Data processing method and device and electronic equipment

Abstract

The invention provides a data processing method and device and electronic equipment, and the method comprises the steps: inputting obtained user problem description information and user input information into a data processing model, and obtaining an output vector; selecting a target data vector matched with the output vector from a preset question and answer library, and determining question reply information corresponding to the target data vector in the preset question and answer library; the preset question and answer library comprises a plurality of question reply pairs and data vectors corresponding to the question reply pairs; the question reply pair comprises question description information and question reply information; and generating target reply information corresponding to the user question description information based on the question reply information corresponding to the target data vector. In the mode, based on the user question description information and the user input information, a more accurate question and answer record can be recalled from the preset question and answer library, and the question reply information is generated based on the question and answer record, so that the mode can accurately understand the user intention to give a correct reply, and the customer service question reply efficiency is improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of data processing design, and in particular to a data processing method, device and electronic device. Background Art

[0002] In the era of rapid development of Internet technology, the number of customer consultations faced by enterprises has shown an explosive growth trend. This growth has posed a huge challenge to the traditional manual customer service model, which is limited by human resources and response speed and cannot meet the increasing customer service needs.

[0003] In the related art, a method for answering user questions based on a large language model is provided. Before answering user questions, this method first retrieves questions related to the user questions from the question-and-answer knowledge base prepared by the product, and then provides the user questions to the large language model, allowing the large language model to generate answers related to the product. However, sometimes it is impossible to retrieve accurate records simply by relying on the semantic information of user questions. There may be a logical relationship between the user's questions and the records stored in the question-and-answer knowledge base. In addition, even for the same question description, the responses that need to be given may be different due to the different status of the user itself. Summary of the invention

[0004] The purpose of the present disclosure is to provide a data processing method, device and electronic device to effectively improve the response efficiency of customer service, thereby reducing operating costs on the basis of providing users with timely responses.

[0005] In a first aspect, the present disclosure provides a data processing method, which includes: obtaining user question description information and user input information; wherein the user input information is used to indicate the information sending status and log information corresponding to the user question description information; inputting the user question description information and / or the user input information into a preset trained data processing model to obtain an output vector; selecting a target data vector matching the output vector from a preset question and answer library, and determining the question answer information corresponding to the target data vector in the preset question and answer library; wherein the preset question and answer library includes multiple question and answer pairs, and data vectors corresponding to the question and answer pairs; the question and answer pairs include question description information and question answer information corresponding to the question description information; based on the question answer information corresponding to the target data vector, generating target answer information corresponding to the user question description information.

[0006] In a second aspect, the present disclosure provides a data processing device, which includes: an information acquisition module, used to acquire user question description information and user input information; wherein the user input information is used to indicate the information sending status and / or log information corresponding to the user question description information; a vector determination module, used to input the user question description information and the user input information into a preset trained data processing model to obtain an output vector; a reply determination module, used to select a target data vector matching the output vector from a preset question and answer library, and determine the question reply information corresponding to the target data vector in the preset question and answer library; wherein the preset question and answer library includes multiple question and answer pairs, and data vectors corresponding to the question and answer pairs; the question and answer pairs include question description information and question reply information corresponding to the question description information; and a reply generation module, used to generate target reply information corresponding to the user question description information based on the question reply information corresponding to the target data vector.

[0007] In a third aspect, the present disclosure provides an electronic device, which includes a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above-mentioned data processing method.

[0008] In a fourth aspect, the present disclosure provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned data processing method.

[0009] The embodiments of the present disclosure bring the following beneficial effects:

[0010] The present disclosure provides a data processing method, device and electronic device, first obtaining user question description information and user input information; wherein the user input information is used to indicate the information sending status and / or log information corresponding to the user question description information; then inputting the user question description information and the user input information into a preset trained data processing model to obtain an output vector; then selecting a target data vector matching the output vector from a preset question and answer library, and determining the question answer information corresponding to the target data vector in the preset question and answer library; wherein the preset question and answer library includes multiple question answer pairs and data vectors corresponding to the question answer pairs; the question answer pairs include question description information and question answer information corresponding to the question description information; then based on the question answer information corresponding to the target data vector, generating the target answer information corresponding to the user question description information. In this method, based on the user question description information and the user input information, more accurate question and answer records can be recalled from the preset question and answer library, and question answer information can be generated based on the question and answer records, so that the method can accurately understand the user's intention to give a correct reply, and improve the efficiency of customer service question reply.

[0011] Other features and advantages of the present disclosure will be set forth in the following description, or some features and advantages may be inferred or unambiguously determined from the description, or may be learned by implementing the above-mentioned technology of the present disclosure.

[0012] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, the following specifically cites preferred implementation modes and describes them in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the specific embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the specific embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1 A flowchart of a data processing method provided by an embodiment of the present disclosure;

[0015] Figure 2 An architectural diagram of a recommended reply generation provided by an embodiment of the present disclosure;

[0016] Figure 3 A schematic diagram of the structure of a data processing device provided in an embodiment of the present disclosure;

[0017] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all the embodiments. The components of the embodiments of the present disclosure described and shown in the drawings here can be arranged and designed in various different configurations.

[0019] Therefore, the following detailed description of the embodiments of the present disclosure provided in the accompanying drawings is not intended to limit the scope of the present disclosure claimed for protection, but merely represents selected embodiments of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present disclosure.

[0020] In the era of rapid development of Internet technology, the number of customer consultations faced by enterprises has shown an explosive growth trend. This growth has posed a huge challenge to the traditional manual customer service model, which is limited by human resources and response speed and cannot meet the increasing customer service needs. In order to meet this challenge, intelligent customer service systems have emerged and have gradually become a key tool for enterprises to improve service efficiency and reduce operating costs.

