Learning condition diagnosis method, system and related device

Abstract

The invention discloses a learning condition diagnosis method and system and a related device, and the method comprises the steps: obtaining target query information, and obtaining multiple pieces of reference information matched with the target query information from a target database; generating a reference diagnosis question matched with the target query information based on an association relationship among the multiple pieces of reference information, and obtaining reply information matched with the reference diagnosis question; and based on the reply information, acquiring a learning condition diagnosis result matched with the target query information. In this way, the learning condition diagnosis effect and accuracy can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a learning situation diagnosis method, system and related devices. Background Art

[0002] In the traditional teaching model, schools mostly refer to unified standards for the teaching process of students. Teachers and parents cannot obtain each student's mastery of each knowledge point in a timely manner, nor can they teach according to each student's personality, interests, specialties and needs, resulting in low learning efficiency. Personalized guidance can be achieved by accurately and comprehensively measuring and monitoring the students' mastery of each knowledge point through a personalized teaching system, which can break the above-mentioned drawbacks of homogeneous teaching and make learning optimization possible. Learning diagnosis is to diagnose the students' mastery of each knowledge point and accurately determine the weak knowledge points that the students have not yet mastered, so that students can study in a targeted manner and improve their learning efficiency. At present, learning diagnosis is mainly performed by teachers based on the teaching situation, or by asking students to take corresponding assessment tests and conducting learning diagnosis based on the test results. This method leads to low diagnostic efficiency and certain limitations in the accuracy of diagnosis.

[0003] In view of this, how to improve the effectiveness and accuracy of learning situation diagnosis has become an urgent problem to be solved. Summary of the invention

[0004] The main technical problem solved by the present application is to provide a learning situation diagnosis method, system and related devices, which can improve the effect and accuracy of learning situation diagnosis.

[0005] In order to solve the above technical problems, a technical solution adopted in the present application is: to provide a learning situation diagnosis method, including: obtaining target query information, obtaining multiple reference information matching the target query information from a target database; based on the association relationship between the multiple reference information, generating a reference diagnostic question matching the target query information, and obtaining reply information matching the reference diagnostic question; based on the reply information, obtaining a learning situation diagnosis result matching the target query information.

[0006] In order to solve the above technical problems, another technical solution adopted in the present application is: to provide a learning situation diagnosis system, including: a first processing module, used to obtain target query information, and obtain multiple reference information matching the target query information from a target database; a second processing module, used to generate a reference diagnostic question matching the target query information based on the association relationship between the multiple reference information, and obtain reply information matching the reference diagnostic question; a diagnosis module, used to obtain a learning situation diagnosis result matching the target query information based on the reply information.

[0007] To solve the above technical problems, another technical solution adopted in the present application is: to provide an electronic device, comprising: a memory and a processor coupled to each other, wherein the memory stores program instructions, and the processor is used to execute the program instructions to implement the method mentioned in the above technical solution.

[0008] In order to solve the above technical problems, another technical solution adopted in the present application is: providing a computer-readable storage medium on which program instructions are stored, and when the program instructions are executed by a processor, the method mentioned in the above technical solution is implemented.

[0009] The beneficial effects of the present application are as follows: Different from the prior art, the learning situation diagnosis method proposed in the present application, after obtaining the target query information, obtains multiple reference information matching the target query information from the target database. Based on the correlation between the multiple reference information, a reference diagnostic question matching the target query information is generated, and the response information matching the reference diagnostic question is obtained. The learning situation diagnosis result is automatically generated based on the response information to improve the efficiency and objectivity of the learning situation diagnosis of the target object. In addition, by decomposing the learning situation diagnosis task into at least one reference diagnostic question, the learning situation diagnosis result is generated based on the response information matching the reference diagnostic question, which can reduce the difficulty of learning situation diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0011] Figure 1 It is a flowchart of an implementation method of the learning situation diagnosis method of the present application;

[0012] Figure 2 yes Figure 1 Step S101 corresponds to a flow chart of another implementation method;

