Answer generation method and device based on wearable device, computer device and medium

Abstract

The invention belongs to the technical field of artificial intelligence, relates to an answer generation method and device based on wearable equipment, computer equipment and a storage medium, is applied to the field of financial science and technology, and comprises the following steps: in a dialogue process between an agent and a client, collecting voice data of the client based on the wearable equipment of the agent; performing voice recognition on the voice data to obtain a text; performing information extraction on the text to obtain key information; performing intention recognition on the text to obtain an intention recognition result; searching a search result matched with the key information and the intention recognition result from a preset knowledge base; determining a target answer matched with the customer from the search result; and converting the target answer into a corresponding specified voice, and outputting the specified voice to the wearable device. In addition, the invention also relates to a block chain technology, and the target answer can be stored in a block chain. Based on the use of the wearable device, the conversation processing efficiency and the reply quality of the customer are effectively improved.

Description

Technical Field

[0001] The present application relates to the fields of artificial intelligence development technology and financial technology, and in particular to a wearable device-based answer generation method, apparatus, computer equipment and storage medium. Background Art

[0002] In the insurance field, agent business development and recruitment are key links for insurance companies to expand the market and increase business volume. The traditional offline dialogue communication method, that is, agents communicate with customers face to face, can directly understand customers' needs and concerns, but it faces many challenges in actual operation.

[0003] First, agents need to manually organize answers to respond to customers' questions. In this process, agents need to rely on their own professional knowledge, experience and judgment to respond to customers' questions immediately and accurately. However, due to differences in each agent's knowledge background, business capabilities and communication skills, there is often inefficiency in answering customer questions. Some agents may need to spend a long time thinking and organizing answers, which prolongs the communication time with customers and reduces the efficiency of business processing.

[0004] Second, the manual organization of answers may also result in low-quality responses. In the insurance field, customers’ questions often involve multiple aspects, including the characteristics of insurance products, coverage, claims process, etc. If the agent does not have a comprehensive or accurate grasp of these knowledge, there may be deviations or omissions when answering customer questions, which will affect customer trust and satisfaction. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose a wearable device-based answer generation method, device, computer equipment and storage medium to solve the technical problems of low business processing efficiency and low response quality in existing offline dialogue communication methods.

[0006] In order to solve the above technical problems, the embodiment of the present application provides a method for generating answers based on a wearable device, which adopts the following technical solution:

[0007] During the conversation between the agent and the customer, voice data of the customer is collected based on the agent's wearable device;

[0008] Performing speech recognition on the speech data to obtain corresponding text;

[0009] Extracting information from the text to obtain corresponding key information;

[0010] Performing intent recognition on the text to obtain a corresponding intent recognition result;

[0011] Querying a preset knowledge base for search results that match the key information and the intent recognition result;

[0012] Determining a target answer matching the customer from the search results;

[0013] The target answer is converted into a corresponding specified voice, and the specified voice is output to the wearable device.

[0014] Furthermore, the step of performing intent recognition on the text to obtain a corresponding intent recognition result specifically includes:

[0015] Performing syntactic analysis on the text to obtain corresponding syntactic analysis features;

[0016] Performing semantic role labeling on the text to obtain corresponding semantic role labeling features;

[0017] Performing a text structure analysis on the text to obtain corresponding text structure analysis features;

[0018] Performing emotion recognition on the text to obtain corresponding emotion recognition features;

[0019] Based on a preset intention recognition model, the syntactic analysis feature, the semantic role labeling feature, the text structure analysis feature, and the emotion recognition feature are processed for intention recognition to obtain a corresponding output result;

[0020] The output result is used as the intention recognition result.

[0021] Furthermore, the step of determining a target answer matching the customer from the search results specifically includes:

[0022] Obtaining a customer profile of the customer, and obtaining a conversation context corresponding to the text;

[0023] Based on the customer profile and the conversation context, determining an initial answer from all the search results;

[0024] Scoring all the initial answers based on a preset scoring strategy to obtain a score for each initial answer;

[0025] Selecting a designated answer with the highest score from all the initial answers;

[0026] A target answer matching the customer is generated based on the specified answer.

[0027] Furthermore, the step of determining an initial answer from all the search results based on the customer profile and the conversation context specifically includes:

[0028] Converting the customer portrait, the conversation context, and the search results into corresponding customer portrait vectors, conversation context vectors, and search result vectors based on a preset natural language processing strategy;

[0029] Calculating a first similarity between the customer portrait vector and the search result vector, and calculating a second similarity between the conversation context vector and the search result vector;

[0030] Filter out, from all the search result vectors, a specified search result vector whose first similarity is greater than a preset first similarity threshold and whose second similarity is greater than a preset second similarity threshold;

[0031] Obtaining a specified search result corresponding to the specified search result vector;

[0032] The specified search result is used as an initial answer.

[0033] Furthermore, the step of scoring all the initial answers based on a preset scoring strategy to obtain a score for each initial answer specifically includes:

[0034] Extracting answer features of each of the initial answers based on a preset extraction strategy; wherein the answer features at least include answer relevance features, answer accuracy features, answer authority features, and answer timeliness features;

[0035] Obtaining a preset weight corresponding to the answer feature;

[0036] Get the preset scoring algorithm;

[0037] The answer features and the preset weights are calculated based on the scoring algorithm to obtain scores for each of the initial answers.