[0021] The feedback system in the related technology is the traditional customer service model. Users of each product submit their own tickets, and customer service staff handle the user's questions one by one and respond. Limited by human resources, when encountering a large number of ticket problems, the response speed is greatly reduced, and the user experience is relatively poor. Since the large language model LLM (for example, chatgpt, etc.) has made a significant breakthrough in the understanding and processing capabilities of natural language, it can effectively understand user intentions and generate more appropriate and natural responses. Therefore, it has also begun to be widely used in customer service scenarios to improve response efficiency. However, due to the "illusion" problem of LLM, and many user questions are also strongly related to the product, it is not possible to directly use LLM to respond to user questions. For this reason, a retrieval-augmented generation (RAG) technology has emerged. The core idea is to retrieve questions related to the user's questions from the question-and-answer knowledge base prepared by the product before responding to the user's questions, and then provide the user's questions and related FAQs as context to LLM, so that it can generate more accurate and product-related answers.

[0022] However, in actual application, this process still has two problems:

[0023] 1. The product needs to prepare a Q&A knowledge base in advance. In order to ensure the response effect, the Q&A knowledge base needs to be maintained and updated due to timeliness and continuous product improvement, which brings certain maintenance costs.

[0024] 2. Currently, RAG technology searches for relevant questions in the question-and-answer knowledge base based on user questions. This step is basically based on the semantic vector model. It calculates the semantic relevance between the user question and each record in the question-and-answer knowledge base, and finally filters the results. However, in actual applications, users' questions are all kinds of strange, and sometimes it is impossible to retrieve accurate records simply by relying on semantic information. User questions and records stored in the question-and-answer knowledge base may have a logical relationship. In addition, even for the same question description, the responses that need to be given may be different due to the different status of the user himself.

[0025] Based on the above problems, the embodiments of the present disclosure provide a data processing method, device and electronic device, which can be applied in question answering scenarios.

[0026] In order to facilitate understanding of the embodiments of the present disclosure, a data processing method disclosed in the embodiments of the present disclosure is first described in detail. Figure 1 As shown, the method comprises the following specific steps:

[0027] Step S102, obtaining user problem description information and user input information; wherein the user input information is used to indicate the information sending status and / or log information corresponding to the user problem description information.

[0028] The user problem description information is the problem text submitted by the user through the user device, which may be a mobile phone, a computer or a tablet computer, etc. The user input information may include one or more of the following: user status information, user behavior information and recorded log information; wherein the user status information is used to indicate the device status when the user sends the user problem description information, the user behavior information is used to indicate the behavior performed on the user device when or before the user sends the user problem description information, and the log information may be information recording some errors or vulnerabilities, etc.

[0029] In a specific implementation, when the user submits the user question description information through the user device, the user input information will be submitted at the same time, so that the question can be answered later based on the user question description information and the user input information.

[0030] Step S104, inputting the user problem description information and the user input information into a preset trained data processing model to obtain an output vector.

[0031] The above data processing model can be a neural network model or a deep learning model, etc., and the specific model can be determined according to the research and development needs. The data processing model is pre-trained, and the training method of the data processing model can be determined according to the research and development needs. The function of the data processing model is to convert the input data into an embedded vector by analyzing the input data, and output the embedded vector, which is also the above output vector.

[0032] Step S106, select a target data vector that matches the output vector from a preset question and answer library, and determine the question answer information corresponding to the target data vector in the preset question and answer library; wherein the preset question and answer library includes multiple question-answer pairs, and data vectors corresponding to the question-answer pairs; the question-answer pairs include question description information and question answer information corresponding to the question description information.

[0033] In specific implementation, the above-mentioned preset question-answer library includes a large number of question-answer pairs, and data vectors corresponding to each question-answer pair; each question-answer pair includes question description information and question-answer information corresponding to the question description information. After obtaining the output vector corresponding to the user's question description information, it is necessary to calculate the similarity between the output vector and the data vector corresponding to each question-answer pair in the preset question-answer library, and then based on the preset rules, select one or more data vectors with high similarity to the output vector from the preset question-answer library as the target data vector, and then determine the target question-answer pair corresponding to the target data vector, and determine the question answer information in the target question-answer pair as the question answer information corresponding to the target data vector.

[0034] The above-mentioned preset rules can be determined according to research and development needs. For example, the preset rule can be to select a data vector with the highest similarity to the output vector from the preset question and answer library as the target data vector, or to select multiple data vectors with high similarity to the output vector from the preset question and answer library as the target data vector, or to select multiple data vectors of different categories with high similarity to the output vector from the preset question and answer library as the target data vector, etc.

[0035] In the related art, the relevant records are retrieved from the preset question and answer library based on the semantic vector, but due to the diversity of user question description information, and the same question description information may correspond to different reply records in different states, simple semantic matching cannot handle these problems. The present invention can recall some more accurate question and answer records from the preset question and answer library based on the user question description information and user input information.

[0036] Step S108, based on the question answer information corresponding to the target data vector, generate target answer information corresponding to the user question description information.

[0037] In specific implementation, if the target data vector corresponds to one piece of question answering information, the question answering information corresponding to the target data vector can be determined as the target answering information corresponding to the user's question description information; if the target data vector corresponds to one or more pieces of question answering information, the target data vector needs to be integrated and analyzed to obtain the target answering information corresponding to the user's question description information. Specifically, the question answering information corresponding to the target data vector can be input into a pre-trained large language model to obtain the target answering information corresponding to the user's question description information; the question answering information corresponding to the target data vector can also be subjected to speech analysis to obtain a coherent target answering information.