[0013] Figure 3 yes Figure 1 Step S102 corresponds to a flow chart of another implementation method;

[0014] Figure 4 yes Figure 3 Step S302 corresponds to a flow chart of another implementation method;

[0015] Figure 5 It is a flowchart of an implementation method of the training method of the reward model of the present application;

[0016] Figure 6 It is a structural diagram of an implementation method of the learning situation diagnosis system of the present application;

[0017] Figure 7 It is a structural schematic diagram of an embodiment of the electronic device of the present application;

[0018] Figure 8 It is a structural schematic diagram of an implementation method of a computer-readable storage medium of the present application. DETAILED DESCRIPTION

[0019] 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 only part of the embodiments of the present application, not all of the embodiments, and different embodiments can be adaptively combined. 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.

[0020] See also Figure 1 , Figure 1 1 is a flow chart of an implementation method of a learning situation diagnosis method of the present application, the method comprising:

[0021] S101: Acquire target query information, and acquire multiple reference information matching the target query information from a target database.

[0022] In one implementation, target query information is acquired, and a plurality of reference information matching the target query information is retrieved from a target database.

[0023] In one implementation scenario, in order to diagnose the learning status of the target subject, target query information is generated according to the target subject's answers to corresponding target questions in a historical period of time. According to the target query information, multiple matching reference information is retrieved from the constructed target database.

[0024] In another embodiment, a diagnosis instruction triggered by a user is obtained, target query information is generated according to the diagnosis instruction, and a plurality of reference information is retrieved from a target database.

[0025] Specifically, the diagnosis instruction includes a target object and a diagnosis scope specified by the user, and the target query information is determined according to the target object and the diagnosis scope. For example, the diagnosis scope is a historical time range specified by the user, and the target query information is generated according to the specified target object and the target object's answer to the target test question within the historical time range; or, the diagnosis scope includes a knowledge point range specified by the user, and the target query information is generated according to the specified target object and the target object's answer to the target test question corresponding to the knowledge point range.

[0026] S102: Based on the association relationship between the multiple reference information, generate a reference diagnosis question matching the target query information, and obtain answer information matching the reference diagnosis question.

[0027] In one embodiment, the plurality of reference information includes corresponding association relationships. According to the association relationships between the plurality of reference information, the learning situation diagnosis task is decomposed into reference diagnosis questions matching the target query information. Response information matching each reference diagnosis question is obtained. The number of reference diagnosis questions obtained by decomposition may be one or more.

[0028] Specifically, after obtaining the reference diagnostic question, the reference diagnostic question is input into the intelligent analysis model, so as to analyze the reference diagnostic question using the intelligent analysis model and output corresponding answer information.

[0029] In one implementation scenario, the intelligent analysis model is a large language model with better data analysis capabilities. Reference diagnostic questions are input into the intelligent analysis model, and the intelligent analysis model is prompted to answer the reference diagnostic questions, so as to obtain the answer information output by the intelligent analysis model.

[0030] In a specific application scenario, the above-mentioned large language model may include but is not limited to deep neural networks (DNNs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short-term memory networks (LSTM), and generative pre-trained Transformer models, etc., and no specific restrictions are made on the specific structure and specific deployment of the large language model. In addition, it should be noted that the specific structure and specific deployment of the intelligent analysis model mentioned in other implementations of this application can refer to this implementation.

[0031] In another embodiment, after obtaining the reference diagnostic question, answer information matching the reference diagnostic question is retrieved from a corresponding database.

[0032] S103: Based on the reply information, obtain the learning situation diagnosis result matching the target query information.

[0033] In one embodiment, an intelligent analysis model is used to analyze and summarize the acquired response information, and output a learning situation diagnosis result that matches the target query information.

[0034] The learning situation diagnosis method proposed in the present application obtains multiple reference information matching the target query information from the target database after obtaining the target query information. Based on the association relationship between the multiple reference information, a reference diagnostic question matching the target query information is generated, and the response information matching the reference diagnostic question is obtained. The learning situation diagnosis result is automatically generated based on the response information to improve the efficiency and objectivity of the learning situation diagnosis of the target object. In addition, by decomposing the learning situation diagnosis task into at least one reference diagnostic question, the learning situation diagnosis result is generated based on the response information matching the reference diagnostic question, which can reduce the difficulty of learning situation diagnosis.