[0038] Furthermore, the step of generating a target answer matching the customer based on the specified answer specifically includes:

[0039] Performing sentence adjustment processing on the designated answer to obtain a corresponding first answer;

[0040] Adding a designated term to the first answer to obtain a corresponding second answer;

[0041] Customize the second answer based on the customer portrait and the conversation context to obtain a corresponding third answer;

[0042] The third answer is used as the target answer.

[0043] Furthermore, the step of extracting information from the text to obtain corresponding key information specifically includes:

[0044] Preprocessing the text to obtain a corresponding processed text;

[0045] Get the preset information extraction strategy;

[0046] Extracting information from the processed text based on the information extraction strategy to obtain corresponding specified information;

[0047] The designated information is used as the key information.

[0048] In order to solve the above technical problems, the embodiment of the present application further provides an answer generation device based on a wearable device, which adopts the following technical solution:

[0049] A collection module, used to collect voice data of the customer based on the agent's wearable device during the conversation between the agent and the customer;

[0050] A first recognition module, used to perform speech recognition on the speech data to obtain corresponding text;

[0051] An extraction module, used to extract information from the text to obtain corresponding key information;

[0052] A second recognition module is used to perform intent recognition on the text to obtain a corresponding intent recognition result;

[0053] A query module, used to query a preset knowledge base for search results that match the key information and the intent recognition result;

[0054] A determination module, used to determine a target answer matching the customer from the search results;

[0055] An output module is used to convert the target answer into a corresponding specified voice and output the specified voice to the wearable device.

[0056] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0057] During the conversation between the agent and the customer, voice data of the customer is collected based on the agent's wearable device;

[0058] Performing speech recognition on the speech data to obtain corresponding text;

[0059] Extracting information from the text to obtain corresponding key information;

[0060] Performing intent recognition on the text to obtain a corresponding intent recognition result;

[0061] Querying a preset knowledge base for search results that match the key information and the intent recognition result;

[0062] Determining a target answer matching the customer from the search results;

[0063] The target answer is converted into a corresponding specified voice, and the specified voice is output to the wearable device.

[0064] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0065] During the conversation between the agent and the customer, voice data of the customer is collected based on the agent's wearable device;

[0066] Performing speech recognition on the speech data to obtain corresponding text;

[0067] Extracting information from the text to obtain corresponding key information;

[0068] Performing intent recognition on the text to obtain a corresponding intent recognition result;

[0069] Querying a preset knowledge base for search results that match the key information and the intent recognition result;

[0070] Determining a target answer matching the customer from the search results;

[0071] The target answer is converted into a corresponding specified voice, and the specified voice is output to the wearable device.

[0072] Compared with the prior art, the embodiments of the present application have the following beneficial effects:

[0073] During the conversation between an agent and a customer, the present application collects the customer's voice data based on the agent's wearable device; then performs voice recognition on the voice data to obtain corresponding text; then performs information extraction on the text to obtain corresponding key information; and performs intent recognition on the text to obtain corresponding intent recognition results; subsequently, search results matching the key information and the intent recognition results are queried from a preset knowledge base; further, a target answer matching the customer is determined from the search results; finally, the target answer is converted into a corresponding designated voice, and the designated voice is output to the wearable device. During the conversation between the agent and the customer, the present application collects the customer's voice data through the use of the agent's wearable device, performs voice recognition on the voice data to obtain the corresponding text, then performs information extraction and intent recognition on the text to obtain the corresponding key information and intent recognition results, and then queries the search results that match the key information and intent recognition results based on the use of the knowledge base, and determines the target answer that matches the customer from the search results, and then converts the target answer into a corresponding specified voice output to the wearable device, so that the agent can use the wearable device to quickly obtain the target answer that matches the customer, and then can assist the agent to quickly and accurately answer the customer's questions, which is conducive to improving the efficiency of conversation processing with the customer and the quality of response. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0075] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0076] Figure 2 A flowchart of an embodiment of a method for generating an answer based on a wearable device according to the present application;

[0077] Figure 3 is a structural schematic diagram of an embodiment of an answer generation device based on a wearable device according to the present application;

[0078] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0079] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0080] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0081] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0082] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0083] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0084] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to the laptop computer 1011, tablet computer 1012 or mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0085] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .

[0086] It should be noted that the answer generation method based on the wearable device provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the answer generation device based on the wearable device is generally set in the server / terminal device.

[0087] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0088] Continue to refer Figure 2 , shows a flowchart of an embodiment of the answer generation method based on a wearable device according to the present application. According to different requirements, the order of the steps in the flowchart can be changed, and some steps can be omitted. The answer generation method based on a wearable device provided in the embodiment of the present application can be applied to any scenario where answer generation is required, and the answer generation method based on a wearable device can be applied to products in these scenarios, for example, answer generation in the field of finance and insurance. The answer generation method based on a wearable device includes the following steps:

[0089] Step S201, during the conversation between the agent and the customer, voice data of the customer is collected based on the agent's wearable device.

