Data processing method and device

Abstract

The embodiment of the invention provides a data processing method and device. The data processing method comprises the steps of actively obtaining user attribute data and historical learning data of a target user in response to a login operation of logging in a learning application program by the target user; and selecting target question data from the historical learning data, and constructing user description data based on the teaching prompt data, the target question data and the user attribute data. And inputting the user description data into the auxiliary learning model to obtain auxiliary learning data, and pushing the auxiliary learning data to the target user through the learning application program. When it is detected that the target user logs in the learning application program, the user attribute data and the historical learning data of the target user are actively obtained, through analysis of the user attribute data and the historical learning data, teaching is actively initiated to the target user, the target user is supervised and urged to learn, the teaching quality is improved, and leakage checking and vacancy filling are achieved; and the target user can better master the knowledge to be learned.

Description

Technical Field

[0001] The embodiments of the present specification relate to the field of computer technology, and in particular, to a data processing method and device. Background Art

[0002] With the continuous development of artificial intelligence technology, the application of big models in the field of teaching has achieved many research results. Big models are used to answer various questions raised by users and meet their information acquisition needs. In the field of teaching, a personalized learning plan can be tailored for users by analyzing user-related data.

[0003] In the prior art, when a user has a learning need, exercises are usually generated for the user according to the user's learning needs, or teaching content related to the learning needs is searched for the user. However, this teaching method focuses on the user's active acquisition and has high requirements for the user's learning initiative. When the user does not have the learning initiative, it is impossible to carry out teaching work for the user, the teaching quality is poor, and it is impossible to ensure that the user can master the knowledge to be learned well. Therefore, a more effective data processing method is urgently needed to solve the above problems. Summary of the invention

[0004] In view of this, an embodiment of this specification provides a data processing method. One or more embodiments of this specification also relate to a data processing device, a computing device, a computer-readable storage medium and a computer program product to solve the technical defects existing in the prior art.

[0005] According to a first aspect of an embodiment of this specification, a data processing method is provided, including: In response to a target user logging into a learning application, acquiring user attribute data and historical learning data of the target user; Selecting target topic data from the historical learning data, and constructing user description data based on teaching prompt data, the target topic data and the user attribute data; The user description data is input into the auxiliary learning model to obtain auxiliary learning data, and the auxiliary learning data is pushed to the target user through the learning application.

[0006] Optionally, the selecting target topic data from the historical learning data includes: Determine at least two pieces of initial question data in the historical learning data, and obtain knowledge video viewing data and question answering data respectively associated with the at least two pieces of initial question data; Based on the knowledge video viewing data and the question answering data respectively associated with the at least two initial question data, the target question data is selected from the at least two initial question data.

[0007] Optionally, inputting the user description data into an auxiliary learning model to obtain auxiliary learning data includes: Extracting knowledge points from the target topic data included in the user description data through the auxiliary learning model to obtain target knowledge points; Determining data restriction rules and knowledge point data generation rules associated with the target knowledge point in the teaching prompt data included in the user description data; The auxiliary learning data matching the user attribute data is generated according to the data restriction rule and the knowledge point data generation rule.

[0008] Optionally, after acquiring the user attribute data and historical learning data of the target user in response to the target user's login operation of logging into the learning application, the method further includes: Detecting the auxiliary attributes of the target user based on the historical learning data; In a case where it is determined based on the detection result that the target user is a non-auxiliary user, receiving question data submitted by the target user based on a question page of the learning application; Using the intention recognition module included in the question-answering model in the learning application to recognize the question intention data of the question data; The reasoning question-answering module included in the question-answering model is used to determine a solution strategy for the question data based on the question intention data, and to generate answer data for the question data based on the solution strategy.

[0009] Optionally, after identifying the question intention data of the question data by using the intention recognition module included in the question-answering model in the learning application, the method further includes: In the case where it is determined based on the question intention data that the question data corresponds to a generation type question, a target resource corresponding to the question data is generated using a data generation module included in the question-answering model; When it is determined based on the question intention data that the question data corresponds to a retrieval-type question, the data retrieval module included in the question-answering model is used to retrieve the target answer corresponding to the question data.

[0010] Optionally, after identifying the question intention data of the question data by using the intention recognition module included in the question-answering model in the learning application, the method further includes: Determine an initial answer to the question data based on the question intention data using a multi-round question-answering module included in the question-answering model, and push the initial answer to the target user; receiving the associated question data submitted by the target user for the initial answer data, and rewriting the associated question data into target associated question data based on the question data; The multi-round question-answering module is used to determine target-associated answer data corresponding to the target-associated question data.

[0011] Optionally, after pushing the auxiliary learning data to the target user through the learning application, the method further includes: Receiving learning feedback data from the target user in response to the auxiliary learning data; In a case where the learning feedback data is related to the auxiliary learning data, determining test question data based on the auxiliary learning data, and pushing the test question data to the target user through the learning application; In a case where the learning feedback data is not related to the auxiliary learning data, a knowledge base is retrieved based on the learning feedback data to obtain learning retrieval data.

[0012] Optionally, the training of the auxiliary learning model includes: Inputting the sample user attribute data in the training sample, the sample user learning data associated with the sample user attribute data, and the content restriction data, knowledge point prompt data, question prompt data, and answer prompt data included in the auxiliary learning prompt data into the initial auxiliary learning model to obtain predicted question data; Calculate the loss value based on the sample user attribute data in the training sample and the sample auxiliary learning data corresponding to the sample user learning data, the prediction problem data and the loss function; The parameters of the initial auxiliary learning model are adjusted based on the loss value until the auxiliary learning model that meets the training stop condition is obtained.