[0035] See also Figure 2 , Figure 2 yes Figure 1 Step S101 in the figure corresponds to a flowchart of another implementation method. Specifically, the target database includes a learning situation sub-database and a knowledge graph sub-database, and the process of obtaining multiple reference information in step S101 includes:

[0036] S201: Acquire identity information and target knowledge information matching the target object from the target query information.

[0037] In one implementation, identity information and target knowledge information matching the target object are extracted from the target query information.

[0038] In one implementation scenario, the target query information is input into the intelligent analysis model, and the intelligent analysis model is prompted to extract identity information and knowledge point information from the target query information to obtain identity information and target knowledge information that match the target object. The identity information and target knowledge information obtained by extraction are helpful for the subsequent diagnosis of the target object corresponding to the identity information's mastery of the target knowledge information.

[0039] In a specific application scenario, the target knowledge information can be obtained according to the knowledge point path in the target query information. For example, the target query information obtained is "Please evaluate the mastery of all knowledge points in Chapter Y of Student M based on the adaptive assessment data of Student M in Chapter Y of Textbook X". The intelligent analysis model is used to extract information such as "student M", "Textbook X" and "Chapter Y" from the above target query information, and "student M" is used as identity information, and the knowledge points in "Chapter Y of Textbook X" are used as target knowledge information.

[0040] Optionally, in other specific application scenarios, the target knowledge information can also be directly extracted from the target query information. For example, the target query information obtained is "Please evaluate the student M's mastery of knowledge point N based on the evaluation data of the student M's test questions corresponding to knowledge point N", and the identity information extracted by the intelligent analysis model is "student M" and the target knowledge information is "knowledge point N".

[0041] S202: Based on the identity information, obtain the learning situation sub-information matching the target object from the learning situation sub-database.

[0042] In one implementation, the acquired identity information is used to perform information retrieval on the learning situation sub-database to obtain learning situation sub-information matching the target object.

[0043] In one implementation scenario, for different reference objects, the above-mentioned learning situation sub-database stores at least one of the answer records and answer behavior information when answering the target questions. The above-mentioned answer record information includes at least one of the answer time, answer time, answer accuracy and answer content. The answer behavior information includes at least one of the number of answers and reference situation information. Among them, each target question is matched with corresponding explanation information, and the explanation information includes at least one of the solution ideas and explanation videos, and the reference situation information is used to indicate whether to refer to the explanation information corresponding to the target question. After obtaining the identity information, the identity information is matched with the reference object corresponding to the learning situation sub-database to determine the target object matching the above-mentioned identity information from multiple reference objects, and at least one of the answer records and answer behavior information matching the target object is used as the above-mentioned learning situation sub-information.

[0044] S203: Based on the target knowledge information, obtain at least one reference knowledge point related to the target knowledge information from the knowledge graph sub-database; wherein a target association relationship is matched between the target knowledge information and the reference knowledge point.

[0045] In one embodiment, a plurality of reference knowledge points are stored in the knowledge graph sub-database, and the acquired target knowledge information is used to perform knowledge point retrieval on the knowledge graph sub-database to obtain at least one reference knowledge point related to the target knowledge information, and determine the target association relationship between the target knowledge information and the related reference knowledge points.

[0046] In one implementation scenario, the knowledge graph sub-database stores multiple reference knowledge points, each of which is matched with a corresponding knowledge point attribute, and the knowledge point attribute is determined based on at least one of the learning level information, difficulty information, contribution information, path information, and reference association relationship of the reference knowledge point. Among them, the reference association relationship matched by the current reference knowledge point is used to characterize the association relationship between the current reference knowledge point and other reference knowledge points.