[0090] In this embodiment, the answer generation method based on the wearable device is run on the electronic device (for example Figure 1The server / terminal device shown in the figure) can obtain the voice data of the customer through a wired connection or a wireless connection. It should be noted that the above-mentioned wireless connection method may include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or developed in the future. The execution subject of the present application may specifically be an answer generation system including a wearable device, referred to as the system. The present application can be applied to business scenarios in which insurance agents in the financial field provide services to customers in business development or recruitment business. Among them, before the agent has a conversation with the customer, the agent first ensures that the wearable device is fully charged and turns on the device. Then, the Bluetooth headset is searched and connected through the terminal APP (central control application) to complete the device pairing, that is, to complete the pairing process between the system and the wearable device. And confirm the device connection status in the APP to ensure smooth communication. In addition, when talking with the customer, the agent wears the above-mentioned wearable device. The microphone integrated in the wearable device captures the conversation voice, that is, the customer's voice data, and transmits it to the system in real time through the Bluetooth module.

[0091] Specifically, wearable devices are divided into the following parts: 1. Bluetooth headset. Shell structure. Including ear shell, earmuff, isolation frame, support frame, etc. Speaker. Used to output the question and answer prompts in the conversation process to the user in the form of sound. Microphone. Integrated with the headset, used to listen to and capture subtle voices for analysis of the conversation context. Bluetooth module. Used to pair and connect with the control end APP to achieve advanced control. Power supply. Provide energy for the operation of the device. Microprocessor. Process the basic logic in the hardware device. Touch sensing module. Used to implement basic functions such as device switch and volume adjustment. 2. Central control application. Using APP as the carrier, it supports running on the mobile phone and APP. Device connection. Supports searching for devices through Bluetooth, pairing and connecting with headsets, and disconnecting. Device information. Supports viewing the device name, connection status, detailed information, etc. of the connected device. Device control. Supports voice listening switch, intelligent answer switch, personnel selection, volume adjustment, etc. 3. Semantic recognition module. Speech recognition. Mainly includes feature extraction technology, pattern matching criteria and model training. Word parsing and information extraction. Including word segmentation, part-of-speech tagging, named entity recognition and word sense disambiguation, extracting important information from a given text. Syntactic parsing and discourse comprehension. Analysis of a certain hierarchical structure and semantic relationship between a series of consecutive clauses, sentences and paragraphs in the text structure, including time, events, causal relationships, etc., and even the emotion recognition carried by the text. Natural language generation. The process of automatically generating text from structured data in a readable way. It includes three stages: text planning (completing the planning of basic content in structured data), sentence planning (combining sentences from structured data to express information flow), and implementation (generating grammatically smooth sentences to express text). 4. Insurance knowledge base. A professional knowledge base built with insurance industry knowledge as the material, supporting semantic parsing and natural language generation. Sales skills. Mainly including knowledge related to appointments, sales interviews, deal interviews, and objection interviews. Customer development. Mainly including knowledge related to list acquisition, list organization, and customer management. Compliance business development. Mainly including knowledge related to laws and regulations, company policies and systems. Product knowledge. Mainly includes product knowledge such as life insurance, health insurance, pension insurance, property insurance, etc. 5. TTS module. Language processing. Mainly simulates the process of human understanding of natural language, text regularization, word segmentation, grammatical analysis and semantic analysis, so that the computer can fully understand the input text and give various pronunciation prompts required for the latter two parts. Rhythm processing. Plan the segment features for the synthesized speech, such as pitch, length and intensity, so that the synthesized speech can correctly express the meaning and sound more natural. Acoustic processing. Output the speech according to the requirements of the processing results of the first two parts, that is, the synthesized speech.

[0092] Step S202: Perform speech recognition on the speech data to obtain corresponding text.

[0093] In this embodiment, the speech data can be converted into text by performing speech recognition on the speech data based on the semantic recognition module. Specifically, the speech is first preprocessed, including denoising, volume normalization, etc., to improve the accuracy of speech recognition. Features such as Mel frequency cepstral coefficients (MFCC) are extracted from the preprocessed speech data, which can reflect the acoustic characteristics of the speech. The extracted features are then matched with the model in the acoustic model library to identify the corresponding phonemes or words. The word sequence recognized by the acoustic model is subsequently decoded using a language model to generate text that best conforms to grammatical and semantic rules.

[0094] Step S203: extract information from the text to obtain corresponding key information.

[0095] In this embodiment, the specific implementation process of extracting information from the text to obtain the corresponding key information will be further described in detail in subsequent specific embodiments of the present application and will not be elaborated on here.

[0096] Step S204, performing intent recognition on the text to obtain a corresponding intent recognition result.

[0097] In this embodiment, the above-mentioned specific implementation process of performing intent recognition on the text to obtain the corresponding intent recognition result will be further described in detail in subsequent specific embodiments of the present application, and will not be elaborated in detail here.

[0098] Step S205, querying a preset knowledge base for search results that match the key information and the intent recognition result.

[0099] In this embodiment, the above-mentioned knowledge base may specifically be an insurance knowledge base. The key information and intent recognition results obtained may be used as query conditions. The query conditions can accurately reflect the customer's query needs and help the system find relevant entries in the insurance knowledge base. Then determine the insurance knowledge base to be searched. The insurance knowledge base is a database containing a large amount of insurance sales skills, customer development, compliance business development and product knowledge. The query conditions are then entered into the knowledge base search system and a search operation is performed. The system will match and filter in the knowledge base according to the query conditions, find out the entries related to the customer's intent and key information, and use them as the corresponding search results. In addition, the search results can also be presented to the agent. This can be presented in a variety of forms such as text, graphics, audio or video. When presenting the results, it is necessary to ensure the accuracy and clarity of the information so that the agent can quickly understand and make decisions.