[0013] According to a second aspect of an embodiment of this specification, there is provided a data processing device, including: an acquisition module, configured to acquire user attribute data and historical learning data of the target user in response to a login operation of the target user logging into the learning application; A selection module is configured to select target topic data from the historical learning data, and construct user description data based on the teaching prompt data, the target topic data and the user attribute data; The push module is configured to input the user description data into the auxiliary learning model, obtain auxiliary learning data, and push the auxiliary learning data to the target user through the learning application.

[0014] According to a third aspect of an embodiment of this specification, a computing device is provided, including: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the above data processing method are implemented.

[0015] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the above-mentioned data processing method are implemented.

[0016] According to a fifth aspect of the embodiments of this specification, a computer program product is provided, including a computer program or instructions, which implement the steps of the above-mentioned data processing method when executed by a processor.

[0017] The data processing method provided by one embodiment of the present specification actively obtains the user attribute data and historical learning data of the target user in response to the target user's login operation of logging into the learning application. The target topic data is selected from the historical learning data, and the user description data is constructed based on the teaching prompt data, the target topic data and the user attribute data. The user description data is input into the auxiliary learning model to obtain the auxiliary learning data, and the auxiliary learning data is pushed to the target user through the learning application. When it is detected that the target user logs into the learning application, the user attribute data and historical learning data of the target user are actively obtained, and through the analysis of the user attribute data and the historical learning data, teaching is actively initiated to the target user, the target user is urged to learn, the teaching quality is improved, and the gaps are found and filled, so that the target user can better master the knowledge that should be learned. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a schematic diagram of a data processing method provided by an embodiment of this specification; Figure 2 is a flow chart of a data processing method provided by an embodiment of this specification; Figure 3 is a processing flow chart of a data processing method provided by an embodiment of this specification; Figure 4 is a structural schematic diagram of a data processing device provided by an embodiment of this specification; Figure 5 It is a structural block diagram of a computing device provided by an embodiment of this specification. DETAILED DESCRIPTION

[0019] Many specific details are described in the following description to facilitate a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of this specification, so this specification is not limited to the specific implementation disclosed below.

[0020] The terms used in one or more embodiments of this specification are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of this specification. The singular forms of "a", "said" and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0021] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0022] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0023] In one or more embodiments of this specification, a large model refers to a deep learning model with large-scale model parameters, which usually contains hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than 10 trillion model parameters. A large model can also be called a foundation model / foundation model. The large model is pre-trained with large-scale unlabeled corpus to produce a pre-trained model with more than 100 million parameters. This model can adapt to a wide range of downstream tasks, and the model has good generalization ability, such as a large-scale language model (LLM), a multi-modal pre-training model, etc.

[0024] When the big model is used in practice, only a small number of samples are needed to fine-tune the pre-trained model and it can be applied to different tasks. The big model can be widely used in natural language processing (NLP), computer vision and other fields. Specifically, it can be applied to computer vision tasks such as visual question answering (VQA), image caption (IC), image generation, as well as natural language processing tasks such as text-based sentiment classification, text summary generation, and machine translation. The main application scenarios of the big model include digital assistants, intelligent robots, search, online education, office software, e-commerce, intelligent design, etc.

[0025] Figure 1 A schematic diagram of a data processing method provided according to an embodiment of the present specification is shown, which is used to actively initiate teaching to the target user based on the user portrait of the target user when the target user logs in to the learning application, so as to improve the learning initiative of the target user. In actual application, in response to the login operation of the target user logging in to the learning application, the user attribute data and historical learning data of the target user are actively obtained. The target topic data is selected from the historical learning data, and the user description data is constructed based on the teaching prompt data, the target topic data and the user attribute data. The user description data is input into the auxiliary learning model to obtain the auxiliary learning data, and the auxiliary learning data is pushed to the target user through the learning application. When it is detected that the target user logs in to the learning application, the user attribute data and historical learning data of the target user are actively obtained, and through the analysis of the user attribute data and the historical learning data, teaching is actively initiated to the target user, and the target user is urged to learn, the teaching quality is improved, and omissions are checked and filled, so that the target user can better master the knowledge to be learned.

[0026] In this specification, a data processing method is provided. This specification also relates to a data processing device, a computing device, a computer-readable storage medium and a computer program product, which are described in detail one by one in the following embodiments.

[0027] See also Figure 2 , Figure 2 A flow chart of a data processing method provided according to an embodiment of the present specification is shown, which specifically includes the following steps.

[0028] Step 202: In response to a target user's login operation to log into a learning application, user attribute data and historical learning data of the target user are obtained.

[0029] Specifically, the learning application may be an application or mini-program with teaching functions, and the learning application may be logged in through a mobile terminal such as a mobile phone or a computer. The learning application may provide learnable content to the target users, and the target users include but are not limited to primary and secondary school students, high school students, college students, teachers, and other persons with learning needs. The login operation refers to the application login operation or account login operation of the target user logging into the learning application. User attribute information includes but is not limited to user personal information and user learning information such as the target user's grade, age, gender, name, and learning time. Historical learning data refers to learning data such as learning content viewing records, question answering records, and wrong question records generated by the target user in the historical learning process.

[0030] Based on this, the target user will generate a login operation when logging into the learning application. The login operation of the target user can trigger the active teaching process of the learning application. The server of the learning application obtains the user attribute data and historical learning data of the target user. Based on the user attribute data and historical learning data of the target user, the user portrait of the target user can be analyzed. Based on the user portrait, the target user's question-answering records and video viewing records can be determined to understand the target user's question-answering accuracy, video viewing completeness and knowledge point mastery.

[0031] Furthermore, considering that the target user may not need the learning application to actively teach during the process of using the target application to learn, when it is detected that the target user does not need auxiliary teaching, other question data raised by the target user can be processed, which is specifically implemented as follows: The auxiliary attributes of the target user are detected based on the historical learning data; when the target user is determined to be a non-auxiliary user based on the detection result, question data submitted by the target user based on the question page of the learning application is received; the question intent data of the question data is identified using the intention recognition module included in the question-answering model in the learning application; the answer strategy of the question data is determined based on the question intent data using the inference question-answering module included in the question-answering model, and answer data of the question data is generated based on the answer strategy.

