Intelligent question and answer method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of intelligent robots, and provides an intelligent question answering method and device, electronic equipment and a storage medium, the intelligent question answering method comprises the following steps: obtaining voice information corresponding to a question asked by a user, and carrying out text conversion on the voice information to obtain a question text of the user; obtaining a user portrait model of the user, wherein the user portrait model is generated according to at least one of a mood feature, an emotion feature, a dialogue habit feature and an age feature of the user; and according to the question text and the user portrait model, querying a plurality of corresponding candidate answers, further determining a target answer from the plurality of candidate answers, performing secondary processing on the target answer to enable the processed target answer to accord with the cognitive level of the age group of the user, and outputting the target answer. The accuracy and correlation of question answering are effectively improved, and the user experience and loyalty are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent robot technology, and in particular to an intelligent question-answering method, device, electronic device and storage medium. Background Art

[0002] An intelligent robot is an intelligent device that can interact with humans in natural language. It can provide corresponding answers based on the user's questions, thereby meeting the user's information and entertainment needs. One of the core technologies of intelligent robots is intelligent question-answering technology, that is, according to the user's question, the most appropriate answer is retrieved or inferred from the massive knowledge base, and returned to the user in the form of natural language. Intelligent question-answering technology involves multiple fields, such as natural language processing, information retrieval, knowledge representation, machine learning, etc. At present, intelligent question-answering technology mainly focuses on how to find the most appropriate answer based on the user's question, while ignoring the impact of the user's personalized characteristics on the selection and generation of answers. Therefore, how to find the most appropriate answer for the user based on the user's personalized characteristics has become an urgent problem to be solved. Summary of the invention

[0003] The present application provides an intelligent question-answering method, device, electronic device and storage medium to find the most appropriate answer for a user based on the user's personalized characteristics and age group characteristics.

[0004] In a first aspect, the present application provides an intelligent question-answering method, the method comprising:

[0005] Acquire voice information corresponding to the user's question, and convert the voice information into text to obtain the user's question text;

[0006] Acquire a user portrait model of the user, where the user portrait model is generated based on at least one of the user's tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics;

[0007] According to the question text and the user portrait model, the corresponding target answer is queried and the target answer is output.

[0008] In a second aspect, the present application further provides an intelligent question-answering device, the device comprising:

[0009] A voice processing module, used to obtain voice information corresponding to a user's question, and convert the voice information into text to obtain the user's question text;

[0010] A portrait acquisition module, used to acquire a user portrait model of the user, wherein the user portrait model is generated based on at least one of the user's tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics;

[0011] The answer query module is used to query the corresponding target answer based on the question text and the user portrait model, and output the target answer.

[0012] In a third aspect, the present application further provides an electronic device, the electronic device comprising a memory and a processor;

[0013] The memory is used to store computer programs;

[0014] The processor is used to execute the computer program and implement the intelligent question-answering method as described above when executing the computer program.

[0015] In a fourth aspect, the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor implements the intelligent question-answering method as described above.

[0016] The present application discloses an intelligent question-answering method, device, electronic device and storage medium, which obtains the voice information corresponding to the user's question and converts the voice information into text to obtain the user's question text; further generates a user portrait model according to at least one of the user's tone characteristics, emotional characteristics, conversation habit characteristics and age characteristics; according to the obtained question text and user portrait model, queries the corresponding target answer and outputs the target answer. Compared with the current technology that lacks the influence of user personalized characteristics and age characteristics on answer selection and generation, this method comprehensively considers the user's tone characteristics, emotional characteristics, conversation habits and age characteristics, and realizes more accurate sending of the most appropriate target answer to the user, improves the accuracy and relevance of answering questions, improves the user's understanding and learning of answers, and improves the user's experience and loyalty. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. 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.

[0018] Figure 1 is a schematic flow chart of an intelligent question-answering method provided in an embodiment of the present application;

[0019] Figure 2 It is a schematic flow chart of another intelligent question-answering method provided in an embodiment of the present application.

[0020] Figure 3 is a schematic flow chart of another intelligent question-answering method provided by an embodiment of the present application;

[0021] Figure 4 is a schematic block diagram of an intelligent question-answering system for an AR glasses device provided in an embodiment of the present application;

[0022] Figure 5 A schematic block diagram of an intelligent question-answering device provided in an embodiment of the present application;

[0023] Figure 6 A schematic block diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0025] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0026] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0027] It should also be understood that the term “and / or” used in the specification and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0028] The intelligent question-answering technology in related technologies mainly focuses on how to find the most appropriate answer based on the user's question, while ignoring the impact of user personalized characteristics and user age group characteristics on answer selection and generation. For example, different users may have different tones, emotions, interests, styles and other characteristics for the same question, which will affect the user's satisfaction and acceptance of the answer. Users of different age groups may have different cognitive levels and educational needs for the same question, which will affect the user's understanding and learning of the answer.