[0100] Step S206: determining a target answer matching the customer from the search results.

[0101] In this embodiment, the specific implementation process of determining the target answer matching the customer from the search results will be described in further detail in subsequent specific embodiments of the present application and will not be elaborated on here.

[0102] Step S207: convert the target answer into a corresponding designated voice, and output the designated voice to the wearable device.

[0103] In this embodiment, the target answer can be input into the TTS module. The TTS module converts the target answer into a voice signal, i.e., the specified voice, and outputs it to the Bluetooth headset in the wearable device, so that the agent can use the wearable device to quickly obtain the optimal target answer, thereby improving the accuracy and quality of the response to the customer, and is conducive to improving the efficiency and accuracy of business development and recruitment.

[0104] During the conversation between the agent and the customer, the present application collects the customer's voice data through the use of the agent's wearable device, performs voice recognition on the voice data to obtain the corresponding text, then performs information extraction and intent recognition on the text to obtain the corresponding key information and intent recognition results, and then queries the search results that match the key information and intent recognition results based on the use of the knowledge base, and determines the target answer that matches the customer from the search results, and then converts the target answer into a corresponding specified voice output to the wearable device, so that the agent can use the wearable device to quickly obtain the target answer that matches the customer, and then can assist the agent to quickly and accurately answer the customer's questions, which is conducive to improving the efficiency of conversation processing with the customer and the quality of response.

[0105] In some optional implementations, step S204 includes the following steps:

[0106] Perform syntactic analysis on the text to obtain corresponding syntactic analysis features.

[0107] In this embodiment, the text can be syntactically analyzed by using a syntactic analyzer (such as Stanford Parser, NLTK Parser). Specifically, the syntactic analyzer will identify sentence components (such as subject, predicate, object) and grammatical relations (such as subject-predicate relationship, verb-object relationship) in the text based on a pre-trained model, wherein the syntactic analysis result, i.e., the syntactic analysis feature, will be output in the form of a tree structure (such as a syntax tree).

[0108] Perform semantic role labeling on the text to obtain corresponding semantic role labeling features.

[0109] In this embodiment, semantic role annotation can be performed on the text after syntactic analysis by using a semantic role annotation tool (such as PropBank, VerbNet). Specifically, the semantic role annotation tool will assign a semantic role (such as agent, patient) to each component in the sentence based on a predefined semantic framework. The semantic role annotation result, that is, the semantic role annotation feature, will be output in the form of sentence component-semantic role pairs.

[0110] Performing a text structure analysis on the text to obtain corresponding text structure analysis features.

[0111] In this embodiment, the text can be analyzed by using a text structure analysis tool (such as Rhetorical Structure TheoryParser). Specifically, the text structure analysis tool will identify the logical relationship (such as cause and effect, transition, and progression) between sentences and paragraphs in the text based on a pre-trained model. The text structure analysis results, that is, the text structure analysis features, will be output in the form of a text structure tree or a relationship matrix.

[0112] Emotion recognition is performed on the text to obtain corresponding emotion recognition features.

[0113] In this embodiment, emotion recognition can be performed on the text by using a sentiment analysis model (such as a method based on SVM or neural network). Specifically, the model sentiment analysis will judge the emotional tendency of the text (such as positive, negative, neutral) based on the vocabulary, syntactic structure and contextual information in the text. Among them, the emotion recognition result, that is, the emotion recognition feature, will be output in the form of an emotion label.

[0114] Based on a preset intention recognition model, the syntactic analysis features, the semantic role labeling features, the text structure analysis features and the emotion recognition features are processed for intention recognition to obtain corresponding output results.

[0115] In this embodiment, all the features extracted above (including syntactic analysis features, semantic role labeling features, paragraph structure analysis features, and emotion recognition features) can be fused to form a comprehensive target feature vector. The target feature vector is then input into the intent recognition model. The intent recognition model will output a probability distribution of one or more intent labels, and then select the most likely intent label as the inference result based on the probability distribution, that is, the above-mentioned intent recognition result. Among them, according to the complexity of the problem and the scale of the data, a suitable intent inference model can be selected, such as a rule-based method (such as regular expressions, decision trees), a machine learning-based method (such as support vector machines, random forests, neural networks, etc.). Then, the labeled data is used to train the model, and the performance is optimized by adjusting the parameters of the model to obtain the final intent recognition model with intent recognition function.

[0116] The output result is used as the intention recognition result.

[0117] The present application performs syntactic analysis on the text to obtain corresponding syntactic analysis features; performs semantic role labeling on the text to obtain corresponding semantic role labeling features; performs paragraph structure analysis on the text to obtain corresponding paragraph structure analysis features; and performs emotion recognition on the text to obtain corresponding emotion recognition features; then performs intent recognition processing on the syntactic analysis features, the semantic role labeling features, the paragraph structure analysis features, and the emotion recognition features based on a preset intent recognition model to obtain corresponding output results; and subsequently uses the output results as the intent recognition results. The present application performs syntactic analysis, semantic role labeling, paragraph structure analysis, and emotion recognition processing on the text, and then performs intent recognition processing on the obtained syntactic analysis features, semantic role labeling features, paragraph structure analysis features, and emotion recognition features based on the use of an intent recognition model, so as to accurately recognize the intent of the text from multiple dimensions, effectively improving the accuracy of the generated intent recognition results.