[0032] Specifically, the auxiliary attribute indicates whether the target user needs the learning application to actively provide auxiliary teaching services. The detection result is whether the target user needs the learning application to actively teach. The target user is a non-auxiliary user, which means that the learning application does not need to actively provide teaching content or learning guidance content for the target user. The question page of the learning application refers to the application page where question data can be entered. Question data refers to the question content that the target user wants to get the answer to. The question-answering model can be a large language model that has the function of giving answers to questions. The intention recognition module is a functional module in the question-answering model, which is used to identify the question intention of the question data. The reasoning question-answering module is a functional module in the question-answering model that has the functions of question reasoning and multi-round question-answering, and is used to answer questions by answer reasoning. The reasoning question-answering module can handle logical reasoning problems such as mathematical calculations and code generation. The answer strategy can be a problem-solving idea. The answer data is the answer predicted by the question-answering model for the question data.

[0033] Based on this, the auxiliary attributes of the target user are detected based on the historical learning data to detect whether the target user needs the learning application to actively teach; if the target user is determined to be a non-auxiliary user based on the detection result, it means that the target user does not need the learning application to actively teach. Receive the question data submitted by the target user based on the question page of the learning application. Use the intention recognition module included in the question-answering model in the learning application to identify the question intent data of the question data, determine the question intent of the question data submitted by the target user, use the reasoning question-answering module included in the question-answering model to determine the answer strategy of the question data based on the question intent data, and generate answer data of the question data based on the answer strategy, and send the answer data to the target user through the learning application.

[0034] In actual applications, after the target user submits question data in response to the Pythagorean theorem knowledge point content sent by the learning application, the question data can be subjected to intent recognition to determine whether the question data is submitted in response to the Pythagorean theorem knowledge point sent by the learning application. If it is determined through intent recognition that the question data is not submitted in response to the Pythagorean theorem knowledge point sent by the learning application, answer retrieval or answer generation is performed for the question data submitted by the target user.

[0035] In summary, the auxiliary attributes of the target user are detected based on the historical learning data, so as to determine whether the target user needs the learning application to actively teach and improve the learning experience of the target user.

[0036] Furthermore, since the learning application can provide target users with a variety of services such as content generation and question-answering, when receiving different types of questions from target users, it is necessary to select a corresponding processing method to answer the target users' questions. The specific implementation is as follows: When it is determined based on the question intention data that the question data corresponds to a generation-type question, the data generation module included in the question-answering model is used to generate a target resource corresponding to the question data; when it is determined based on the question intention data that the question data corresponds to a retrieval-type question, the data retrieval module included in the question-answering model is used to retrieve a target answer corresponding to the question data.

[0037] Specifically, generation problems can be text content generation, image generation, video generation, code generation or execution, etc.; retrieval problems can be question-answering problems, and the target answer corresponding to the question data can be obtained by searching in the knowledge base or searching the Internet.

[0038] Based on this, when it is determined based on the question intent data that the question data corresponds to generation-type questions such as text content generation, image generation, video generation, code generation or execution, the data generation module contained in the question-answering model is used to generate the target resources corresponding to the question data; when it is determined based on the question intent data that the question data corresponds to retrieval-type questions such as question and answer, the data retrieval module contained in the question-answering model is used to retrieve the target answer corresponding to the question data.

[0039] To summarize, based on the question intent data, different types of question data are determined, and different modules included in the question-answering model are selected for processing to improve the question-answering effect and ensure the accuracy of the answer.

[0040] Furthermore, in the multi-round question-answering scenario, the learning application has the ability to answer multiple questions, which is specifically implemented as follows: The multi-round question-and-answer module included in the question-and-answer model is used to determine an initial answer to the question data based on the question intention data, and the initial answer is pushed to the target user; the associated question data submitted by the target user for the initial answer data is received, and the associated question data is rewritten into target-associated question data based on the question data; the multi-round question-and-answer module is used to determine the target-associated answer data corresponding to the target-associated question data.

[0041] Specifically, the multi-round question-and-answer module is used to process multiple questions raised continuously by the target user. The initial answer to the question data refers to the answer data obtained from the first round of question-and-answering by the target user. The associated question data refers to the question data that is relevant to the question data in terms of the question essence dimension. The target associated question data can be the question data obtained by adding an object to the associated question data. The target associated answer data is the answer to the target associated question data.

[0042] Based on this, the multi-round question-answering module included in the question-answering model is used to determine the initial answer to the question data based on the question intention data, and the initial answer is pushed to the target user. When the associated question data submitted by the target user for the initial answer data is received, it means that the target user has asked a follow-up question associated with the question data based on the question data, and it is necessary to rewrite the associated question data into target associated question data based on the question data, so that the target associated question data is a question short sentence containing a subject, a predicate, and an object. The multi-round question-answering module is used to determine the target associated answer data corresponding to the target associated question data.

[0043] Using the above example, when the question data submitted by the target user is: "Give a few examples of practical application of knowledge point A", the question intent data is determined to be a retrieval question and answer, and the multi-round question and answer module included in the question and answer model is used to determine the initial answer to the question data based on the question intent data, and the initial answer is pushed to the target user. When the associated question data submitted by the target user for the initial answer is received: "What about knowledge point B?", it means that the target user has asked a follow-up question after the question data, and the associated question data "What about knowledge point B?" needs to be rewritten based on the question data to: "Give a few examples of practical application of knowledge point A", and then the target associated answer data corresponding to the target associated question data is determined using the multi-round question and answer module, and the associated answer data is sent to the target user.