[0029] There are three main types of existing intelligent question-answering technologies:

[0030] 1. Rule-based intelligent question-answering technology: This technology parses user questions and generates answers by pre-defining some rules and templates. The advantage of this technology is that it can ensure the correctness and consistency of the answers, but the disadvantage is that it is difficult to cover diverse and complex questions, and requires a lot of manual writing and maintenance of rules and templates.

[0031] 2. Statistical-based intelligent question-answering technology: This technology uses statistical models and machine learning algorithms to learn the mapping relationship between user questions and answers from a large amount of labeled data. The advantage of this technology is that it can handle diverse and complex questions and does not require manual writing of rules and templates. However, the disadvantage is that it requires a large amount of labeled data and it is difficult to ensure the correctness and consistency of the answers.

[0032] 3. Intelligent question-answering technology based on deep learning: This technology uses deep neural networks to learn the implicit relationship between user questions and answers from a large amount of unlabeled data. The advantage of this technology is that it can handle diverse and complex questions, and does not require manual writing of rules and templates, nor does it require a large amount of labeled data. However, the disadvantage is that it requires a lot of computing resources and it is difficult to explain the generation process of the answer.

[0033] In view of the shortcomings of the prior art, the embodiments of the present application provide an intelligent question-answering method, device, electronic device and storage medium. The intelligent question-answering method can be applied to an electronic device, which includes a server or a terminal device, and outputs a target answer that best suits the user's personalized characteristics and age characteristics based on the question text and user portrait model obtained from the user's question, thereby improving the user's understanding and learning of the answer, and improving the user's experience and loyalty.

[0034] The server may be an independent server or a server cluster, and the terminal device may be an AR glasses device, a smart phone, a laptop computer, an intelligent robot or other electronic device.

[0035] It should be noted that in the embodiments of the present application, the intelligent question-answering method can be applied to the AR glasses device scenario, and the target answer that best suits the user's personalized characteristics and age characteristics is output based on the question text and user portrait model obtained from the user's question, thereby improving the user's experience and loyalty. The most suitable target answer is screened for the user, further improving the user's understanding and learning of the answer. The intelligent question-answering method can also be applied to other intelligent question-answering scenarios, which are not limited here.

[0036] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0037] Figure 1is a schematic flow chart of an intelligent question-answering method provided by an embodiment of the present application, such as Figure 1 As shown, the intelligent question-answering method includes steps S101 to S103.

[0038] S101: Acquire voice information corresponding to a question asked by a user, and convert the voice information into text to obtain a text of the user's question.

[0039] When the user needs to ask questions, he needs to ask questions to the AR glasses device. For example, the user can wake up the AR glasses device and say to the AR glasses device "I want to know what the weather is like today". In this way, the voice information corresponding to the user's question can be obtained, and the user's voice information can be converted into text information using speech recognition technology, thereby realizing the conversion from voice to question text.

[0040] For example, the way to wake up the AR glasses device can be manually touching the power button or directly calling by voice. For example, when the user asks a question to the AR glasses device, the voice calls "Xiao A", "Xiao A", and the AR glasses device responds "I'm here" after hearing the user's call, and then wakes up the user's AR glasses device. Then the user can ask questions to the AR glasses device.

[0041] Exemplarily, speech recognition technology can use Azure Speech Services to convert speech information to obtain corresponding question text.

[0042] S102: Obtain a user portrait model of the user, where the user portrait model is generated based on at least one of the user's tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics.

[0043] The user portrait model is mainly used to describe the personalized characteristics of the user, such as the user's tone characteristics, emotional characteristics, conversation habit characteristics and age characteristics. Specifically, based on multi-source information fusion technology, a user portrait model can be generated by comprehensively considering the user's tone characteristics, emotional characteristics, conversation habit characteristics and age characteristics.

[0044] For example, multi-source information fusion technology, such as multi-layer perceptron (MLP) technology, takes various personalized features of users as input and outputs a multi-dimensional vector as a representation of the user portrait model, which can reflect the user's personality, needs, expectations, etc. The purpose is to generate a model that can describe user characteristics so as to generate more appropriate answers later.