[0118] In some optional implementations of this embodiment, step S206 includes the following steps:

[0119] Obtain a customer profile of the customer, and obtain a conversation context corresponding to the text.

[0120] In this embodiment, the customer profile of the above-mentioned customer can be obtained from a preset customer database. The customer profile may include information such as the customer's age, gender, occupation, interests, etc. In addition, key conversation contexts can be extracted from the conversation between the customer and the system, including questions raised by the user, previous answers, and the context of the conversation.

[0121] Based on the customer profile and the conversation context, an initial answer is determined from all the search results.

[0122] In this embodiment, the specific implementation process of determining the initial answer from all the search results based on the customer portrait and the conversation context will be described in further detail in subsequent specific embodiments of the present application and will not be elaborated on here.

[0123] All the initial answers are scored based on a preset scoring strategy to obtain a score for each of the initial answers.

[0124] In this embodiment, the specific implementation process of scoring all the initial answers based on the preset scoring strategy to obtain the scores of each initial answer will be further described in detail in subsequent specific embodiments of the present application and will not be elaborated on here.

[0125] A designated answer with the highest score is selected from all said initial answers.

[0126] In this embodiment, the scores of the initial answers may be numerically compared to select the designated answer with the highest score from all the initial answers.

[0127] A target answer matching the customer is generated based on the specified answer.

[0128] In this embodiment, the specific implementation process of generating the target answer matching the customer based on the specified answer will be described in further detail in subsequent specific embodiments of the present application, and will not be elaborated on here.

[0129] This application obtains the customer portrait of the customer and the conversation context corresponding to the text; then determines the initial answer from all the search results based on the customer portrait and the conversation context; then scores all the initial answers based on a preset scoring strategy to obtain the score of each of the initial answers; then filters out the highest-scoring designated answer from all the initial answers; and finally generates a target answer that matches the customer based on the designated answer. This application determines the initial answer from all search results based on the obtained customer portrait and conversation context; then scores all the initial answers based on the use of a scoring strategy to obtain the score of each of the initial answers; and then automatically, quickly and accurately determines the target answer that matches the customer based on the highest-scoring designated answer filtered out from all the initial answers, effectively improving the intelligence and accuracy of the generation of the target answer.

[0130] In some optional implementations, determining an initial answer from all the search results based on the customer profile and the conversation context includes the following steps:

[0131] Based on a preset natural language processing strategy, the customer portrait, the conversation context and the search results are converted into corresponding customer portrait vectors, conversation context vectors and search result vectors.

[0132] In this embodiment, the customer portrait, conversation context, and text in the search results can be converted into corresponding vector representations by using natural language processing technology (such as TF-IDF, Word2Vec, BERT, etc.) to obtain the corresponding customer portrait vector, conversation context vector, and search result vector.

[0133] A first similarity between the customer portrait vector and the search result vector is calculated, and a second similarity between the conversation context vector and the search result vector is calculated.

[0134] In this embodiment, the first similarity between the customer portrait vector and the search result vector and the second similarity between the conversation context vector and the search result vector can be calculated by using methods such as cosine similarity and Jaccard similarity.

[0135] A designated search result vector whose first similarity is greater than a preset first similarity threshold and whose second similarity is greater than a preset second similarity threshold is selected from all the search result vectors.

[0136] In this embodiment, there is no specific limitation on the values ​​of the first similarity threshold and the second similarity threshold, which can be set according to actual business requirements, for example, they can be set to 0.86 and 0.88 respectively.

[0137] Obtain a specified search result corresponding to the specified search result vector.

[0138] In this embodiment, the above-mentioned specified search results refer to the results whose similarity with the customer portrait vector is greater than the above-mentioned first similarity threshold, and whose similarity with the conversation context vector is greater than the above-mentioned second similarity threshold.

[0139] The specified search result is used as an initial answer.

[0140] The present application converts the customer portrait, the conversation context and the search results into corresponding customer portrait vectors, conversation context vectors and search result vectors based on a preset natural language processing strategy; then calculates the first similarity between the customer portrait vector and the search result vector, and calculates the second similarity between the conversation context vector and the search result vector; then selects from all the search result vectors a specified search result vector whose first similarity is greater than a preset first similarity threshold and whose second similarity is greater than a preset second similarity threshold; subsequently obtains a specified search result corresponding to the specified search result vector, and uses the specified search result as the initial answer. The present application converts the customer portrait, the conversation context and the search results into corresponding customer portrait vectors, conversation context vectors and search result vectors based on the use of a natural language processing strategy, then calculates the first similarity between the customer portrait vector and the search result vector, and calculates the second similarity between the conversation context vector and the search result vector, and then selects from all the search results a specified search result whose similarity with the customer portrait vector is greater than the above-mentioned first similarity threshold and whose similarity with the conversation context vector is greater than the above-mentioned second similarity threshold as the final initial answer, effectively improving the accuracy of generating the initial answer.

[0141] In some optional implementations, scoring all the initial answers based on a preset scoring strategy to obtain a score for each initial answer includes the following steps:

[0142] The answer features of each of the initial answers are extracted based on a preset extraction strategy; wherein the answer features at least include answer relevance features, answer accuracy features, answer authority features, and answer timeliness features.

[0143] In this embodiment, the first similarity and the second similarity obtained by calculation may be used as the answer correlation feature.