[0044] In summary, the associated question data submitted by the target user for the initial answer data are summarized and rewritten as target associated question data based on the question data, thereby improving the accuracy of the target associated answer data.

[0045] Step 204: Select target topic data from the historical learning data, and construct user description data based on the teaching prompt data, the target topic data and the user attribute data.

[0046] Specifically, after obtaining the user attribute data and historical learning data of the target user in response to the login operation of the target user logging into the learning application, the target question data can be selected from the historical learning data, and the user description data can be constructed based on the teaching prompt data, the target question data and the user attribute data, wherein the target question data can be questions that the target user answered incorrectly when answering questions, or can be questions with simple knowledge points but easy to make mistakes in answering questions determined by analyzing the knowledge points in the historical learning data and the questions corresponding to the knowledge points. Teaching prompt data refers to prompt words used to guide the auxiliary learning model to process data. User description data refers to data that can be directly input into the auxiliary learning model for processing.

[0047] Based on this, after obtaining the user attribute data and historical learning data of the target user in response to the target user's login operation to log into the learning application, target question data that can be used for the target user to continue learning or review knowledge points are selected from the historical learning data, and user description data is constructed based on the teaching prompt data, target question data and user attribute data, so as to facilitate the subsequent use of the auxiliary learning model to predict the auxiliary learning data that can be provided to the target user for question testing or knowledge point review.

[0048] Furthermore, considering that the target users need to watch knowledge videos or learn and understand knowledge points before answering questions during the learning process, the degree of knowledge video viewing and the depth of knowledge point learning will affect the accuracy of answering questions. When selecting target question data, you can refer to the knowledge video viewing data and question answering data. The specific implementation is as follows: Determine at least two initial question data in the historical learning data, and obtain the knowledge video viewing data and question answering data respectively associated with the at least two initial question data; based on the knowledge video viewing data and question answering data respectively associated with the at least two initial question data, select the target question data from the at least two initial question data.

[0049] Specifically, at least two pieces of initial question data may be question data with incorrect answers in the historical learning data. The knowledge video viewing data may be data such as the viewing time, viewing duration, and viewing completeness of the knowledge video corresponding to the initial question data. The question answering data may be the question answering time (including the answering start time and answering end time) and the answering result.

[0050] Based on this, at least two question data with wrong answer results are determined in the historical learning data as initial question data, and the knowledge video viewing data and question answering data respectively associated with the at least two initial question data are obtained. Based on the analysis of the knowledge video viewing data and question answering data respectively associated with the at least two initial question data, target question data with a long answering time, wrong answer results, and low knowledge video viewing completeness are selected from the at least two initial question data. In practical applications, the learning depth of the knowledge points corresponding to the question data can also be used as a factor in selecting the target question data.

[0051] Continuing with the above example, select wrong questions 1, wrong questions 2, and wrong questions 3 as the initial question data in the target question data. Determine the completeness of the knowledge video viewing corresponding to each wrong question, as well as the depth of learning the knowledge point learning content. When the completeness of the knowledge video viewing of wrong question 1 is 70%, the depth of learning the knowledge point learning content is 80%, the completeness of the knowledge video viewing of wrong question 2 is 50%, the depth of learning the knowledge point learning content is 50%, the completeness of the knowledge video viewing of wrong question 3 is 40%, and the depth of learning the knowledge point learning content is 40%, you can select wrong question 1, which has a high degree of knowledge video viewing completeness, a high degree of learning depth of knowledge point learning content, but is still wrong when answering, as the target question data.

[0052] To sum up, based on the analysis of the knowledge video viewing data and question answering data respectively associated with at least two initial question data, selecting target question data with longer answering time, wrong answer results, and lower knowledge video viewing completeness from at least two initial question data can improve the accuracy of target question data selection.

[0053] Step 206: Input the user description data into the auxiliary learning model to obtain auxiliary learning data, and push the auxiliary learning data to the target user through the learning application.

[0054] Specifically, after selecting the target topic data in the historical learning data, and constructing the user description data based on the teaching prompt data, the target topic data and the user attribute data, the user description data can be input into the auxiliary learning model to obtain the auxiliary learning data, and the auxiliary learning data can be pushed to the target user through the learning application, wherein the auxiliary learning model can be a large language model with the capabilities of language understanding, context perception, learning, and personalized generation. The auxiliary learning data can be topic data associated with the target topic data and recommended to the target user for answering, knowledge point data associated with the target topic data and recommended to the target user for learning, or video data associated with the target topic data and recommended to the target user for watching. It can also be learning skills and learning lists associated with the target topic data and recommended to the target user. It can also be reminder information generated for the target topic data to remind the target user to review relevant knowledge points.

[0055] Based on this, after selecting the target topic data from the historical learning data, and constructing the user description data based on the teaching prompt data, the target topic data, and the user attribute data, the user description data is input into the auxiliary learning model to obtain the auxiliary learning data, and the auxiliary learning data is pushed to the target user through the learning application for the target user to view, and multiple rounds of interaction with the learning application are continued based on the auxiliary learning data. In the subsequent multiple rounds of interaction, the knowledge points can be continued to be learned according to the auxiliary learning data sent by the learning application, and question data unrelated to the target topic data can also be submitted to the learning application.

[0056] Furthermore, considering that the auxiliary learning model has a high flexibility in data generation, when calling the auxiliary learning model to predict the auxiliary learning data, it is necessary to restrict the data prediction process of the auxiliary learning model, which is specifically implemented as follows: The target topic data contained in the user description data is subjected to knowledge point extraction through the auxiliary learning model to obtain target knowledge points; data restriction rules and knowledge point data generation rules associated with the target knowledge points are determined in the teaching prompt data contained in the user description data; and the auxiliary learning data matching the user attribute data is generated according to the data restriction rules and the knowledge point data generation rules.