[0045] In one embodiment, in order to improve the user experience, before obtaining the user portrait model of the user, it also includes: obtaining the user's setting operation instruction, the setting operation instruction includes a voice instruction or a gesture operation instruction; according to the setting operation instruction, setting the user's age characteristics and conversation habit characteristics, wherein the set age characteristics and conversation habit characteristics of the user are used to update the user portrait model. The setting operation instruction can be a voice instruction or a gesture operation instruction, for example, the voice instruction can specifically be "change the age characteristic to xx years old", etc., and the gesture operation instruction can predefine different characteristics corresponding to specific gestures, such as the OK gesture corresponding to the age characteristic of xx years old, etc. It should also be noted that the operation can be set or modified, and then the user portrait model can be updated. This makes it convenient for users to choose according to their own preferences, thereby improving the user experience.

[0046] It should be noted that the above setting operation instructions are selectively executed by the user according to his or her actual needs. If the user does not need this service, no setting is required.

[0047] S103: According to the question text and the user portrait model, query the corresponding target answer and output the target answer.

[0048] Specifically, based on the knowledge graph technology, the target answer can be determined from multiple candidate answers according to the question text and the user portrait model. Among them, the knowledge graph technology, for example, can adopt the graph neural network (GNN) technology to use the entities and relationships in the knowledge graph as nodes and edges, the question text and the user portrait model as query vectors, and output a sorted list as candidate answers; and use the sorting learning technology to select a target answer that best matches the user's question text and user portrait model from the candidate answers, and output the target answer. Among them, the sorting learning technology, for example, can be implemented using LambdaRank technology. Outputting the target answer can specifically be displaying the target answer, such as displaying the target answer through an AR glasses device, or broadcasting the voice corresponding to the target answer through voice.

[0049] In some embodiments, a database corresponding to questions and answers, i.e., a preset knowledge base, may also be pre-set. According to the question text and the user portrait model, the corresponding target answer is queried, specifically including: according to the question text and the user portrait model, multiple candidate answers are queried from the preset knowledge base; and the target answer is determined from the multiple candidate answers. The purpose is to retrieve or infer the most appropriate answer from the massive knowledge base.

[0050] In some embodiments, outputting the target answer includes: determining a corresponding text processing technology based on the age characteristics of the user; performing secondary processing on the target answer based on the determined text processing technology so that the processed target answer meets the cognitive level of the user's age group; converting the processed target answer into voice information, and playing the voice information.

[0051] Exemplarily, the technology used for the secondary processing of the target answer includes natural language generation technology. For example, the ChatGPT model is used to perform secondary processing on the target answer to make it consistent with the user's age group and cognitive level, and return it to the user in the form of natural language.

[0052] Preferably, according to the personalized characteristics of the user, before the target answer after secondary processing is returned to the user in the form of natural language, the target answer can be modified using text simplification technology, for example, Text Simplification API technology, to replace complex words and sentences in the target answer with simple and easy-to-understand words and sentences; or text modification technology, such as Text Augmentation API, can be used to add some appropriate rhetoric and emotional colors to the target answer according to the user's conversation habits. This can improve the user's understanding and learning of the answer and increase the user's trust and love for the intelligent electronic device.

[0053] Exemplarily, the technologies used to convert the target answer into speech include speech synthesis technology, such as Azure Speech Services API, to convert the target answer into a speech signal that is the same or similar to the user's language; and speech adjustment technology, such as OpenSMILE toolkit, to add some appropriate tone changes and emotional expressions to the speech signal according to the user's tone characteristics. The purpose is to interact with the user in the form of natural language and speech, and to enhance the communication effect and educational significance between the smart device and the user.

[0054] The intelligent question-answering method provided in the above embodiment can obtain the voice information corresponding to the user's question, convert the voice information into text to obtain the user's question text; further generate a user portrait model based on at least one of the user's tone characteristics, emotional characteristics, conversation habit characteristics and age characteristics; determine the target answer from multiple candidate answers based on the obtained question text and user portrait model, and perform secondary processing on the target answer to output the target answer after secondary processing. Adding the secondary processing steps of obtaining multiple candidate answers and target answers makes the processed target answer meet the cognitive level of the user's age group, thereby improving the user's understanding and learning of the answer. Compared with the current technology that lacks the influence of user personalized characteristics and age characteristics on answer selection and generation, this method comprehensively considers the user's tone characteristics, emotional characteristics, conversation habits and age characteristics, and realizes more accurate sending of the most appropriate target answer to the user, improves the accuracy and relevance of answering questions, and improves the user's experience and loyalty.

[0055] See also Figure 2 , Figure 2 It is a schematic flow chart of another intelligent question-answering method provided by an embodiment of the present application, which may specifically include the following steps S201 and S205.

