method according to any one of claims 1 to 3, characterized in that: The inference model performs inference based on the comprehensive scoring vector and outputs the customer's intention classification, including: The number of customer intent categories is M, and the Top-K intents with the highest scores are selected according to the M intent categories; or Filtering the low-confidence intents among the M intents by setting a threshold to obtain K intents that meet the requirements; or sorting the M intentions by priority; Or execute the M intentions in parallel.
5. The method according to claim 1, characterized in that The step of fusing the first scoring vector and the second scoring vector to obtain a comprehensive scoring vector includes: Through agent arrangement and prompt word engineering, the output weights of the first scoring vector and the second scoring vector are arranged.
6. The method according to claim 1 or 5, characterized in that: The step of arranging the output weights of the first scoring vector and the second scoring vector by agent arrangement and prompt word engineering includes: Determine the key nodes of the task flow of the inference large model, the key nodes include: customer original intention input node, customer portrait classification model inference call node, fine-tuning large model inference call node, weight merging decision node, behavior execution node and multi-module combination process node; Convert the output of the inference call node of the customer portrait classification model into a description statement based on semantic understanding of the fine-tuned large model; The prompt word engineering is configured for the first scoring vector and the second scoring vector, and different tendency weights are adjusted in different business scenarios.
7. The method according to claim 1 or 5, characterized in that: The driving and executing corresponding business logic according to the customer intent classification includes: Determine an agent behavior library, wherein the agent behavior library pre-sets tools and / or atomic capability services for registering business operation behaviors; According to the customer intention classification, a corresponding agent combination is selected in the agent behavior library, and the agent combination is used to choreograph and execute behaviors.
8. The method according to claim 1, characterized in that The method further comprises: Obtaining the result of the interactive behavior of the customer feedback based on the context text of the customer interaction; Using the results of the interactive behavior of the customer feedback as training corpus for retraining the feature weights of the customer portrait classification model; The customer portrait classification model retrains the feature weights according to the training corpus; including data integration and feature expansion, feature quantification of feedback annotations, automatic feature engineering, AutoML automatic machine learning training, incremental learning and online updating.
9. The method according to claim 1 or 8, characterized in that: The method further comprises: The fine-tuned large model automatically generates new annotations according to the results of the interactive behavior fed back by the customer; The conversation interaction results and new annotations are used as input-output pairs to continuously fine-tune the fine-tuned large model to align the conversation and intent recognition.
10. An electronic device comprising: at least one memory for storing a program; At least one processor is used to execute the program stored in the memory, and when the program stored in the memory is executed, the processor is used to execute the method according to any one of claims 1 to 8.
Citation Information
Patent Citations
Intelligent conversation recommendation method and device based on knowledge graph
CN113127646A
Dialogue emotion recognition method and system based on character character dynamic modeling
CN117195914A
Emotion support dialogue generation method, system and device based on thinking chain reasoning
CN117932041A
Usage of emotion recognition in trade monitoring
US20240127335A1
Cited By
Deep learning driven order dispatching management system and method, and storage medium
CN119443734A
Consumer protection method and device based on artificial intelligence and storage medium
CN119539821A
Large language model fingerprint adding method and device based on weight superposition
CN119598433A
Large model-based interpretable emotion recognition method
CN119943095A
Interaction method, system, equipment and medium
CN120196214A