[0057] Specifically, the target knowledge point is the knowledge point corresponding to the target question in the target question data. The data restriction rule is used to restrict the content of the output auxiliary learning data during the data prediction process of the auxiliary learning model to prevent the auxiliary learning model from predicting content irrelevant to the user description data. The knowledge point data generation rule is used to guide the auxiliary learning model to generate auxiliary learning data that meets the knowledge point specification, question specification or answer verification specification.

[0058] Based on this, the target topic data contained in the user description data is determined, and the knowledge points of the target topic data contained in the user description data are extracted through the auxiliary learning model to obtain the target knowledge points. The teaching prompt data contained in the user description data is determined, and the data restriction rules and the knowledge point data generation rules associated with the target knowledge points are determined in the teaching prompt data contained in the user description data. The auxiliary learning model is used to generate auxiliary learning data matching the user attribute data according to the data restriction rules and the knowledge point data generation rules.

[0059] Continuing with the above example, the data restriction rule can be "interactive content is limited to the push of knowledge points, the asking of questions and the verification of answers, and does not involve irrelevant topics. Interact with users strictly in accordance with the set process and language specifications. All pushed knowledge points and questions should come from internal API calls. The recommendation of all knowledge points cannot exceed the following 3 ranges: hash table, Pythagorean theorem, TCP three-way handshake"; in the stage of using auxiliary learning models to push knowledge points, the knowledge point data generation rule can be "**Push knowledge points**1. Select a clear, accurate and in-depth new knowledge point to push to users. 2. Ensure that the knowledge points cover key concepts, principles and application scenarios", in the stage of using auxiliary learning models to generate questions, the knowledge point data generation rule can be "**Pose questions**1. Based on the pushed knowledge points, design a targeted and inspiring question. 2. The question should effectively test the user's understanding and application ability of the knowledge point", in the stage of using auxiliary learning models to verify answers, the knowledge point data generation rule can be "**Verify answers**1. If the answer is correct, give affirmation and encouragement, such as: "Well done! You have mastered this knowledge point. ”2. If the answer is wrong, point it out patiently and give the correct answer, such as: "It’s okay, keep trying! The correct answer is [correct answer]. Let’s review this knowledge point again: [knowledge point content]. ”3. If all knowledge points have been recommended within 1 day, please inform the reply user and ask the user whether to repeat the learning. 4. If the user answers incorrectly, you can actively ask the user’s choice: continue to do the questions or re-explain the knowledge point, or change a knowledge point to continue learning". The auxiliary learning model can be used to generate auxiliary learning data that matches the user attribute data according to the data restriction rules and knowledge point data generation rules. When the target question data is determined, that is, the knowledge point of wrong question 1 is the Pythagorean theorem, the relevant knowledge points of the Pythagorean theorem can be pushed to the target user as auxiliary learning data.

[0060] To sum up, by using the auxiliary learning model according to the data restriction rules and knowledge point data generation rules, auxiliary learning data that matches the user attribute data can be generated, so that the generated auxiliary learning data can be more in line with the learning situation and learning progress of the target users, and help the target users to fill in the gaps and further learn the knowledge points that they have not mastered well.

[0061] Furthermore, considering that the learning application can provide a variety of question-answering services, after the auxiliary learning data is pushed to the target user, the target user can learn the knowledge points based on the auxiliary learning data and answer questions, or can answer content unrelated to the auxiliary learning data. The specific implementation is as follows: Receive learning feedback data from the target user regarding the auxiliary learning data; when the learning feedback data is related to the auxiliary learning data, determine test question data based on the auxiliary learning data, and push the test question data to the target user through the learning application; when the learning feedback data is not related to the auxiliary learning data, retrieve a knowledge base based on the learning feedback data to obtain learning retrieval data.

[0062] Specifically, the learning feedback data may be data fed back to the learning application by the target user after receiving the auxiliary learning data. The test question data refers to question data related to the target knowledge point of the target question data, which is sent to the target user and recommended to the target user to answer in order to detect the target user's mastery of the target knowledge point.

[0063] Based on this, learning feedback data from the target user regarding the auxiliary learning data is received; when the learning feedback data is related to the auxiliary learning data, it indicates that the target user wishes to continue learning regarding the auxiliary learning data. At this time, test question data related to the target knowledge points corresponding to the target question data can be generated based on the auxiliary learning data, and the test question data can be pushed to the target user through a learning application for the target user to answer, thereby detecting the target user's mastery of the target knowledge points; when the learning feedback data is unrelated to the auxiliary learning data, it indicates that the target user's learning feedback data is unrelated to the target question data in the auxiliary learning data, as well as the target knowledge points corresponding to the target question data. Based on the learning feedback data, the knowledge base is retrieved to obtain learning retrieval data, and answers to the questions raised by the target user are given.

[0064] Using the above example, after sending the knowledge point data of the Pythagorean theorem as auxiliary learning data to the target user of the learning application, the learning feedback data of the target user is received. If the learning feedback data expresses that the target user wants to learn the knowledge point of the Pythagorean theorem, a test question related to the Pythagorean theorem can be generated and sent to the target user for the target user to answer. If the learning feedback data is irrelevant to the knowledge point of the Pythagorean theorem, such as: what is a hash table, it is determined that the learning feedback data is irrelevant to the auxiliary learning data, and the knowledge base is searched based on "what is a hash table", the search results are obtained, and the search results are sent to the target user as learning search data.

[0065] In summary, in the case where the learning feedback data is related or unrelated to the auxiliary learning data, the learning feedback data is processed and the processing result is sent to the target user to improve the user experience of the target user.