[0047] Specifically, the above difficulty information is used to characterize the learning difficulty of the corresponding reference knowledge point. The above learning level information is used to characterize the learning level of the corresponding reference knowledge point; for example, the learning level includes "basic level", "advanced level" and "advanced level", etc., and the corresponding learning level is assigned according to the learning difficulty and importance of the corresponding reference knowledge point. The above path information is used to characterize the acquisition path of the corresponding reference knowledge point; for example, when the reference knowledge point is obtained from a book, the acquisition path of the reference knowledge point is determined according to the position of the reference knowledge point in the corresponding book, and the acquisition path includes at least one of the directory and page number of the corresponding reference knowledge point. The reference association relationship matched by the current reference knowledge point is used to characterize the association relationship between the current reference knowledge point and other reference knowledge points. The association relationship includes the dependency relationship between different reference knowledge points. For example, the dependency relationship includes the predecessor dependency relationship and the successor dependency relationship. The mastery of the current reference knowledge point depends on the predecessor reference knowledge point; or, the association relationship also includes the connection between the reference knowledge points corresponding to different path information. The contribution information of the current reference knowledge point characterizes the contribution of the predecessor knowledge point to the current reference knowledge point.

[0048] In one implementation scenario, after acquiring the target knowledge information, a GraphRAG (Graph-based Retrieval-Augmented Generation) model is used to acquire at least one reference knowledge point related to the target knowledge information from the knowledge graph sub-database.

[0049] S204: Obtain a preset prompt template, and obtain reference information based on the preset prompt template, learning situation sub-information, target knowledge information and at least one reference knowledge point related thereto.

[0050] In one embodiment, the above-obtained learning situation sub-information and at least one reference knowledge point related to the target knowledge information are used as reference information. A preset prompt template is obtained, and the acquired learning situation sub-information, target knowledge information and at least one reference knowledge point related thereto are filled into corresponding positions in the prompt template to obtain a diagnostic prompt text. The prompt text is used to prompt the intelligent analysis model to perform learning situation diagnosis.

[0051] In a specific application scenario, the preset prompt template is: "You are an experienced junior high school mathematics teacher who is good at evaluating the mastery of each target knowledge point by "student ××" based on the "learning situation sub-information" and "reference knowledge points matching the target knowledge points" of "student ××". By filling the learning situation sub-information, target knowledge information and at least one related reference knowledge point into the prompt template, a diagnostic prompt text is obtained.

[0052] In another embodiment, a target knowledge list is constructed based on the target knowledge information. A learning situation list is constructed based on the learning situation sub-information. And, a reference knowledge list is constructed based on the reference knowledge points that match the target knowledge information. The above target knowledge list, learning situation list, and reference knowledge list are filled into corresponding positions in the prompt template to obtain a diagnostic prompt text.

[0053] Specifically, the target knowledge information is arranged in sequence to obtain a target knowledge list. For example, [target knowledge point 1, target knowledge point 2…]. The corresponding learning situation list is formed according to the answer record and answer behavior information when the target object answers the target test questions related to the target knowledge information. For example, {target knowledge point 1: [(test question 1: answer record, answer behavior information), (test question 2: answer record, answer behavior information)]; target knowledge point 2: [(test question 3: answer record, answer behavior information), (test question 4: answer record, answer behavior information)]}. And, based on the knowledge point attributes of the reference knowledge points that match the target knowledge information, a reference knowledge list is constructed. For example, [target knowledge point 1: knowledge point attributes; target knowledge point 2: knowledge point attributes].

[0054] The above scheme obtains learning situation sub-information from the learning situation sub-database and reference knowledge points from the knowledge graph sub-database to provide a reference basis for subsequent learning situation diagnosis, thereby improving the accuracy of diagnosis.

[0055] See also Figure 3 , Figure 3 yes Figure 1 Step S102 in the flowchart corresponds to another embodiment. Specifically, the implementation process of step S102 includes:

[0056] S301: Based on the association relationship between reference information, link information matching the target query information is acquired.