[0038] A data processing method provided by an embodiment of the present disclosure can recall more accurate question and answer records from a preset question and answer library based on user question description information and user input information, and generate question answer information based on the question and answer records. This method can accurately understand the user's intention and give a correct reply, thereby improving the efficiency of customer service question reply.

[0039] The following embodiments are used to describe a method for determining a target data vector and a method for determining a preset question and answer library.

[0040] Specifically, the specific process of selecting the target data vector that matches the output vector from the preset question and answer library may include: determining the similarity between the output vector and the data vector corresponding to each question-answer pair in the preset question and answer library; based on the similarity, selecting the target data vector that matches the output vector from the preset question and answer library.

[0041] In the specific implementation, when determining the target data vector, it is first necessary to calculate the similarity between the output vector and the data vector corresponding to each question-answer pair in the preset question-answer library, obtain the similarity corresponding to each data vector, and then select one or more data vectors with higher similarity from the preset question-answer library as the target data vector.

[0042] In an optional embodiment, the specific process of selecting the target data vector matching the output vector from the preset question and answer library can be any of the following methods:

[0043] Method 1: A data vector in a preset question and answer database whose similarity is greater than a first preset similarity threshold is determined as a target data vector that matches the output vector.

[0044] In specific implementation, the first preset similarity threshold can be determined according to research and development requirements, for example, the first preset similarity threshold can be 85% or 90%, etc. Specifically, all data vectors in the preset question and answer library whose similarity is greater than the first preset similarity threshold can be determined as target data vectors.

[0045] Method 2: Sort the similarities between the output vector and the data vectors corresponding to the question-answer pairs in the preset question-answer library in order from large to small to obtain the sorting result, and determine a specified number of data vectors ranked at the top of the sorting result as the target data vectors that match the output vector.

[0046] In specific implementation, firstly, the similarity corresponding to each data vector in the preset question and answer database can be sorted in descending order of similarity to obtain the sorting result, and then a specified number of data vectors before sorting are selected from the sorting result as the target data vectors, and the specified number can be determined according to the research and development requirements. For example, the specified number can be 5 or 6.

[0047] In another optional embodiment, the above-mentioned specific process of selecting a target data vector that matches the output vector from a preset question and answer library based on similarity may also include: selecting a first data vector with the highest similarity to the output vector from each cluster group in the preset question and answer library; determining the first data vector in each cluster group as the target data vector that matches the output vector; wherein the number of vectors contained in the target data vector matches the number of cluster groups contained in the preset question and answer library.

[0048] In a specific implementation, the preset question-answer library contains multiple cluster groups, each of which contains one or more question-answer pairs, and a data vector corresponding to each question-answer pair. At the same time, the similarity between the data vectors contained in each cluster group is high. Therefore, when determining the target data vector, a data vector with the highest similarity to the output vector can be selected from each cluster group as the target data vector, so that the number of vectors contained in the target data vector is the same as the number of cluster groups contained in the preset question-answer library.

[0049] In actual applications, the above-mentioned preset question and answer library is determined in the following manner: obtain multiple historical reply records; wherein the historical reply records include: question-answer pairs consisting of question description information and question answer information; for each historical reply record in the multiple historical reply records, input the historical reply record into the data processing model to obtain a data vector corresponding to the historical reply record; determine the similarity between the data vectors corresponding to every two historical reply records in the multiple historical reply records; divide the historical reply records whose similarity is less than a second preset threshold into the same cluster group to obtain multiple cluster groups; wherein the cluster group includes at least one multiple historical reply record and the data vectors corresponding to the multiple historical reply records; determine the multiple cluster groups as the preset question and answer library.

[0050] In specific implementation, when initially building a preset question-and-answer library, it is first necessary to obtain the historical response records of user questions (the historical response records are also question-answer pairs), and then use the data processing model to convert all these historical response records into data vectors, that is, each historical response record corresponds to a data vector; after obtaining all the data vectors, these data vectors and historical response records are clustered, that is, similar historical response records are clustered into a group. If the similarity between a historical response record and the data that has been clustered is lower than the second preset similarity threshold, a new cluster group is added, and finally the obtained historical response records are converted into N cluster groups, each of which includes at least one historical response record and the data vector corresponding to the historical response record. The initial historical question-and-answer library is these N cluster groups.

[0051] In order to ensure the accuracy of recommended replies, the comprehensiveness and real-time performance of the preset question and answer library are very important. However, relying on people from various products to maintain the preset question and answer library every day will waste a lot of manpower. Therefore, the present disclosure provides a method for self-updating the preset question and answer library. Specifically, the method for self-updating the preset question and answer library includes: forming a target question-answer pair with user question description information and target answer information, and determining the output vector as the target data vector corresponding to the target question-answer pair; based on the similarity between the target data vector and the data vector corresponding to each question-answer pair in the preset question and answer library, determining the target clustering group to which the target data vector belongs in the preset question and answer library; associating the target question-answer pair with the target data vector and storing it in the target clustering group to update the preset question and answer library.

[0052] In the specific implementation, every time the customer service responds to a new user question, that is, each time the target reply information corresponding to the user question description information sent by the user is obtained, the output vector corresponding to the user question description information will be obtained, or the user question description information and the target reply information will be converted into an output vector again, and then the similarity between the output vector and each data vector in the preset question and answer library will be calculated. If the similarity between the output vector and a certain data vector is greater than the second preset similarity threshold, the target question-answer pair corresponding to the output vector is divided into the cluster group corresponding to the data vector; if the similarity between the output vector and each data vector in the preset question and answer library is not greater than the second preset similarity threshold, the output vector and the corresponding target question reply are divided into a new cluster group, thereby ensuring that uncovered hot issues can also be effectively handled.