[0066] Furthermore, the training of the auxiliary learning model includes: inputting the sample user attribute data in the training sample, the sample user learning data associated with the sample user attribute data, and the content restriction data, knowledge point prompt data, question prompt data, and answer prompt data contained in the auxiliary learning prompt data into the initial auxiliary learning model to obtain predicted question data; calculating the loss value based on the sample user attribute data in the training sample and the sample auxiliary learning data corresponding to the sample user learning data, the predicted question data, and the loss function; and adjusting the parameters of the initial auxiliary learning model based on the loss value until the auxiliary learning model that meets the training stop condition is obtained.

[0067] Specifically, the training samples are used to train the initial auxiliary learning model. The sample user attribute data refers to the user personal information and user learning information such as the grade, age, gender, name, and learning time of the sample user. The sample user learning data associated with the sample user attribute data refers to the learning data such as the learning content viewing records, question answering records, and wrong question records generated by the sample user in the historical learning process. The auxiliary learning prompt data, as prompt words in the initial auxiliary learning model training process, can guide the knowledge point recommendation, question generation and answer processing of the initial auxiliary learning model, and limit the initial auxiliary learning model to generate learning-related data. The training stop condition may be that the preset number of iterative training times is reached during the model training process, the prediction accuracy of the pre-trained data processing model reaches the accuracy threshold, or the preset pre-training time is reached. This embodiment does not impose any restrictions on the pre-training stop condition.

[0068] Based on this, the sample user attribute data in the training sample, the sample user learning data associated with the sample user attribute data, and the content restriction data, knowledge point prompt data, question prompt data, and answer prompt data contained in the auxiliary learning prompt data are input into the initial auxiliary learning model to obtain the predicted question data output by the initial auxiliary learning model. The loss value is calculated based on the sample auxiliary learning data, predicted question data, and loss function corresponding to the sample user attribute data and the sample user learning data in the training sample; the initial auxiliary learning model is adjusted based on the loss value to obtain an intermediate auxiliary learning model, and sample data for inputting into the intermediate auxiliary learning model and training the intermediate auxiliary learning model are continuously selected in the training sample, and the intermediate auxiliary learning model is continuously trained, and so on, until an auxiliary learning model that meets the training stop condition is obtained.

[0069] In summary, the initial auxiliary learning model is trained based on the sample user attribute data in the training samples, the sample user learning data associated with the sample user attribute data, the content restriction data, knowledge point prompt data, question prompt data and answer prompt data contained in the auxiliary learning prompt data, and the loss function to obtain the auxiliary learning model, realize targeted model training, and improve the usability of the auxiliary learning model.

[0070] The data processing method provided by one embodiment of the present specification actively obtains the user attribute data and historical learning data of the target user in response to the target user's login operation of logging into the learning application. The target topic data is selected from the historical learning data, and the user description data is constructed based on the teaching prompt data, the target topic data and the user attribute data. The user description data is input into the auxiliary learning model to obtain the auxiliary learning data, and the auxiliary learning data is pushed to the target user through the learning application. When it is detected that the target user logs into the learning application, the user attribute data and historical learning data of the target user are actively obtained, and through the analysis of the user attribute data and the historical learning data, teaching is actively initiated to the target user, the target user is urged to learn, the teaching quality is improved, and the gaps are found and filled, so that the target user can better master the knowledge that should be learned.

[0071] The following combination Figure 3 , taking the application of the data processing method provided in this specification in a teaching application as an example, the data processing method is further described. Figure 3 A processing flow chart of a data processing method provided by an embodiment of the present specification is shown, and the specific execution process is described as follows.

[0072] like Figure 3 As shown, at the beginning, the user enters the teaching page of the teaching application (in the form of mini program, web page, etc.) to determine whether it is necessary to actively initiate a teaching task; if it is necessary to actively initiate a teaching task, a question task is initiated according to the business logic, and the user answers the questions and enters multiple rounds of dialogue; if it is detected that the user does not continue the dialogue, the teaching ends.

[0073] In actual applications, whether it is necessary to actively initiate teaching tasks can be determined based on the user's user profile. Collect the user's learning records (answer records), and determine whether it is necessary to actively initiate teaching for the user by analyzing the user. When it is determined that it is necessary to actively initiate teaching for the user, a question-and-answer task can be initiated to the user through prompt words, and the knowledge points can be passed to the big model. The workflow of the big model can be set through prompt words, and multiple rounds of Socratic question-and-answer sessions can be conducted. The big model will determine whether the user has mastered the knowledge points based on the basic logic of the user's answer content. If not, it will actively initiate teaching (ask questions to the user). In addition to using the ability of the big model to ask questions, you can also use the tool call function of the big model to select questions from the question bank. Therefore, by analyzing the user portrait, actively initiate teaching to the user, urge the user to learn, improve the teaching quality, and make up for the gaps, so that the user can better master the knowledge that should be learned.

[0074] Without actively initiating a teaching task, enter the teaching page and wait for the user to ask questions. When receiving a user's question, detect the content of the user's question to see if it contains special intents (text images, code execution, image search, network search, etc.). If it contains special intents, execute the special intent logic; if it does not contain special intents, determine whether it is necessary to conduct Q&A by searching the knowledge base. If it is necessary to conduct Q&A by searching the knowledge base, search the knowledge base. If necessary, continue to call the big model to answer the user's question and enter the processing logic of multiple rounds of dialogue. In the processing logic of multiple rounds of dialogue, it is necessary to determine whether the question needs to be rewritten for each round of dialogue. If the question needs to be rewritten, rewrite the question before determining the answer to the question. When it is determined that it is not necessary to conduct Q&A by searching the knowledge base, the big model can be directly called to answer the user's question and enter the processing logic of multiple rounds of dialogue. In the processing logic of multiple rounds of dialogue, it is necessary to determine whether the question needs to be rewritten for each round of dialogue. If the question needs to be rewritten, rewrite the question before determining the answer to the question.