[0056] S201: Acquire voice information corresponding to a question asked by a user, and convert the voice information into text to obtain a text of the user's question.

[0057] Exemplarily, the user's voice information is converted into text information using speech recognition technology, thereby realizing the conversion from speech to question text, such as Azure Speech Services.

[0058] S202. Obtain an initial user portrait, where the initial user portrait is a user portrait established by asking questions to the user in the form of a questionnaire to obtain the user's personalized characteristics, where the personalized characteristics include at least one of tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics.

[0059] S203: Obtain at least a tone feature and an age feature of the user according to the voice information.

[0060] Specifically, the tone characteristics of the user include but are not limited to happiness, anger, sadness, tension, calmness, etc., and the voice characteristics of the user include but are not limited to volume, speaking speed, pauses, etc. The age characteristics of the user include children, teenagers, adults, the elderly, etc.

[0061] Exemplarily, speech signal processing technology is used to identify the user's tone of voice from the user's voice input. For example, the OpenSMILE toolkit can be used to extract the user's voice features, such as volume, speech speed, pauses and other tone features, and the support vector machine (SVM) can determine whether the user's tone is happy, angry, sad, nervous or calm, and whether the user's age is a child, teenager, adult or elderly based on a pre-trained model.

[0062] S204: Adjust the initial user portrait according to the tone characteristics and age characteristics of the user to obtain the user portrait model.

[0063] Exemplarily, generating a user portrait model includes using questionnaire technology, such as the SurveyMonkey platform, to design some simple and interesting questions, such as "What is your favorite color?", "What is your favorite animal?", "Which is your favorite city?", "What sport are you best at?", "What kind of jokes do you like best?", etc. Based on the user's answers, the user's personalized characteristics are preliminarily determined to generate an initial user portrait model.

[0064] Furthermore, according to the tone characteristics and age characteristics of the user, the initial user portrait model is gradually learned and adjusted to generate a user portrait model.

[0065] For example, the user's initial tone characteristic is pleasant, and in the subsequent question and answer, the user's tone is sad. When the smart electronic device senses the change in the user's tone, the user's tone characteristic setting on the smart electronic device switches from pleasant to sad. At this time, the initial user portrait model is adjusted to generate a user portrait model.

[0066] S205. According to the question text and the user portrait model, query the corresponding target answer and output the target answer.

[0067] Exemplarily, according to the user's question and the user portrait model, based on the knowledge graph technology, a target answer is determined from multiple candidate answers.

[0068] For example, natural language generation technology is used to perform secondary processing on the target answer to make it suitable for the user's age group and cognitive level, thereby improving the user's understanding and learning of the answer.

[0069] The intelligent question-answering method provided in the above embodiment obtains the voice information corresponding to the user's question, converts the voice information into text to obtain the user's question text; further generates a user portrait model based on at least one of the user's tone characteristics, emotional characteristics, conversation habit characteristics and age characteristics; generates an initial user portrait model, and adjusts the initial user portrait based on the user's tone characteristics and age characteristics to obtain a user portrait model. According to the obtained question text and user portrait model, the corresponding target answer is queried, and the target answer is output. This method comprehensively considers the user's tone characteristics, emotional characteristics, conversation habits and age characteristics factors to achieve more accurate sending of the most appropriate target answer to the user, and the user maintains a continuous and in-depth understanding and trust with the intelligent electronic device, which provides a reference for subsequent voice interaction and improves the accuracy and relevance of answering questions.

[0070] See also Figure 3 , Figure 3 This is a schematic flowchart of another intelligent question-answering method provided by an embodiment of the present application, which may specifically include the following steps S301 and S306.

[0071] S301: Acquire voice information corresponding to a question asked by a user, and convert the voice information into text to obtain a text of the question asked by the user.

[0072] S302: Process the voice information to extract voice features of the user.

[0073] S303: Based on a pre-trained feature recognition model, the voice feature is recognized to obtain the tone feature and age feature of the user.

[0074] For example, based on speech signal processing technology, the user's tone characteristics and age characteristics are obtained from the user's voice. For example, the OpenSMILE toolkit extracts the user's voice characteristics, such as volume, speech speed, pauses and other tone characteristics, and the support vector machine (SVM) can judge whether the user's tone is happy, angry, sad, nervous or calm, and whether the user's age is a child, teenager, adult or elderly based on a pre-trained model.

[0075] S304: Acquire the historical conversation records of the user, and determine the emotional characteristics and conversation habit characteristics of the user according to the historical conversation records.

[0076] Specifically, key information can be extracted from the historical conversation records based on a preset BERT model; based on a preset sentiment analysis model, the key information is analyzed to determine the sentiment characteristics and conversation habit characteristics of the user.