[0053] The following embodiments are used to describe the method of generating target reply information.

[0054] Specifically, the specific process of generating target answer information corresponding to the user's question description information based on the question answer information corresponding to the target data vector may include: determining the model prompt words based on the user's question description information, the question answer information corresponding to the target data vector and the information output rules; inputting the model prompt words into a pre-trained large language model to obtain the target answer information corresponding to the user's question description information.

[0055] In a specific implementation, the above-mentioned large language model refers to a natural language processing model obtained by training with a large amount of data, which can understand natural language and generate human-readable text. The large speech model used in the embodiment of the present disclosure is pre-trained, and the large language model can automatically generate text with a certain logical structure and semantics by learning and analyzing a large amount of data. Specifically, when using a large language model, how to design the input prompt words is very important, because the output of the large language model (i.e., the generated text) will be generated according to the input prompt words, and good prompt words can guide the large language model to generate useful and accurate text, while poor prompt words may cause the output of the large language model to deviate from expectations, or produce ambiguous and semantically unclear results.

[0056] In specific implementation, the above model prompts include user question description information, question answer information corresponding to the target data vector, and information output rules. For example, the model prompt can be: "You are a customer service representative of product xxx, and you are required to give a reply. The user question is the user description information. Here are some historical replies for reference: question answer information corresponding to the target data vector. Please output the target answer information according to the information output rule yyy". Specifically, the above information output rules can be determined according to R&D needs and are not specifically limited here.

[0057] The following embodiments are used to describe the manner of translating data.

[0058] Specifically, after obtaining the user's question description information and the user's input information, the first language corresponding to the user's question description information can be identified; then the user's question description information and the user's input information are input into a preset trained data processing model to obtain an output vector; a target data vector matching the output vector is selected from a preset question-and-answer library, the question answer information corresponding to the target data vector in the preset question-and-answer library is determined, and the target answer information corresponding to the user's question description information is generated based on the question answer information corresponding to the target data vector. After generating the target answer information corresponding to the user's question description information based on the question answer information corresponding to the target data vector, the second language corresponding to the target answer information corresponding to the user's question description information is identified; it is determined whether the first language is the same as the second language; if they are the same, the target answer information is sent to the user device that sent the user's question description information; if they are not the same, the target answer information is translated into the first language, and the target answer information in the first language is returned to the user device that sent the user's question description information.

[0059] In actual application scenarios, multi-language problems are often encountered. For example, users may ask questions in English, Korean or some minority languages ​​to obtain user problem description information. Therefore, for recommended replies, a translation step is also required to give replies in the corresponding language based on the language of the user's question. Specifically, it is first necessary to identify the language of the user's problem description information submitted by the user. Here, the machine learning-based language identification method CompactLanguage Detector can be used for identification to obtain the first language to be translated. The first language is also the language corresponding to the user's problem description information. If the target reply information obtained is also in the first language, no translation is required. Otherwise, the second language corresponding to the target reply information needs to be translated into the first language.

[0060] In a specific implementation, the specific process of translating the second language corresponding to the target answer information into the first language may include: firstly, it is necessary to extract the target terminology words matching the terminology words in the terminology word list from the target answer information, then construct a set of prompt words by combining the extracted target answer information in the second language, the target terminology words and the first language information, and input the prompt words into the translation model to obtain the target answer information in the first language. The terminology word list includes multiple terminology words and the corresponding translation texts of each terminology word in different languages.

[0061] In a specific embodiment, the prompt words may be: "You are a language expert, and you are required to translate from the second language to the first language. The sentence to be translated is the target reply information in the first language. Here are some translation terms: target data words in the first language. Please output the translation result." The translation model may also be a large language model.

[0062] The following embodiments are used to describe the method of classifying user questions and filtering tags.

[0063] Specifically, after obtaining the user problem description information and the user input information, determine whether the user problem description information is valid information; if it is valid information, execute the step of inputting the user problem description information and the user input information into a preset trained data processing model; if it is not valid information, return the preset reply information to the user device that sent the user problem description information.

[0064] In the specific implementation, for the user problem description information submitted by the user, it is first necessary to determine whether it is a validity issue. In actual scenarios, some invalid questions submitted by users are often received. If these questions need to be reviewed one by one by the customer service students, it will waste too much time. For this reason, the present invention designs a validity analysis module, which can analyze the submitted questions using NLP text classification technology. For example, if the problem text description information submitted by the user does not have enough semantic information or is some messy descriptions (for example, 111111, asdfgg, etc.), it can be judged that this problem description information is invalid. For the problem description information judged to be invalid, a preset reply information can be given immediately to prompt the user to supplement the information to complete the problem, avoid users waiting too long, and also reduce the intervention of customer service students.

[0065] In a specific embodiment, the method of determining whether the user problem description information is valid information may also include inputting the user problem description information into a classification model, and the classification model may output whether the user problem description information is valid. The classification model is trained based on training data, and the training data includes a large number of training samples, each training sample includes problem description information and a validity label for the problem description information annotated by the annotator, and the validity label is used to indicate whether the problem description information is valid or invalid. The above classification model may be a Bert model or other neural network model, which may be determined according to R&D needs.

[0066] In an optional embodiment, after obtaining user problem description information and user input information, the user problem description information is input into a pre-trained label model to obtain a target label corresponding to the user problem description information; wherein the target label is a label among multiple preset labels.