[0075] Corresponding to the above method embodiment, this specification also provides a data processing device embodiment, Figure 4 FIG. 1 is a schematic diagram showing the structure of a data processing device provided by an embodiment of the present specification. Figure 4 As shown, the device comprises: An acquisition module 402 is configured to acquire user attribute data and historical learning data of the target user in response to a login operation of the target user logging into the learning application; A selection module 404, configured to select target topic data from the historical learning data, and construct user description data based on the teaching prompt data, the target topic data and the user attribute data; The push module 406 is configured to input the user description data into the auxiliary learning model, obtain auxiliary learning data, and push the auxiliary learning data to the target user through the learning application.

[0076] In an optional embodiment, the selection module 404 is further configured to: Determine at least two pieces of initial question data in the historical learning data, and obtain knowledge video viewing data and question answering data respectively associated with the at least two pieces of initial question data; Based on the knowledge video viewing data and the question answering data respectively associated with the at least two initial question data, the target question data is selected from the at least two initial question data.

[0077] In an optional embodiment, the push module 406 is further configured to: Extracting knowledge points from the target topic data included in the user description data through the auxiliary learning model to obtain target knowledge points; Determining data restriction rules and knowledge point data generation rules associated with the target knowledge point in the teaching prompt data included in the user description data; The auxiliary learning data matching the user attribute data is generated according to the data restriction rule and the knowledge point data generation rule.

[0078] In an optional embodiment, the acquisition module 402 is further configured to: Detecting the auxiliary attributes of the target user based on the historical learning data; In a case where it is determined based on the detection result that the target user is a non-auxiliary user, receiving question data submitted by the target user based on a question page of the learning application; Using the intention recognition module included in the question-answering model in the learning application to recognize the question intention data of the question data; The reasoning question-answering module included in the question-answering model is used to determine a solution strategy for the question data based on the question intention data, and to generate answer data for the question data based on the solution strategy.

[0079] In an optional embodiment, the acquisition module 402 is further configured to: In the case where it is determined based on the question intention data that the question data corresponds to a generation type question, a target resource corresponding to the question data is generated using a data generation module included in the question-answering model; When it is determined based on the question intention data that the question data corresponds to a retrieval-type question, the data retrieval module included in the question-answering model is used to retrieve the target answer corresponding to the question data.

[0080] In an optional embodiment, the acquisition module 402 is further configured to: Determine an initial answer to the question data based on the question intention data using a multi-round question-answering module included in the question-answering model, and push the initial answer to the target user; receiving the associated question data submitted by the target user for the initial answer data, and rewriting the associated question data into target associated question data based on the question data; The multi-round question-answering module is used to determine target-associated answer data corresponding to the target-associated question data.

[0081] In an optional embodiment, the push module 406 is further configured to: Receiving learning feedback data from the target user in response to the auxiliary learning data; In a case where the learning feedback data is related to the auxiliary learning data, determining test question data based on the auxiliary learning data, and pushing the test question data to the target user through the learning application; In a case where the learning feedback data is not related to the auxiliary learning data, a knowledge base is retrieved based on the learning feedback data to obtain learning retrieval data.

[0082] In an optional embodiment, the push module 406 is further configured to: Inputting the sample user attribute data in the training sample, the sample user learning data associated with the sample user attribute data, and the content restriction data, knowledge point prompt data, question prompt data, and answer prompt data included in the auxiliary learning prompt data into the initial auxiliary learning model to obtain predicted question data; Calculate the loss value based on the sample user attribute data in the training sample and the sample auxiliary learning data corresponding to the sample user learning data, the prediction problem data and the loss function; The parameters of the initial auxiliary learning model are adjusted based on the loss value until the auxiliary learning model that meets the training stop condition is obtained.

[0083] The data processing device provided by one embodiment of the present specification actively obtains the user attribute data and historical learning data of the target user in response to the target user's login operation of logging into the learning application. The target topic data is selected from the historical learning data, and the user description data is constructed based on the teaching prompt data, the target topic data and the user attribute data. The user description data is input into the auxiliary learning model to obtain the auxiliary learning data, and the auxiliary learning data is pushed to the target user through the learning application. When it is detected that the target user logs into the learning application, the user attribute data and historical learning data of the target user are actively obtained, and through the analysis of the user attribute data and the historical learning data, teaching is actively initiated to the target user, the target user is urged to learn, the teaching quality is improved, and the gaps are found and filled, so that the target user can better master the knowledge that should be learned.

[0084] The above is a schematic scheme of a data processing device of this embodiment. It should be noted that the technical scheme of the data processing device and the technical scheme of the above data processing method belong to the same concept, and the details of the technical scheme of the data processing device that are not described in detail can be referred to the description of the technical scheme of the above data processing method.

[0085] Figure 5 The structure block diagram of a computing device 500 provided according to an embodiment of the present specification is shown. The components of the computing device 500 include but are not limited to a memory 510 and a processor 520. The processor 520 is connected to the memory 510 via a bus 530, and the database 550 is used to store data.

[0086] The computing device 500 also includes an access device 540 that enables the computing device 500 to communicate via one or more networks 560. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 540 may include one or more of any type of network interface (e.g., a network interface card (NIC)) that is wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a world-wide interoperability for microwave access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, and a near field communication (NFC).

[0087] In one embodiment of the present specification, the above components of the computing device 500 and Figure 5 Other components not shown in the figure may also be connected to each other, for example, via a bus. It should be understood that Figure 5 The computing device structure block diagram shown is only for the purpose of illustration, and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0088] The computing device 500 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smart phone), a wearable computing device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or a personal computer (PC). The computing device 500 may also be a mobile or stationary server.

[0089] The processor 520 is used to execute the following computer executable instructions, which implement the steps of the above data processing method when executed by the processor.