[0144] A database containing known facts, professional knowledge, etc. is pre-built. Then, the answer is compared with the information in the knowledge base to check whether the information in the answer is consistent with the information in the knowledge base. Finally, based on the comparison result, a feature value representing the accuracy of the answer (such as accuracy rate, error rate, etc.) is extracted and used as the above-mentioned answer accuracy feature.

[0145] By evaluating the authority of the source of the answer. This is achieved by checking the domain name, website reputation, author identity, etc. of the source of the answer. Then, based on the evaluation results, a feature value representing the authority of the answer is extracted and used as the above answer authority feature. This can be a discrete authority level (such as high, medium, low) or a continuous authority score.

[0146] By extracting the publishing time or updating time from the answer. Then, the timeliness of the answer is evaluated based on the current time and the publishing time or updating time of the answer. This can be achieved by calculating the time difference, comparing timestamps, etc. Finally, based on the evaluation result, a feature value representing the timeliness of the answer is extracted and used as the above answer timeliness feature. This can be a numerical value representing the time difference or a timeliness level divided according to the time difference.

[0147] Obtain a preset weight corresponding to the answer feature.

[0148] In this embodiment, a weight is assigned to each answer feature according to a preset weight setting mechanism, such as principal component analysis. These weights can be adjusted according to business needs and user preferences.

[0149] Get the preset scoring algorithm.

[0150] In this embodiment, the scoring algorithm may specifically adopt a linear weighted sum algorithm.

[0151] The answer features and the preset weights are calculated based on the scoring algorithm to obtain scores for each of the initial answers.

[0152] In this embodiment, a scoring algorithm may be used to multiply the answer feature of each initial answer by the corresponding weight, and the sum is calculated to obtain the score of each initial answer.

[0153] The present application extracts the answer features of each of the initial answers based on a preset extraction strategy; then obtains the preset weights corresponding to the answer features; then obtains a preset scoring algorithm; and subsequently calculates and processes the answer features and the preset weights based on the scoring algorithm to obtain the scores of each of the initial answers. The present application extracts the answer features of each of the initial answers based on the use of an extraction strategy, and obtains the preset weights corresponding to the answer features, and then calculates and processes the answer features and the preset weights based on the use of a scoring algorithm, so that the scores of each of the initial answers can be obtained efficiently and accurately, thereby improving the data accuracy of the scores of the guaranteed initial answers.

[0154] In some optional implementations of this embodiment, generating a target answer matching the customer based on the specified answer includes the following steps:

[0155] The specified answer is sentence adjusted to obtain a corresponding first answer.

[0156] In this embodiment, the sentence structure of the selected specified answer can be adjusted to make it clearer, more concise and easier to understand, which may include reorganizing the sentence, deleting redundant information, using simpler vocabulary, etc.

[0157] The first answer is processed by adding designated terms to obtain a corresponding second answer.

[0158] In this embodiment, appropriate polite words can be added to the first answer to improve the affinity of the answer and customer satisfaction. Specifically, polite words such as "please", "thank you", "sorry" and the like, as well as a more friendly tone and intonation, can be used.

[0159] The second answer is customized based on the customer portrait and the conversation context to obtain a corresponding third answer.

[0160] In this embodiment, the answer can be personalized based on the customer's customer profile and the conversation context, which may include using the customer's name, mentioning the customer's interests or concerns, and providing suggestions related to the customer's career or life scenario.

[0161] The third answer is used as the target answer.

[0162] The present application performs sentence adjustment processing on the designated answer to obtain the corresponding first answer; then performs designated terminology addition processing on the first answer to obtain the corresponding second answer; then performs content customization processing on the second answer based on the customer portrait and the conversation context to obtain the corresponding third answer; and subsequently uses the third answer as the target answer. The present application performs sentence adjustment processing, designated terminology addition processing, and content customization processing on the designated answer to automatically and intelligently optimize the designated answer and obtain the final target answer, effectively improving the quality of the target answer and customer satisfaction.

[0163] In some optional implementations of this embodiment, step S203 includes the following steps:

[0164] The text is preprocessed to obtain a corresponding processed text.

[0165] In this embodiment, the above-mentioned preprocessing may at least include word segmentation, part-of-speech tagging, named entity recognition, and word sense elimination processing. Specifically, word segmentation includes: segmenting the recognized text, that is, dividing the continuous text into independent words. Part-of-speech tagging includes: marking each word with its part of speech, such as noun, verb, adjective, etc. Named entity recognition includes: identifying specific entities in the text, such as names of people, places, institutions, etc. Word sense elimination includes: selecting the word sense that best fits the text context from multiple possible word senses.

[0166] Get the preset information extraction strategy.

[0167] In this embodiment, the above-mentioned information extraction strategy refers to extracting key information from text using key information extraction technology (such as rule-based methods, machine learning-based methods).

[0168] Information is extracted from the processed text based on the information extraction strategy to obtain corresponding specified information.

[0169] In this embodiment, according to the policy content of the above information extraction policy, based on preset rules or models, key information in the text, such as customer questions, needs, etc., can be identified. The key information extraction result is output in the form of key-value pairs or structured data.

[0170] The designated information is used as the key information.

[0171] The present application obtains the corresponding processed text by preprocessing the text; then obtains a preset information extraction strategy; then extracts information from the processed text based on the information extraction strategy to obtain corresponding designated information; and subsequently uses the designated information as the key information. The present application obtains the processed text by preprocessing the text, and then extracts information from the processed text based on the use of an information extraction strategy, thereby efficiently and accurately extracting key information from the text, improving the efficiency of extracting key information and ensuring the accuracy of the key information obtained.