[0067] In specific implementation, the above-mentioned label model is trained based on preset training data, which includes multiple samples, each of which includes problem description information and the category label represented by the labeler for the problem description information. The category label is any one of multiple preset labels. For example, preset labels may include but are not limited to general problems, abnormal errors, and recharge problems. Specifically, for user problem description information that is judged to be a valid problem, it is necessary to use the label model to further analyze it to determine the category to which the user problem description information belongs (equivalent to the above-mentioned target label). This step is mainly to determine the importance of the user problem description information. When customer service colleagues receive too much user problem description information at the same time, they can filter by category to prioritize responses to more important issues, thereby bringing a better user experience.

[0068] like Figure 2The figure shows an architecture diagram for generating a recommended reply provided by an embodiment of the present disclosure. First, the user inputs user question description information, and then performs validity analysis. When it is determined that the user question description information is an invalid question, preset reply information is output. When it is determined that the user question description information is a valid question, the target tag corresponding to the user question description information is identified to determine the processing priority of the user question description information; then the valid user question description information is input into a personalized retrieval model (equivalent to the above-mentioned data processing model) to obtain an output vector corresponding to the user question description information, and the similarity between the output vector and the data vector corresponding to each question-answer pair in the preset question-answer library is determined, and K target data vectors with a high similarity to the output vector are selected from the preset question-answer library; then the question answer information corresponding to the target data vector in the preset question-answer library is determined, and the model prompt word is determined based on the user question description information, the question answer information corresponding to the target data vector, and the information output rule; the model prompt word is input into a large language model (equivalent to Figure 2 In the LLM model in the example above, the target answer information corresponding to the user's question description information is obtained (equivalent to the recommended answer). If the target answer information is in a different language than the user's question description information, the target answer information needs to be translated into the language corresponding to the user's question description information to obtain the final output result.

[0069] By utilizing the above technical solution, the response efficiency of customer service can be effectively improved, thereby reducing operating costs while providing users with timely responses.

[0070] Corresponding to the above method embodiment, the present disclosure embodiment also provides a data processing device, such as Figure 3 As shown, the device comprises:

[0071] The information acquisition module 30 is used to acquire user problem description information and user input information; wherein the user input information is used to indicate the information sending status and / or log information corresponding to the user problem description information.

[0072] The vector determination module 31 is used to input the user problem description information and the user input information into a preset trained data processing model to obtain an output vector.

[0073] The answer determination module 32 is used to select a target data vector that matches the output vector from the preset question and answer library, and determine the question answer information corresponding to the target data vector in the preset question and answer library; wherein the preset question and answer library includes multiple question-answer pairs, and data vectors corresponding to the question-answer pairs; the question-answer pairs include question description information and question answer information corresponding to the question description information.

[0074] The answer generation module 33 is used to generate target answer information corresponding to the user question description information based on the question answer information corresponding to the target data vector.

[0075] The above-mentioned data processing device can recall more accurate question and answer records from the preset question and answer library based on the user question description information and user input information, and generate question answer information based on the question and answer records. In this way, this method can accurately understand the user's intention and give the correct response, and improve the efficiency of customer service question response.

[0076] Specifically, the above-mentioned answer determination module 32 is used to: determine the similarity between the output vector and the data vector corresponding to each question-answer pair in the preset question-answer library; based on the similarity, select a target data vector that matches the output vector from the preset question-answer library.

[0077] Furthermore, the above-mentioned answer determination module 32 is used to: determine the data vectors in the preset question and answer library whose similarity is greater than a first preset similarity threshold as target data vectors that match the output vector; or, sort the similarities between the output vector and the data vectors corresponding to the question and answer pairs in the preset question and answer library in order from large to small to obtain the sorting result, and determine a specified number of data vectors ranked at the top of the sorting result as target data vectors that match the output vector.

[0078] Furthermore, the above-mentioned device also includes a question and answer library determination module, which is used to: obtain multiple historical reply records; wherein the historical reply records include: question-answer pairs consisting of question description information and question answer information; for each historical reply record in the multiple historical reply records, input the historical reply record into a data processing model to obtain a data vector corresponding to the historical reply record; determine the similarity between the data vectors corresponding to every two historical reply records in the multiple historical reply records; divide the historical reply records whose similarity is less than a second preset threshold into the same cluster group to obtain multiple cluster groups; wherein the cluster group includes at least one multiple historical reply record and the data vectors corresponding to the multiple historical reply records; determine the multiple cluster groups as a preset question and answer library.

[0079] Furthermore, the above-mentioned device also includes a question and answer library updating module, which is used to: combine user question description information and target answer information into a target question and answer pair, and determine the output vector as the target data vector corresponding to the target question and answer pair; based on the similarity between the target data vector and the data vector corresponding to each question and answer pair in the preset question and answer library, determine the target clustering group to which the target data vector belongs in the preset question and answer library; associate the target question and answer pair and the target data vector and store them in the target clustering group to update the preset question and answer library.

[0080] Furthermore, the above-mentioned answer determination module 32 is used to: select the first data vector with the highest similarity to the output vector from each cluster group in the preset question and answer library; determine the first data vector in each cluster group as the target data vector matching the output vector; wherein the number of vectors contained in the target data vector matches the number of cluster groups contained in the preset question and answer library.

[0081] Furthermore, the above-mentioned answer generation module 33 is used to: determine the model prompt words based on the user's question description information, the question answer information corresponding to the target data vector and the information output rules; input the model prompt words into the pre-trained large language model to obtain the target answer information corresponding to the user's question description information.