[0090] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the above data processing method belong to the same concept, and the details not described in detail in the technical scheme of the computing device can be referred to the description of the technical scheme of the above data processing method.

[0091] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which can implement the steps of the above-mentioned data processing method when executed by a processor.

[0092] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the above data processing method belong to the same concept, and the details not described in detail in the technical scheme of the storage medium can be referred to the description of the technical scheme of the above data processing method.

[0093] An embodiment of the present specification also provides a computer program product, including a computer program or instructions, which implement the steps of the above data processing method when executed by a processor.

[0094] The above is a schematic scheme of a computer program product of this embodiment. It should be noted that the technical scheme of the computer program product and the technical scheme of the above data processing method belong to the same concept, and the details not described in detail in the technical scheme of the computer program product can be referred to the description of the technical scheme of the above data processing method.

[0095] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0096] The computer instructions include computer program codes, which may be in source code form, object code form, executable files or some intermediate forms, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0097] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0098] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0099] The preferred embodiments of this specification disclosed above are only used to help explain this specification. The optional embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that technicians in the relevant technical field can understand and use this specification well.

Claims

1. A data processing method, characterized in that: include: In response to a target user logging into a learning application, acquiring user attribute data and historical learning data of the target user; Selecting target topic data from the historical learning data, and constructing user description data based on teaching prompt data, the target topic data and the user attribute data; The user description data is input into the auxiliary learning model to obtain auxiliary learning data, and the auxiliary learning data is pushed to the target user through the learning application.

2. The data processing method according to claim 1, characterized in that: The selecting target topic data from the historical learning data comprises: Determine at least two pieces of initial question data in the historical learning data, and obtain knowledge video viewing data and question answering data respectively associated with the at least two pieces of initial question data; Based on the knowledge video viewing data and the question answering data respectively associated with the at least two initial question data, the target question data is selected from the at least two initial question data.

3. The data processing method according to claim 1, characterized in that: The step of inputting the user description data into the auxiliary learning model to obtain the auxiliary learning data includes: Extracting knowledge points from the target topic data included in the user description data through the auxiliary learning model to obtain target knowledge points; Determining data restriction rules and knowledge point data generation rules associated with the target knowledge point in the teaching prompt data included in the user description data; The auxiliary learning data matching the user attribute data is generated according to the data restriction rule and the knowledge point data generation rule.

4. The data processing method according to claim 1, characterized in that: After acquiring the user attribute data and historical learning data of the target user in response to the target user's login operation of logging into the learning application, the method further includes: Detecting the auxiliary attributes of the target user based on the historical learning data; In a case where it is determined based on the detection result that the target user is a non-auxiliary user, receiving question data submitted by the target user based on a question page of the learning application; Using the intention recognition module included in the question-answering model in the learning application to recognize the question intention data of the question data; The reasoning question-answering module included in the question-answering model is used to determine a solution strategy for the question data based on the question intention data, and to generate answer data for the question data based on the solution strategy.

5. The data processing method according to claim 4, characterized in that: After the intention recognition module included in the question-answering model in the learning application is used to identify the question intention data of the question data, the method further includes: In the case where it is determined based on the question intention data that the question data corresponds to a generation type question, a target resource corresponding to the question data is generated using a data generation module included in the question-answering model; When it is determined based on the question intention data that the question data corresponds to a retrieval-type question, the data retrieval module included in the question-answering model is used to retrieve the target answer corresponding to the question data.

6. The data processing method according to claim 4, characterized in that: After the intention recognition module included in the question-answering model in the learning application is used to identify the question intention data of the question data, the method further includes: Determine an initial answer to the question data based on the question intention data using a multi-round question-answering module included in the question-answering model, and push the initial answer to the target user; receiving the associated question data submitted by the target user for the initial answer data, and rewriting the associated question data into target associated question data based on the question data; The multi-round question-answering module is used to determine target-associated answer data corresponding to the target-associated question data.

7. The data processing method according to claim 1, characterized in that: After the auxiliary learning data is pushed to the target user through the learning application, the method further includes: Receiving learning feedback data from the target user in response to the auxiliary learning data; In a case where the learning feedback data is related to the auxiliary learning data, determining test question data based on the auxiliary learning data, and pushing the test question data to the target user through the learning application; In a case where the learning feedback data is not related to the auxiliary learning data, a knowledge base is retrieved based on the learning feedback data to obtain learning retrieval data.

8. The data processing method according to claim 1, characterized in that: The training of the auxiliary learning model includes: Inputting the sample user attribute data in the training sample, the sample user learning data associated with the sample user attribute data, and the content restriction data, knowledge point prompt data, question prompt data, and answer prompt data included in the auxiliary learning prompt data into the initial auxiliary learning model to obtain predicted question data; Calculate the loss value based on the sample user attribute data in the training sample and the sample auxiliary learning data corresponding to the sample user learning data, the prediction problem data and the loss function; The parameters of the initial auxiliary learning model are adjusted based on the loss value until the auxiliary learning model that meets the training stop condition is obtained.

9. A data processing device, characterized in that: include: an acquisition module, configured to acquire user attribute data and historical learning data of the target user in response to a login operation of the target user logging into the learning application; A selection module is configured to select target topic data from the historical learning data, and construct user description data based on teaching prompt data, the target topic data and the user attribute data; The push module is configured to input the user description data into the auxiliary learning model, obtain auxiliary learning data, and push the auxiliary learning data to the target user through the learning application.

10. A computing device, characterized in that include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the data processing method according to any one of claims 1 to 8 are implemented.

11. A computer-readable storage medium, characterized in that: It stores computer executable instructions, which, when executed by a processor, implement the steps of the data processing method described in any one of claims 1 to 8.

12. A computer program product, characterized in that The method comprises a computer program or an instruction, which, when executed by a processor, implements the steps of the data processing method according to any one of claims 1 to 8.

Images