[0172] In some optional implementations, the user information obtained is subject to the user's consent and complies with relevant laws and policies.

[0173] In addition, the non-Company software tools or components that appear in the embodiments of the present application are merely examples and do not represent actual use.

[0174] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0175] It should be emphasized that in order to further ensure the privacy and security of the above target answer, the above target answer can also be stored in a node of a blockchain.

[0176] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0177] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0178] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0179] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0180] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of an answer generation device based on a wearable device. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0181] like Figure 3As shown, the answer generation device 300 based on the wearable device described in this embodiment includes: a collection module 301, a first recognition module 302, an extraction module 303, a second recognition module 304, a query module 305, a determination module 306 and an output module 307. Among them:

[0182] The collection module 301 is used to collect the voice data of the customer based on the wearable device of the agent during the conversation between the agent and the customer;

[0183] A first recognition module 302, configured to perform speech recognition on the speech data to obtain corresponding text;

[0184] An extraction module 303 is used to extract information from the text to obtain corresponding key information;

[0185] The second recognition module 304 is used to perform intent recognition on the text to obtain a corresponding intent recognition result;

[0186] A query module 305 is used to query a preset knowledge base for search results that match the key information and the intention recognition result;

[0187] A determination module 306, configured to determine a target answer matching the customer from the search results;

[0188] The output module 307 is used to convert the target answer into a corresponding specified voice and output the specified voice to the wearable device.

[0189] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the answer generation method based on the wearable device in the aforementioned embodiment, and are not repeated here.

[0190] In some optional implementations of this embodiment, the second identification module 304 includes:

[0191] A first analysis submodule is used to perform syntactic analysis on the text to obtain corresponding syntactic analysis features;

[0192] A labeling submodule, used to label the text with semantic roles to obtain corresponding semantic role labeling features;

[0193] The second analysis submodule is used to perform a text structure analysis on the text to obtain corresponding text structure analysis features;

[0194] A first recognition submodule is used to perform emotion recognition on the text to obtain corresponding emotion recognition features;

[0195] A second recognition submodule is used to perform intent recognition processing on the syntactic analysis feature, the semantic role labeling feature, the text structure analysis feature and the emotion recognition feature based on a preset intent recognition model to obtain a corresponding output result;

[0196] The first determination submodule is used to use the output result as the intention recognition result.

[0197] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the answer generation method based on the wearable device in the aforementioned embodiment, and are not repeated here.

[0198] In some optional implementations of this embodiment, the determination module 306 includes:

[0199] A first acquisition submodule is used to acquire a customer portrait of the customer and acquire a conversation context corresponding to the text;

[0200] A second determination submodule is used to determine an initial answer from all the search results based on the customer portrait and the conversation context;

[0201] A scoring submodule, used to score all the initial answers based on a preset scoring strategy to obtain a score for each initial answer;

[0202] A screening submodule, used to screen out a designated answer with the highest score from all the initial answers;

[0203] A generating submodule is used to generate a target answer matching the customer based on the specified answer.

[0204] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the answer generation method based on the wearable device in the aforementioned embodiment, and are not repeated here.

[0205] In some optional implementations of this embodiment, the second determining submodule includes:

[0206] A conversion unit, configured to convert the customer portrait, the conversation context, and the search results into corresponding customer portrait vectors, conversation context vectors, and search result vectors based on a preset natural language processing strategy;

[0207] A first calculation unit, configured to calculate a first similarity between the customer portrait vector and the search result vector, and to calculate a second similarity between the conversation context vector and the search result vector;

[0208] A screening unit, configured to screen out, from all the search result vectors, a specified search result vector whose first similarity is greater than a preset first similarity threshold and whose second similarity is greater than a preset second similarity threshold;

[0209] A first acquisition unit, configured to acquire a specified search result corresponding to the specified search result vector;

[0210] The first determining unit is configured to use the specified search result as an initial answer.

[0211] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the answer generation method based on the wearable device in the aforementioned embodiment, and are not repeated here.

[0212] In some optional implementations of this embodiment, the scoring submodule includes:

[0213] An extraction unit, configured to extract answer features of each of the initial answers based on a preset extraction strategy; wherein the answer features at least include answer relevance features, answer accuracy features, answer authority features, and answer timeliness features;

[0214] A second acquisition unit, used to acquire a preset weight corresponding to the answer feature;

[0215] A third acquisition unit, used to acquire a preset scoring algorithm;

[0216] The second calculation unit is used to calculate the answer features and the preset weights based on the scoring algorithm to obtain the scores of each of the initial answers.

[0217] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the answer generation method based on the wearable device in the aforementioned embodiment, and are not repeated here.

[0218] In some optional implementations of this embodiment, the generating submodule includes:

[0219] A first processing unit, configured to perform sentence adjustment processing on the specified answer to obtain a corresponding first answer;

[0220] A second processing unit, configured to perform a process of adding a designated term to the first answer to obtain a corresponding second answer;

[0221] A third processing unit, configured to perform content customization processing on the second answer based on the customer portrait and the conversation context to obtain a corresponding third answer;

[0222] The second determining unit is configured to use the third answer as the target answer.

[0223] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the answer generation method based on the wearable device in the aforementioned embodiment, and are not repeated here.