[0082] Furthermore, the above-mentioned device also includes a validity judgment module, which is used to: after obtaining the user problem description information and the user input information, determine whether the user problem description information is valid information; if it is valid information, execute the step of inputting the user problem description information and the user input information into a preset trained data processing model; if it is not valid information, return preset reply information to the user device that sent the user problem description information.

[0083] Furthermore, the above-mentioned device also includes a label setting module, which is used to: after obtaining user problem description information and user input information, input the user problem description information into a pre-trained label model to obtain a target label corresponding to the user problem description information; wherein the target label is a label among multiple preset labels.

[0084] Furthermore, the above device also includes a language identification module, which is used to: after obtaining the user question description information and the user input information, identify the first language corresponding to the user question description information. Based on this, the above device also includes a translation module, which is used to: after generating the target answer information corresponding to the user question description information based on the question answer information corresponding to the target data vector, identify the second language corresponding to the target answer information corresponding to the user question description information; determine whether the first language is the same as the second language; if they are the same, send the target answer information to the user device that sent the user question description information; if they are not the same, translate the target answer information into the first language, and return the target answer information in the first language to the user device that sent the user question description information.

[0085] The data processing device provided in the embodiment of the present disclosure has the same implementation principle and technical effects as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0086] The present disclosure also provides an electronic device, such as Figure 4As shown, the electronic device includes a processor and a memory, the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above data processing method.

[0087] Specifically, the above-mentioned data processing method includes: obtaining user question description information and user input information; wherein the user input information is used to indicate the information sending status and / or log information corresponding to the user question description information; inputting the user question description information and the user input information into a preset trained data processing model to obtain an output vector; selecting a target data vector matching the output vector from a preset question and answer library, and determining the question answer information corresponding to the target data vector in the preset question and answer library; wherein the preset question and answer library includes multiple question and answer pairs, and data vectors corresponding to the question and answer pairs; the question and answer pairs include question description information and question answer information corresponding to the question description information; based on the question answer information corresponding to the target data vector, generating target answer information corresponding to the user question description information.

[0088] In the above data processing method, based on the user question description information and user input information, more accurate question and answer records can be recalled from the preset question and answer library, and question answer information can be generated based on the question and answer records. In this way, this method can accurately understand the user's intention and give the correct response, and improve the efficiency of customer service question response.

[0089] In an optional embodiment, the above-mentioned step of selecting a target data vector that matches the output vector from a preset question and answer library includes: determining the similarity between the output vector and the data vector corresponding to each question-answer pair in the preset question and answer library; based on the similarity, selecting a target data vector that matches the output vector from the preset question and answer library.

[0090] In an optional embodiment, the above-mentioned selecting a target data vector that matches the output vector from a preset question and answer library based on similarity includes the following methods: determining a data vector in the preset question and answer library whose similarity is greater than a first preset similarity threshold as a target data vector that matches the output vector; and sorting the similarities between the output vector and the data vectors corresponding to the question and answer pairs in the preset question and answer library in order from large to small to obtain a sorting result, and determining a specified number of data vectors ranked at the top of the sorting result as the target data vectors that match the output vector.

[0091] In an optional embodiment, the above-mentioned preset question and answer library is determined in the following manner: obtaining multiple historical reply records; wherein the historical reply records include: question-answer pairs consisting of question description information and question answer information; for each historical reply record in the multiple historical reply records, inputting the historical reply record into a data processing model to obtain a data vector corresponding to the historical reply record; determining the similarity between the data vectors corresponding to every two historical reply records in the multiple historical reply records; dividing the historical reply records whose similarity is less than a second preset threshold into the same cluster group to obtain multiple cluster groups; wherein the cluster group includes at least one multiple historical reply record and the data vectors corresponding to the multiple historical reply records; and determining the multiple cluster groups as the preset question and answer library.

[0092] In an optional embodiment, the above method also includes: combining user question description information and target answer information to form a target question-answer pair, and determining the output vector as the target data vector corresponding to the target question-answer pair; determining the target clustering group to which the target data vector belongs in the preset question-answer library based on the similarity between the target data vector and the data vector corresponding to each question-answer pair in the preset question-answer library; and associating the target question-answer pair and the target data vector and storing them in the target clustering group to update the preset question-answer library.

[0093] In an optional embodiment, the above-mentioned step of selecting a target data vector that matches the output vector from a preset question and answer library based on similarity includes: selecting the first data vector with the highest similarity to the output vector from each cluster group in the preset question and answer library; determining the first data vector in each cluster group as the target data vector that matches the output vector; wherein the number of vectors contained in the target data vector matches the number of cluster groups contained in the preset question and answer library.

[0094] In an optional embodiment, the above-mentioned step of generating target answer information corresponding to the user question description information based on the question answer information corresponding to the target data vector includes: determining the model prompt words based on the user question description information, the question answer information corresponding to the target data vector and the information output rules; inputting the model prompt words into a pre-trained large language model to obtain the target answer information corresponding to the user question description information.

[0095] In an optional embodiment, after the step of obtaining user problem description information and user input information, the above method also includes: determining whether the user problem description information is valid information; if it is valid information, executing the step of inputting the user problem description information and user input information into a preset trained data processing model; if it is not valid information, returning preset reply information to the user device that sent the user problem description information.

[0096] In an optional embodiment, after the step of obtaining user problem description information and user input information, the method further includes: inputting the user problem description information into a pre-trained label model to obtain a target label corresponding to the user problem description information; wherein the target label is a label among multiple preset labels.