[0057] In one embodiment, the target query information is matched with a diagnostic task, and multiple reference information is input into the intelligent analysis model, prompting the intelligent analysis model to analyze the reference information according to the diagnostic task and the association relationship between each reference information, and obtain multiple reasoning steps corresponding to the reference information. In response to the reference information including at least one reference knowledge point obtained from the knowledge graph sub-database, the association relationship between the reference information includes the association relationship between different reference knowledge points, and the target association relationship between the target knowledge information and the corresponding reference knowledge point.

[0058] Furthermore, the intelligent analysis model is used to summarize all the reasoning steps to obtain the link information.

[0059] Specifically, after the reference information is input into the intelligent analysis model, the intelligent analysis model is prompted to imitate the human way of thinking and divide the complex diagnostic task into multiple reasoning steps according to the association between each reference information. The intelligent analysis model is prompted to summarize and conclude all reasoning steps to obtain more concise link information.

[0060] In a specific application scenario, when the reference information includes reference knowledge point a and reference knowledge point b, and reference knowledge point a is the predecessor knowledge of reference knowledge point b, the intelligent analysis model is used to generate corresponding reasoning steps, including "diagnosing the target object's mastery of reference knowledge point a" and "based on the target object's mastery of reference knowledge point a, diagnosing the target object's mastery of reference knowledge point b".

[0061] In another embodiment, in response to obtaining the diagnostic prompt text through the above-mentioned corresponding embodiment, the diagnostic prompt text is input into the intelligent analysis model, prompting the intelligent analysis model to obtain multiple reasoning steps corresponding to the reference information according to the diagnostic task and the association relationship between the various reference information in the diagnostic prompt text. The specific implementation process can refer to the above-mentioned embodiment.

[0062] S302: Generate a reference diagnostic question based on decomposing the link information, and obtain answer information matching the reference diagnostic question.

[0063] In one embodiment, the link information is obtained after summarization, and the link information is used to prompt the learning situation diagnosis task, decompose the link information into at least one reference diagnosis question, and obtain the response information matching each reference diagnosis information.

[0064] Specifically, the link information is decomposed into at least one reference diagnostic question using the intelligent analysis model, and the answer information matching each reference diagnostic question is obtained. The answer information can be obtained using the intelligent analysis model, or can also be retrieved from a corresponding database.

[0065] The above scheme generates link information that matches the target query information through an intelligent analysis model, which improves the logic of academic diagnosis and helps to improve the accuracy of diagnosis.

[0066] See also Figure 4 , Figure 4 yes Figure 3 Step S302 in the flowchart corresponds to another embodiment. Specifically, the implementation process of step S302 includes:

[0067] S401: Decompose the link information using an intelligent analysis model to generate multiple link sub-information, and obtain a reference diagnosis question matching each link sub-information.

[0068] In one embodiment, the acquired link information is input into the intelligent analysis model to prompt the intelligent analysis model to decompose the link information to obtain a plurality of link sub-information. For each link sub-information, a matching reference diagnosis question is generated using the intelligent analysis model.

[0069] Specifically, the link information is obtained after being summarized by the intelligent analysis model, and is used to prompt the diagnosis task of the learning situation. In order to reduce the difficulty of diagnosis, the link information is decomposed into multiple link sub-information using the intelligent analysis model, and reference diagnostic questions matching each link sub-information are generated, that is, the complex learning situation diagnosis task is decomposed to obtain multiple simplified reference diagnostic questions.

[0070] S402: Obtain multiple candidate responses that match each reference diagnostic question, and use the trained reward model to output a response score that matches each candidate response.

[0071] In one embodiment, for each reference diagnostic question, a plurality of matching candidate answers are generated using the intelligent analysis model. The plurality of candidate answers corresponding to the reference diagnostic question are input into the trained reward model to generate a matching answer score for each candidate answer using the reward model.

[0072] Optionally, in other implementations, the candidate responses may also be retrieved through a network or database.

[0073] S403: For each reference diagnostic question, the candidate answer corresponding to the maximum answer score is used as the answer information.

[0074] In one embodiment, a plurality of matching candidate answers are obtained for each reference diagnostic question, and the candidate answer corresponding to the maximum answer score is used as the corresponding answer information.