[0077] S305: Construct a user portrait model of the user according to at least one of the tone feature, age feature, emotion feature, and conversation habit feature.

[0078] Exemplary technologies that can be used to generate user portrait models include multi-source information fusion technology, such as multi-layer perceptron (MLP) technology, which is implemented by taking various characteristics of the user as input and outputting a multi-dimensional vector as a representation of the user portrait model. The multi-dimensional vector can reflect information such as the user's personality, needs, and expectations.

[0079] S306: According to the question text and the user portrait model, query the corresponding target answer and output the target answer.

[0080] It should be noted that steps S301 to S305 are a process for constructing a user portrait model, which may be constructed before obtaining the user portrait model of the user, or may be constructed during the process of obtaining the user portrait model of the user.

[0081] The intelligent question-answering method provided in the above embodiment obtains the voice information corresponding to the user's question, converts the voice information into text to obtain the user's question text; further generates a user portrait model based on at least one of the user's tone characteristics, emotional characteristics, conversation habit characteristics and age characteristics; queries the corresponding target answer based on the obtained question text and user portrait model, and outputs the target answer. The user's emotional characteristics and conversation habits are obtained based on historical conversation records, and the user portrait model is gradually learned and adjusted to achieve more accurate sending of the most appropriate target answer to the user. The accuracy and relevance of answering questions are improved.

[0082] Specifically, here we take the user using AR glasses as an example to analyze the implementation process of intelligent question answering. Figure 4 , Figure 4 It is a schematic block diagram of an intelligent question-answering system for an AR glasses device provided in an embodiment of the present application, which may specifically include the following steps S401 and S405:

[0083] S401: Generate an initial user portrait model.

[0084] When a user uses an AR glasses device for the first time, the AR glasses device will first ask the user a few questions. For example, the user is asked questions in the form of a questionnaire, such as the SurveyMonkey platform. Through the interaction of several questions, such as "What color do you like", "What is your favorite animal", "Where do you most want to travel", "What subject are you best at", "What kind of jokes do you like best", etc., which can reflect the user's personalized characteristics, the AR glasses device preliminarily determines at least one of the user's tone characteristics, emotional characteristics, conversation habits and age characteristics, and generates an initial user portrait model. When a user uses an AR glasses device for the first time, the trust between the user and the AR glasses device is improved, and a reference is provided for subsequent voice interaction.

[0085] S402: Optimize the initial user portrait model.

[0086] When users use AR glasses devices later, in order to maintain continuous and in-depth understanding and trust, and provide more appropriate personalized feature references for subsequent voice interactions. According to the long-term dialogue between the user and the AR glasses device, the user's emotional characteristics and conversation habits are analyzed from the historical dialogue records between the user and the AR glasses device. Specifically, deep learning technology is used to obtain the user's emotional characteristics and conversation habits. For example, the BERT model, which can be used to extract keywords, topics, intentions and other information from the user's questions; and the Azure Text Analytics API sentiment analysis technology, which is used to determine whether the user's emotions are likes, dislikes, curiosity, boredom or doubts; the LSTM model proposed based on the sequence model technology can learn the user's preferences, interests, style and other information based on the user's historical dialogue, and determine whether the user's dialogue habits are formal, casual, humorous, serious or educational. Gradually learn and adjust the user portrait model. Exemplarily, an online learning technology, such as the ChatGPT model, is used to update and optimize the user portrait model based on each dialogue between the user and the AR glasses device, so that it can more accurately reflect the user's personality, needs, expectations and other information.

[0087] It should be noted that the user can actively set the user's age characteristics, conversation habit characteristics and voice interaction answer style through voice or gesture operation instructions, and update and optimize the user portrait model of the user portrait model.

[0088] S403: Convert the voice information corresponding to the user's question into text to obtain the user's question text.

[0089] Exemplarily, the AR glasses device uses speech recognition technology to convert the user's voice information into text information, thereby realizing the conversion from voice to question text.

[0090] S404: Query the corresponding target answer based on the question text and the user portrait model corresponding to the user voice information.

[0091] Exemplarily, the AR glasses device queries multiple candidate answers from a preset knowledge base based on knowledge graph technology.

[0092] Furthermore, the AR glasses device determines the target answer from multiple candidate answers, and a ranking learning technology, such as LambdaRank technology, can be used here to query the most appropriate answer from a preset knowledge base.

[0093] S405: Perform secondary processing on the target answer.