[0224] In some optional implementations of this embodiment, the extraction module 303:

[0225] A preprocessing submodule, used for preprocessing the text to obtain a corresponding processed text;

[0226] The second acquisition submodule is used to acquire a preset information extraction strategy;

[0227] An extraction submodule, used for extracting information from the processed text based on the information extraction strategy to obtain corresponding specified information;

[0228] The third determining submodule is configured to use the designated information as the key information.

[0229] In this embodiment, the operations performed by the above modules or units correspond one-to-one to the steps of the answer generation method based on the wearable device in the aforementioned embodiment, and are not repeated here.

[0230] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0231] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 with components 41-43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0232] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0233] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the answer generation method based on the wearable device. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0234] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer-readable instructions stored in the memory 41 or process data, such as computer-readable instructions for running the answer generation method based on the wearable device.

[0235] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0236] The present application also provides another embodiment, namely, providing a computer-readable storage medium, which stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the answer generation method based on the wearable device as described above.

[0237] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0238] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for generating answers based on a wearable device, characterized in that: The steps include: During the conversation between the agent and the customer, voice data of the customer is collected based on the agent's wearable device; Performing speech recognition on the speech data to obtain corresponding text; Extracting information from the text to obtain corresponding key information; Performing intent recognition on the text to obtain a corresponding intent recognition result; Querying a preset knowledge base for search results that match the key information and the intent recognition result; Determining a target answer matching the customer from the search results; The target answer is converted into a corresponding specified voice, and the specified voice is output to the wearable device.

2. The answer generation method based on a wearable device according to claim 1, characterized in that: The step of performing intent recognition on the text to obtain a corresponding intent recognition result specifically includes: Performing syntactic analysis on the text to obtain corresponding syntactic analysis features; Performing semantic role labeling on the text to obtain corresponding semantic role labeling features; Performing a text structure analysis on the text to obtain corresponding text structure analysis features; Performing emotion recognition on the text to obtain corresponding emotion recognition features; Based on a preset intention recognition model, the syntactic analysis feature, the semantic role labeling feature, the text structure analysis feature, and the emotion recognition feature are processed for intention recognition to obtain a corresponding output result; The output result is used as the intention recognition result.

3. The answer generation method based on a wearable device according to claim 1, characterized in that: The step of determining a target answer matching the customer from the search results specifically includes: Obtaining a customer profile of the customer, and obtaining a conversation context corresponding to the text; Based on the customer profile and the conversation context, determining an initial answer from all the search results; Scoring all the initial answers based on a preset scoring strategy to obtain a score for each initial answer; Selecting a designated answer with the highest score from all the initial answers; A target answer matching the customer is generated based on the specified answer.

4. The answer generation method based on a wearable device according to claim 3, characterized in that: The step of determining an initial answer from all the search results based on the customer portrait and the conversation context specifically includes: Converting the customer portrait, the conversation context, and the search results into corresponding customer portrait vectors, conversation context vectors, and search result vectors based on a preset natural language processing strategy; Calculating a first similarity between the customer portrait vector and the search result vector, and calculating a second similarity between the conversation context vector and the search result vector; Filter out, from all the search result vectors, a specified search result vector whose first similarity is greater than a preset first similarity threshold and whose second similarity is greater than a preset second similarity threshold; Obtaining a specified search result corresponding to the specified search result vector; The specified search result is used as an initial answer.

5. The answer generation method based on a wearable device according to claim 3, characterized in that: The step of scoring all the initial answers based on a preset scoring strategy to obtain the score of each initial answer specifically includes: Extracting answer features of each of the initial answers based on a preset extraction strategy; wherein the answer features at least include answer relevance features, answer accuracy features, answer authority features, and answer timeliness features; Obtaining a preset weight corresponding to the answer feature; Get the preset scoring algorithm; The answer features and the preset weights are calculated based on the scoring algorithm to obtain scores for each of the initial answers.

6. The answer generation method based on a wearable device according to claim 3, characterized in that: The step of generating a target answer matching the customer based on the specified answer specifically includes: Performing sentence adjustment processing on the designated answer to obtain a corresponding first answer; Adding a designated term to the first answer to obtain a corresponding second answer; Customize the second answer based on the customer portrait and the conversation context to obtain a corresponding third answer; The third answer is used as the target answer.

7. The answer generation method based on a wearable device according to claim 1, characterized in that: The step of extracting information from the text to obtain corresponding key information specifically includes: Preprocessing the text to obtain a corresponding processed text; Get the preset information extraction strategy; Extracting information from the processed text based on the information extraction strategy to obtain corresponding specified information; The designated information is used as the key information.

8. An answer generation device based on a wearable device, characterized in that: include: A collection module, used to collect voice data of the customer based on the agent's wearable device during the conversation between the agent and the customer; A first recognition module, used to perform speech recognition on the speech data to obtain corresponding text; An extraction module, used to extract information from the text to obtain corresponding key information; A second recognition module is used to perform intent recognition on the text to obtain a corresponding intent recognition result; A query module, used to query a preset knowledge base for search results that match the key information and the intent recognition result; A determination module, used to determine a target answer matching the customer from the search results; An output module is used to convert the target answer into a corresponding specified voice and output the specified voice to the wearable device.

9. A computer device, characterized in that: It includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the answer generation method based on a wearable device as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps of the answer generation method based on a wearable device as described in any one of claims 1 to 7 are implemented.

Images