[0075] The above scheme improves the fit between the response information and the reference diagnostic question by selecting the response information from multiple candidate responses matching the reference diagnostic question according to the response score, and helps to improve the diagnostic accuracy of the learning situation.

[0076] See also Figure 5 , Figure 5 1 is a flowchart of an implementation method of a reward model training method. Specifically, the reward model training process includes:

[0077] S501: Acquire multiple training samples; wherein the training samples include training diagnostic questions and training responses that match the training diagnostic questions, each training response is matched with a corresponding training label, and the training label is used to characterize the accuracy of the corresponding training response.

[0078] In one embodiment, a plurality of training samples are obtained, the training samples including training diagnostic questions and training responses matching the training diagnostic questions. Each training response is matched with a corresponding training label, and the training label is used to characterize the accuracy of the corresponding training response. Setting a training label matching each training response is helpful in the subsequent training process.

[0079] In one implementation scenario, the training label that matches each training response is annotated by relevant technical personnel.

[0080] S502: Utilize the constructed reward model to output a prediction score that matches the training response.

[0081] In one embodiment, a constructed reward model is obtained, training diagnostic questions and matching training responses are input into the reward model, and the reward model is used to predict predicted scores that match each training response.

[0082] In one implementation scenario, the above-mentioned reward model is a neural network model with semantic analysis capabilities, and the specific structure of the model can refer to the existing neural network model structure.

[0083] S503: For the training samples, based on the training labels and prediction scores matched by each training response, the parameters of the reward model are adjusted until the preset convergence conditions are met to obtain a trained reward model.

[0084] In one embodiment, for each training sample, the training loss is calculated based on the training labels and prediction scores matched by each training response. The training loss is used to adjust the parameters of the reward model until a preset convergence condition is met, thereby obtaining a trained reward model.

[0085] Specifically, for each training sample, the training responses are sorted according to the training labels matched by the corresponding training responses. Based on the sorting results, the training loss is calculated using the Pairwise Ranking Loss function. The obtained training loss is used to adjust the parameters of the reward model until the number of training times reaches the threshold or the training loss converges, then the training is stopped and the trained reward model is obtained.

[0086] In a specific application scenario, the training sample matches three different training responses, namely, training response A, training response B, and training response C. The accuracy of the training responses is sorted according to the corresponding training labels, and it is determined that the accuracy of training response A is greater than the accuracy of training response B, and the accuracy of training response B is greater than the accuracy of training response C. After obtaining the prediction scores corresponding to training response A, training response B, and training response C, the training loss is calculated according to the accuracy sorting order and the prediction scores corresponding to each training response.

[0087] See also Figure 6 , Figure 6 1 is a structural diagram of an embodiment of a learning situation diagnosis system of the present application. The learning situation diagnosis system comprises a first processing module 10, a second processing module 20 and a diagnosis module 30 coupled to each other.

[0088] Specifically, the first processing module 10 is used to obtain target query information, and obtain a plurality of reference information matching the target query information from a target database.

[0089] The second processing module 20 is used to generate a reference diagnostic question matching the target query information based on the association relationship between the multiple reference information, and obtain answer information matching the reference diagnostic question.

[0090] The diagnosis module 30 is used to obtain the learning situation diagnosis results matching the target query information based on the reply information.

[0091] In one embodiment, the target database includes a learning situation sub-database and a knowledge graph sub-database, and the first processing module 10 obtains multiple reference information matching the target query information from the target database, including: obtaining identity information and target knowledge information matching the target object from the target query information; based on the identity information, obtaining learning situation sub-information matching the target object from the learning situation sub-database; and, based on the target knowledge information, obtaining at least one reference knowledge point related to the target knowledge information from the knowledge graph sub-database; wherein a target association relationship is matched between the target knowledge information and the reference knowledge point; obtaining a preset prompt template, and obtaining reference information based on the preset prompt template, the learning situation sub-information, the target knowledge information and at least one reference knowledge point related to it.