[0094] Exemplarily, in order to make the target answer suitable for the user's age group and cognitive level, the AR glasses device performs secondary processing on the target answer, using natural language generation technology, such as the ChatGPT model, and returns it to the user in the form of natural language. Use text simplification technology, such as the Text Simplification API, to replace complex vocabulary and sentence patterns in the target answer with simple and easy-to-understand vocabulary and sentence patterns; you can also use text modification technology, such as the Text Augmentation API, to add some appropriate rhetoric and emotional colors to the target answer based on the user's conversation habits. The purpose is to improve the user's understanding and learning of the answer, and increase the user's trust and love for the AR glasses device.

[0095] S406: Output the target answer after secondary processing.

[0096] Exemplarily, the AR glasses device displays the secondary processed target answer in the user's field of view through projection technology.

[0097] The intelligent question-answering method provided by the above-mentioned AR glasses device embodiment obtains the voice information corresponding to the user's question, converts the voice information into text to obtain the user's question text; further generates a user portrait model based on at least one of the user's tone characteristics, emotional characteristics, conversation habit characteristics and age characteristics; based on the obtained question text and user portrait model, queries the corresponding target answer, and outputs the target answer. Among them, the emotional characteristics and conversation habits are analyzed based on the historical conversation records between the user and the AR glasses device, and the user and the AR glasses device maintain continuous and in-depth understanding and trust, so as to achieve more accurate sending of the most appropriate target answer to the user. The accuracy and relevance of answering questions are improved, and the user's love and trust in the AR glasses device are increased.

[0098] Furthermore, taking the voice interaction between users and intelligent robots as an example, the implementation process of intelligent question and answer is analyzed in detail:

[0099] A 6-year-old child uses an intelligent robot to ask a question: "Why is the sky blue?" The intelligent robot device will perform the following steps:

[0100] Identify the user's tone and age characteristics from the user's voice input. For example, by analyzing the user's voice pitch, volume, speaking speed, pauses and other parameters, determine that the user's tone is pleasant; by analyzing the user's voice timbre, range, pronunciation and other parameters, determine that the user is a child.

[0101] Analyze the user's emotional characteristics and conversation habits from the user's conversation content and historical records. For example, by using natural language understanding technology, extract keywords, topics, intentions and other information from the user's questions, and judge the user's emotion to be curious; by using sequence model technology, learn the user's preferences, interests, style and other information from the user's historical conversations, and judge the user's conversation habits to be casual and humorous. Comprehensively consider at least one of the user's tone characteristics, emotional characteristics, conversation habits and age characteristics to generate a user portrait model.

[0102] According to the user's question and the user portrait model, the target answer is determined from multiple candidate answers. Assume there are three candidate answers: "1. The sky is blue because the ability of molecules in the air to scatter the blue part of sunlight is higher than its ability to scatter red light", "2. The sky is blue because the various colors of sunlight are scattered by molecules in the atmosphere, and blue light is most likely to reach our eyes", "3. The sky is blue because the wavelength of blue light is shorter and is more easily scattered by gas molecules in the atmosphere".

[0103] Taking into account at least one of the user's tone characteristics, emotional characteristics, conversation habit characteristics, and age characteristics, it is assumed that the target answer obtained by the intelligent robot is: "The sky is blue because the various colors of sunlight are scattered by molecules in the atmosphere, and blue light is most likely to reach our eyes."

[0104] After the target answer is processed twice, the target answer after secondary processing is: "The sky is blue because there are many colors in sunlight, but only blue is the most powerful and can pass through the air to reach our eyes. So we see the sky as blue!"

[0105] The target answer after the secondary processing is converted into speech. According to the tone characteristics of the user, some appropriate tone changes and emotional expressions are added to the speech signal, and the speech signal is played through the speaker: "The sky is blue because there are many colors in the sunlight, but only blue is the most powerful and can pass through the air to reach our eyes. So we see the sky is blue! (high-pitched and brisk voice)". At this point, the intelligent robot has completed the answer to the question asked by the 6-year-old child.

[0106] See also Figure 5 , Figure 5 The embodiment of the present application provides a schematic block diagram of an intelligent question-answering device, which is used to execute the aforementioned intelligent question-answering method. The intelligent question-answering device can be configured on a server.

[0107] like Figure 5 As shown, the intelligent question-answering device 500 includes: a speech processing module 501, a portrait acquisition module 502 and an answer query module 503.

[0108] The voice processing module 501 is used to obtain voice information corresponding to the user's question, and convert the voice information into text to obtain the user's question text.

[0109] The portrait acquisition module 502 is used to acquire a user portrait model of the user, where the user portrait model is generated based on at least one of the user's tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics.

[0110] The answer query module 503 is used to query the corresponding target answer according to the question text and the user portrait model, and output the target answer.