[0097] In an optional embodiment, after the step of obtaining the user's question description information and the user's input information, the method further includes: identifying the first language corresponding to the user's question description information. After the step of generating the target answer information corresponding to the user's question description information based on the question answer information corresponding to the target data vector, the method further includes: identifying the second language corresponding to the target answer information corresponding to the user's question description information; determining whether the first language is the same as the second language; if they are the same, sending the target answer information to the user device that sent the user's question description information; if they are not the same, translating the target answer information into the first language, and returning the target answer information in the first language to the user device that sent the user's question description information.

[0098] Further, Figure 4 The electronic device shown further includes a bus 102 and a communication interface 103 , and the processor 101 , the communication interface 103 and the memory 100 are connected via the bus 102 .

[0099] The memory 100 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 103 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 102 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0100] The processor 101 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 101. The above processor 101 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present disclosure can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present disclosure can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 100, and the processor 101 reads the information in the memory 100 and completes the steps of the method of the above embodiment in combination with its hardware.

[0101] The embodiment of the present disclosure also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned data processing method.

[0102] Specifically, the above-mentioned data processing method includes: obtaining user question description information and user input information; wherein the user input information is used to indicate the information sending status and / or log information corresponding to the user question description information; inputting the user question description information and the user input information into a preset trained data processing model to obtain an output vector; selecting a target data vector matching the output vector from a preset question and answer library, and determining the question answer information corresponding to the target data vector in the preset question and answer library; wherein the preset question and answer library includes multiple question and answer pairs, and data vectors corresponding to the question and answer pairs; the question and answer pairs include question description information and question answer information corresponding to the question description information; based on the question answer information corresponding to the target data vector, generating target answer information corresponding to the user question description information.

[0103] In the above data processing method, based on the user question description information and user input information, more accurate question and answer records can be recalled from the preset question and answer library, and question answer information can be generated based on the question and answer records. In this way, this method can accurately understand the user's intention and give the correct response, and improve the efficiency of customer service question response.

[0104] In an optional embodiment, the above-mentioned step of selecting a target data vector that matches the output vector from a preset question and answer library includes: determining the similarity between the output vector and the data vector corresponding to each question-answer pair in the preset question and answer library; based on the similarity, selecting a target data vector that matches the output vector from the preset question and answer library.

[0105] In an optional embodiment, the above-mentioned selecting a target data vector that matches the output vector from a preset question and answer library based on similarity includes the following methods: determining a data vector in the preset question and answer library whose similarity is greater than a first preset similarity threshold as a target data vector that matches the output vector; and sorting the similarities between the output vector and the data vectors corresponding to the question and answer pairs in the preset question and answer library in order from large to small to obtain a sorting result, and determining a specified number of data vectors ranked at the top of the sorting result as the target data vectors that match the output vector.

[0106] In an optional embodiment, the above-mentioned preset question and answer library is determined in the following manner: obtaining multiple historical reply records; wherein the historical reply records include: question-answer pairs consisting of question description information and question answer information; for each historical reply record in the multiple historical reply records, inputting the historical reply record into a data processing model to obtain a data vector corresponding to the historical reply record; determining the similarity between the data vectors corresponding to every two historical reply records in the multiple historical reply records; dividing the historical reply records whose similarity is less than a second preset threshold into the same cluster group to obtain multiple cluster groups; wherein the cluster group includes at least one multiple historical reply record and the data vectors corresponding to the multiple historical reply records; and determining the multiple cluster groups as the preset question and answer library.

[0107] In an optional embodiment, the above method also includes: combining user question description information and target answer information to form a target question-answer pair, and determining the output vector as the target data vector corresponding to the target question-answer pair; determining the target clustering group to which the target data vector belongs in the preset question-answer library based on the similarity between the target data vector and the data vector corresponding to each question-answer pair in the preset question-answer library; and associating the target question-answer pair and the target data vector and storing them in the target clustering group to update the preset question-answer library.

[0108] In an optional embodiment, the above-mentioned step of selecting a target data vector that matches the output vector from a preset question and answer library based on similarity includes: selecting the first data vector with the highest similarity to the output vector from each cluster group in the preset question and answer library; determining the first data vector in each cluster group as the target data vector that matches the output vector; wherein the number of vectors contained in the target data vector matches the number of cluster groups contained in the preset question and answer library.

[0109] In an optional embodiment, the above-mentioned step of generating target answer information corresponding to the user question description information based on the question answer information corresponding to the target data vector includes: determining the model prompt words based on the user question description information, the question answer information corresponding to the target data vector and the information output rules; inputting the model prompt words into a pre-trained large language model to obtain the target answer information corresponding to the user question description information.

[0110] In an optional embodiment, after the step of obtaining user problem description information and user input information, the above method also includes: determining whether the user problem description information is valid information; if it is valid information, executing the step of inputting the user problem description information and user input information into a preset trained data processing model; if it is not valid information, returning preset reply information to the user device that sent the user problem description information.

[0111] In an optional embodiment, after the step of obtaining user problem description information and user input information, the method further includes: inputting the user problem description information into a pre-trained label model to obtain a target label corresponding to the user problem description information; wherein the target label is a label among multiple preset labels.

[0112] In an optional embodiment, after the step of obtaining the user's question description information and the user's input information, the method further includes: identifying the first language corresponding to the user's question description information. After the step of generating the target answer information corresponding to the user's question description information based on the question answer information corresponding to the target data vector, the method further includes: identifying the second language corresponding to the target answer information corresponding to the user's question description information; determining whether the first language is the same as the second language; if they are the same, sending the target answer information to the user device that sent the user's question description information; if they are not the same, translating the target answer information into the first language, and returning the target answer information in the first language to the user device that sent the user's question description information.