[0092] In one embodiment, the learning situation sub-database stores at least one of answer record information and answer behavior information when answering target questions, the answer record information includes at least one of answer time, answer time, answer accuracy and answer content, and the answer behavior information includes at least one of the number of answers and reference situation information; wherein, the reference situation information is used to characterize whether the explanation information corresponding to the target question is referred to; the knowledge graph sub-database stores multiple reference knowledge points, each reference knowledge point is matched with a corresponding knowledge point attribute, and the knowledge point attribute is determined based on at least one of the learning level information, difficulty information, contribution information, path information and reference association relationship of the reference knowledge point; wherein, the reference association relationship matched by the current reference knowledge point is used to characterize the association relationship between the current reference knowledge point and other reference knowledge points.

[0093] In one embodiment, the second processing module 20 generates a reference diagnostic question that matches the target query information based on the association relationship between multiple reference information, and obtains response information that matches the reference diagnostic question, including: obtaining link information that matches the target query information based on the association relationship between the reference information; generating a reference diagnostic question based on decomposing the link information, and obtaining response information that matches the reference diagnostic information.

[0094] In one embodiment, the target query information is matched with a diagnostic task, and the second processing module 20 obtains link information matching the target query information based on the association relationship between the reference information, including: inputting the reference information into the intelligent analysis model, prompting the intelligent analysis model to analyze the reference information according to the association relationship between the diagnostic task and the reference information, and obtaining multiple reasoning steps corresponding to the reference information; using the intelligent analysis model to summarize all reasoning steps to obtain link information.

[0095] In one embodiment, the second processing module 20 generates reference diagnostic questions based on the decomposition of link information, and obtains reply information matching the reference diagnostic questions, including: using an intelligent analysis model to decompose the link information to generate multiple link sub-information, and obtaining reference diagnostic questions matching each link sub-information; obtaining multiple candidate replies matching each reference diagnostic question, and using a trained reward model to output a reply score matching each candidate reply; for each reference diagnostic question, using the candidate reply corresponding to the maximum reply score as reply information.

[0096] In one embodiment, please continue to refer to Figure 6 The learning situation diagnosis system proposed in the present application also includes a training module 40 coupled to the second processing module 20. The method for the training module 40 to train the reward model includes: obtaining multiple training samples; wherein the training samples include training diagnostic questions and training responses matching the training diagnostic questions, each training response is matched with a corresponding training label, and the training label is used to characterize the accuracy of the corresponding training response; using the constructed reward model to output a predicted score matching the training response; for the training samples, based on the training labels and predicted scores matching each training response, adjusting the parameters of the reward model until the preset convergence conditions are met to obtain the trained reward model.

[0097] See also Figure 7 , Figure 7: It is a structural diagram of an embodiment of an electronic device of the present application. The electronic device includes: a memory 50 and a processor 60 coupled to each other. The memory 50 stores program instructions, and the processor 60 is used to execute the program instructions to implement the method mentioned in any of the above embodiments. Specifically, the electronic device includes but is not limited to: a desktop computer, a laptop computer, a tablet computer, a server, etc., which are not limited here. In addition, the processor 60 can also be called a CPU (Center Processing Unit). The processor 60 may be an integrated circuit chip with signal processing capabilities. The processor 60 can also be a general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field-programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. In addition, the processor 60 can be implemented by an integrated circuit chip.

[0098] See also Figure 8 , Figure 8 It is a structural diagram of an embodiment of a computer-readable storage medium of the present application. The computer-readable storage medium 70 stores program instructions 80 that can be executed by a processor. When the program instructions 80 are executed by the processor, the method mentioned in any of the above embodiments is implemented.

[0099] In the several embodiments provided in the present application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device implementation described above is only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0100] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0101] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0102] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each implementation method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program code.

[0103] The above description is only an implementation method of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly used in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A learning situation diagnosis method, characterized in that: include: Acquire target query information, and acquire multiple reference information matching the target query information from a target database; Based on the association relationship between the plurality of reference information, a reference diagnostic question matching the target query information is generated, and answer information matching the reference diagnostic question is obtained; Based on the reply information, obtain the learning situation diagnosis result that matches the target query information.