[0111] In some embodiments, the speech processing module 501 is specifically configured to:

[0112] Acquire voice information corresponding to the user's question, and convert the voice information into text to obtain the user's question text.

[0113] In some embodiments, the portrait acquisition module 502 is specifically used to:

[0114] An initial user portrait is obtained, wherein the initial user portrait is a user portrait established by asking questions to the user in the form of a questionnaire to obtain the user's personalized characteristics, wherein the personalized characteristics include at least one of tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics; at least the tone characteristics and age characteristics of the user are obtained based on the voice information; the initial user portrait is adjusted based on the tone characteristics and age characteristics of the user to obtain the user portrait model.

[0115] In some embodiments, the intelligent question-answering device 500 further includes a model building module, which is specifically used to:

[0116] Obtain a user portrait, process the voice information, and extract the voice features of the user; based on a pre-trained feature recognition model, identify the voice features to obtain the tone features and age features of the user; obtain the historical conversation records of the intelligent electronic device, and determine the emotional features and conversation habit features of the user based on the historical conversation records: extract key information from the historical conversation records based on a preset BERT model; analyze the key information based on a preset sentiment analysis model to determine the emotional features and conversation habit features of the user. Construct a user portrait model for the user based on at least one of the tone features, age features, emotional features, and conversation habit features.

[0117] In some embodiments, the intelligent question-answering device 500 further includes a feature setting module, which is specifically used to:

[0118] Before obtaining the user portrait model of the user, the user's age characteristics and conversation habit characteristics can be selectively set by setting voice commands or gesture operation commands, wherein the set age characteristics and conversation habit characteristics of the user are used to update the user portrait model.

[0119] In some embodiments, the answer query module 503 is specifically configured to:

[0120] According to the question text and the user portrait model, query the corresponding target answer, and according to the question text and the user portrait model, query multiple candidate answers from a preset knowledge base;

[0121] A target answer is determined from the plurality of candidate answers.

[0122] In some embodiments, the answer query module 503 is further configured to:

[0123] Determine a corresponding text processing technology according to the age characteristics of the user, perform secondary processing on the target answer according to the determined text processing technology, so that the processed target answer meets the cognitive level of the user's age group; convert the processed target answer into voice information, and play the voice information.

[0124] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0125] The above-mentioned device can be implemented in the form of a computer program. Figure 6 The electronic device shown is running.

[0126] See also Figure 6 , Figure 6 1 is a schematic block diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may be a server or a terminal device, wherein the terminal device includes a mobile phone, a tablet computer, a notebook or smart glasses.

[0127] See also Figure 6 The electronic device 600 includes a processor 601 and a memory 602 connected via a system bus, wherein the memory may include a non-volatile storage medium and an internal memory. The electronic device may also include a network interface and the like.

[0128] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any intelligent question-answering method.

[0129] The processor is used to provide computing and control capabilities to support the operation of the entire electronic device.

[0130] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any intelligent question-answering method.

[0131] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will appreciate that Figure 6 The structure shown in the figure is merely a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. Specifically, the electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0132] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0133] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0134] Acquire voice information corresponding to a user's question, and convert the voice information into text to obtain the user's question text; acquire a user portrait model of the user, wherein the user portrait model is generated based on at least one of the user's tone characteristics, emotional characteristics, conversation habit characteristics, and age characteristics; query the corresponding target answer based on the question text and the user portrait model, and output the target answer.

[0135] In one embodiment, when acquiring the user portrait model of the user, the processor is used to implement:

[0136] An initial user portrait is obtained, wherein the initial user portrait is a user portrait established by asking questions to the user in the form of a questionnaire to obtain the user's personalized characteristics, wherein the personalized characteristics include one or more of tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics; at least the tone characteristics and age characteristics of the user are obtained based on the voice information; the initial user portrait is adjusted based on the tone characteristics and age characteristics of the user to obtain the user portrait model.

[0137] The voice information is processed to extract the voice features of the user; based on a pre-trained feature recognition model, the voice features are recognized to obtain the tone features and age features of the user; historical conversation records of the intelligent electronic device are obtained, and the emotional features and conversation habit features of the user are determined according to the historical conversation records; a user portrait model of the user is constructed according to at least one of the tone features, age features, emotional features and conversation habit features.

[0138] In one embodiment, when determining the emotion characteristics and conversation habit characteristics of the user according to the historical conversation record, the processor is used to implement:

[0139] Based on a preset BERT model, key information is extracted from the historical conversation records; based on a preset sentiment analysis model, the key information is analyzed to determine the sentiment characteristics and conversation habit characteristics of the user.