[0113] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a terminal device, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0114] In the description of the present disclosure, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present disclosure and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present disclosure. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0115] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed in the present disclosure, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A data processing method, characterized in that: The method comprises: Obtaining user problem description information and user input information; wherein the user input information is used to indicate the information sending status and / or log information corresponding to the user problem description information; Inputting the user problem description information and the user input information into a preset trained data processing model to obtain an output vector; Selecting a target data vector that matches the output vector from a preset question-and-answer library, and determining question answer information corresponding to the target data vector in the preset question-and-answer library; wherein the preset question-and-answer library includes a plurality of question-answer pairs, and data vectors corresponding to the question-answer pairs; the question-answer pairs include question description information and question answer information corresponding to the question description information; Based on the question answer information corresponding to the target data vector, target answer information corresponding to the user question description information is generated.

2. The method according to claim 1, characterized in that The step of selecting a target data vector matching the output vector from a preset question-and-answer library comprises: Determine the similarity between the output vector and the data vector corresponding to each question-answer pair in the preset question-answer library; Based on the similarity, a target data vector matching the output vector is selected from the preset question and answer library.

3. The method according to claim 2, characterized in that The selecting, based on the similarity, a target data vector matching the output vector from the preset question and answer library comprises the following manner: Determine the data vector in the preset question-and-answer database whose similarity is greater than a first preset similarity threshold as a target data vector that matches the output vector; and, The similarities between the output vector and the data vectors corresponding to the question-answer pairs in the preset question-answer library are sorted in order from large to small to obtain a sorting result, and a specified number of data vectors ranked at the top of the sorting result are determined as target data vectors matching the output vector.

4. The method according to claim 2, characterized in that: The preset question and answer library is determined in the following manner: Acquire multiple historical reply records; wherein the historical reply records include: question-reply pairs consisting of question description information and question-reply information; For each of the plurality of historical reply records, input the historical reply record into the data processing model to obtain a data vector corresponding to the historical reply record; Determine the similarity between the data vectors corresponding to every two of the historical reply records in the plurality of historical reply records; The historical reply records whose similarity is less than a second preset threshold are divided into the same cluster group to obtain a plurality of cluster groups; wherein the cluster group includes at least one of the plurality of historical reply records and data vectors corresponding to the plurality of historical reply records; The multiple cluster groups are determined as the preset question and answer library.

5. The method according to claim 4, characterized in that The method further comprises: The user question description information and the target answer information form a target question-answer pair, and the output vector is determined as a target data vector corresponding to the target question-answer pair; Determine a target cluster group to which the target data vector belongs in the preset question and answer library based on the similarity between the target data vector and the data vector corresponding to each question-answer pair in the preset question and answer library; The target question-answer pair and the target data vector are associated and stored in the target cluster group to update the preset question-answer library.

6. The method according to claim 4, characterized in that The step of selecting a target data vector matching the output vector from the preset question and answer library based on the similarity comprises: Selecting, from each cluster group in the preset question-and-answer library, a first data vector having the highest similarity to the output vector; The first data vector in each cluster group is determined as a target data vector that matches the output vector; wherein the number of vectors included in the target data vector matches the number of cluster groups included in the preset question and answer library.

7. The method according to claim 1, characterized in that The step of generating target answer information corresponding to the user question description information based on the question answer information corresponding to the target data vector comprises: Determining a model prompt word based on the user question description information, the question answer information corresponding to the target data vector, and an information output rule; The model prompt words are input into a pre-trained large language model to obtain target answer information corresponding to the user question description information.

8. The method according to claim 1, characterized in that After the step of obtaining user problem description information and user input information, the method further includes: Determine whether the user problem description information is valid information; If the information is valid, the step of inputting the user problem description information and the user input information into a preset trained data processing model is executed; If it is not valid information, return preset reply information to the user device that sent the user question description information.

9. The method according to claim 1, characterized in that: After the step of obtaining user problem description information and user input information, the method further includes: The user question description information is input into a pre-trained label model to obtain a target label corresponding to the user question description information; wherein the target label is a label among multiple preset labels.

10. The method according to claim 1, characterized in that After the step of obtaining user problem description information and user input information, the method further includes: Identify the first language corresponding to the user problem description information; After the step of generating target answer information corresponding to the user question description information based on the question answer information corresponding to the target data vector, the method further includes: Identify a second language corresponding to the target answer information corresponding to the user question description information; Determine whether the first language is the same as the second language; if they are the same, send the target reply information to the user device that sent the user question description information; If they are not the same, the target answer information is translated into the first language, and the target answer information in the first language is returned to the user device that sent the user question description information.

11. A data processing device, characterized in that: The device comprises: An information acquisition module, used to acquire user problem description information and user input information; wherein the user input information is used to indicate the information sending status and / or log information corresponding to the user problem description information; A vector determination module, used for inputting the user question description information and the user input information into a preset trained data processing model to obtain an output vector; A reply determination module is used to select a target data vector that matches the output vector from a preset question-answer library, and determine the question reply information corresponding to the target data vector in the preset question-answer library; wherein the preset question-answer library includes a plurality of question-answer pairs, and data vectors corresponding to the question-answer pairs; the question-answer pairs include question description information and question reply information corresponding to the question description information; The answer generation module is used to generate target answer information corresponding to the user question description information based on the question answer information corresponding to the target data vector.

12. An electronic device, characterized in that: The electronic device includes a processor and a memory, the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the data processing method according to any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the data processing method according to any one of claims 1 to 10.

Images