2. The method according to claim 1, characterized in that The target database includes a learning situation sub-database and a knowledge graph sub-database, and the obtaining of a plurality of reference information matching the target query information from the target database includes: Acquire identity information and target knowledge information matching the target object from the target query information; Based on the identity information, acquiring learning situation sub-information matching the target object from the learning situation sub-database; and, Based on the target knowledge information, obtaining at least one reference knowledge point related to the target knowledge information from the knowledge graph sub-database; wherein a target association relationship is matched between the target knowledge information and the reference knowledge point; Obtain a preset prompt template, and obtain the reference information based on the preset prompt template, the learning situation sub-information, the target knowledge information and at least one of the reference knowledge points related thereto.

3. The method according to claim 2, characterized in that The learning situation sub-database stores at least one of answer record information and answer behavior information when answering the target test question, the answer record information includes at least one of the answer time, the answer time, the answer accuracy rate and the answer content, and the answer behavior information includes at least one of the number of answers and reference information; wherein the reference information is used to indicate whether the explanation information corresponding to the target test question is referred to; The knowledge graph sub-database stores a plurality of the reference knowledge points, each of the reference knowledge points is matched with a corresponding knowledge point attribute, and the knowledge point attribute is determined based on at least one of the learning level information, difficulty information, contribution information, path information and reference association relationship of the reference knowledge point; wherein, the reference association relationship matched by the current reference knowledge point is used to characterize the association relationship between the current reference knowledge point and other reference knowledge points.

4. The method according to claim 1, characterized in that: The step of generating a reference diagnostic question matching the target query information based on the association relationship between the plurality of reference information, and obtaining answer information matching the reference diagnostic question, comprises: Based on the association relationship between the reference information, obtaining link information matching the target query information; Based on decomposing the link information, the reference diagnostic question is generated, and the answer information matching the reference diagnostic question is acquired.

5. The method according to claim 4, characterized in that The target query information is matched with a diagnostic task, and the acquiring link information matching the target query information based on the association relationship between the reference information includes: Inputting the reference information into an intelligent analysis model, prompting the intelligent analysis model to analyze the reference information according to the association between the diagnostic task and the reference information, and obtaining a plurality of reasoning steps corresponding to the reference information; All the reasoning steps are summarized using the intelligent analysis model to obtain the link information.

6. The method according to claim 4, characterized in that The step of generating the reference diagnostic question based on decomposing the link information and acquiring the response information matching the reference diagnostic question comprises: Decomposing the link information by using an intelligent analysis model to generate a plurality of link sub-information, and obtaining the reference diagnostic question matching each link sub-information; Acquire multiple candidate answers that match each of the reference diagnostic questions, and use the trained reward model to output an answer score that matches each of the candidate answers; For each of the reference diagnostic questions, the candidate answer corresponding to the maximum answer score is used as the answer information.

7. The method according to claim 6, characterized in that The training method of the reward model comprises: Acquire multiple training samples; wherein the training samples include training diagnostic questions and training answers matching the training diagnostic questions, each of the training answers is matched with a corresponding training label, and the training label is used to characterize the accuracy of the corresponding training answer; Outputting a prediction score matching the training response using the constructed reward model; For the training samples, based on the training labels and the predicted scores matched by each of the training responses, the parameters of the reward model are adjusted until a preset convergence condition is met, thereby obtaining the trained reward model.

8. A learning situation diagnosis system, characterized in that: include: A first processing module is used to obtain target query information and obtain multiple reference information matching the target query information from a target database; A second processing module is used to generate a reference diagnostic question matching the target query information based on the association relationship between the plurality of reference information, and obtain answer information matching the reference diagnostic question; A diagnosis module is used to obtain a learning situation diagnosis result that matches the target query information based on the response information.

9. An electronic device, characterized in that: include: A memory and a processor coupled to each other, wherein the memory stores program instructions, and the processor is used to execute the program instructions to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having program instructions stored thereon, characterized in that: When the program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images