[0140] In one embodiment, before obtaining the user portrait model of the user, the processor is used to implement:

[0141] According to the set voice command or gesture operation command, the age characteristics and conversation habit characteristics of the user are set, wherein the set age characteristics and conversation habit characteristics of the user are used to update the user portrait model.

[0142] In one embodiment, when the processor queries the corresponding target answer according to the question text and the user portrait model, it is used to implement:

[0143] According to the question text and the user portrait model, multiple candidate answers are queried from a preset knowledge base; and a target answer is determined from the multiple candidate answers.

[0144] In one embodiment, when outputting the target answer, it is used to achieve:

[0145] Determine a corresponding text processing technology according to the age characteristics of the user; perform secondary processing on the target answer according to the determined text processing technology, so that the processed target answer meets the cognitive level of the user's age group; convert the processed target answer into voice information, and play the voice information.

[0146] A computer-readable storage medium is also provided in an embodiment of the present application, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and the processor executes the program instructions to implement any one of the intelligent question-answering methods provided in the embodiments of the present application.

[0147] The computer-readable storage medium may be an internal storage unit of the electronic device described in the foregoing embodiment, such as a hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, 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 (Flash Card), etc. equipped on the electronic device.

[0148] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. An intelligent question-answering method, characterized in that: include: Acquire voice information corresponding to the user's question, and convert the voice information into text to obtain the user's question text; Acquire a user portrait model of the user, where the user portrait model is generated based on at least one of the user's tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics; According to the question text and the user portrait model, the corresponding target answer is queried and the target answer is output.

2. The intelligent question-answering method according to claim 1, characterized in that: The obtaining of the user portrait model of the user includes: Obtaining an initial user portrait, wherein the initial user portrait is a user portrait established by asking questions to the user in the form of a questionnaire to obtain the user's personalized characteristics, wherein the personalized characteristics include at least one of a tone characteristic, an emotional characteristic, a conversation habit characteristic, and an age characteristic; Acquire at least the tone feature and age feature of the user according to the voice information; The initial user portrait is adjusted according to the tone characteristics and age characteristics of the user to obtain the user portrait model.

3. The intelligent question-answering method according to claim 1, characterized in that: The method further comprises: Processing the voice information to extract voice features of the user; Based on a pre-trained feature recognition model, the voice feature is recognized to obtain the tone feature and age feature of the user; Acquire the historical conversation records of the user, and determine the emotional characteristics and conversation habit characteristics of the user according to the historical conversation records; A user portrait model of the user is constructed according to at least one of the tone characteristics, age characteristics, emotion characteristics and conversation habit characteristics.

4. The intelligent question-answering method according to claim 3, characterized in that: The determining the user's emotional characteristics and conversation habit characteristics according to the historical conversation records includes: Extracting key information from the historical conversation records based on a preset BERT model; Based on a preset sentiment analysis model, the key information is analyzed to determine the sentiment characteristics and conversation habit characteristics of the user.

5. The intelligent question-answering method according to claim 1, characterized in that: The method further comprises: Acquire a setting operation instruction of the user, wherein the setting operation instruction includes a voice instruction or a gesture operation instruction; According to the setting operation instruction, the age characteristics and conversation habit characteristics of the user are set, wherein the set age characteristics and conversation habit characteristics of the user are used to update the user portrait model.

6. The intelligent question-answering method according to claim 1, characterized in that: The querying of the corresponding target answer according to the question text and the user portrait model includes: According to the question text and the user portrait model, query multiple candidate answers from a preset knowledge base; A target answer is determined from the plurality of candidate answers.

7. The intelligent question-answering method according to claim 1, characterized in that: The outputting the target answer comprises: Determining a corresponding text processing technology according to the age characteristics of the user; Performing secondary processing on the target answer according to a determined text processing technique; The processed target answer is converted into voice information, and the voice information is played.

8. An intelligent question-answering device, characterized in that: include: A voice processing module, used to obtain voice information corresponding to a user's question, and convert the voice information into text to obtain the user's question text; A portrait acquisition module, used to acquire a user portrait model of the user, wherein the user portrait model is generated based on at least one of the user's tone characteristics, emotion characteristics, conversation habit characteristics, and age characteristics; The answer query module is used to query the corresponding target answer based on the question text and the user portrait model, and output the target answer.

9. An electronic device, characterized in that: The electronic device comprises a memory and a processor; The memory is used to store computer programs; The processor is used to execute the computer program and implement the intelligent question-answering method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor enables the processor to implement the intelligent question-answering method according to any one of claims 1 to 7.

Images