Test question tutoring method, related device, equipment and storage medium

Abstract

The invention discloses a test question tutoring method, a related device, equipment and a storage medium, and the test question tutoring method comprises the steps: obtaining a personal answer of a target object to a recommended test question and a reference answer of the recommended test question; obtaining a first directed acyclic graph based on the personal answer, and obtaining a second directed acyclic graph based on the reference answer; wherein the directed acyclic graph comprises a plurality of nodes which are connected in sequence, and the plurality of nodes respectively represent each answer step in test question answers; and performing comparison based on the first directed acyclic graph and the second directed acyclic graph to obtain tutoring data for helping the target object to correctly answer the recommended test question. According to the scheme, the refinement degree of test question tutoring can be improved, especially for complex test questions.

Description

Technical Field

[0001] The present application relates to the field of natural language processing technology, and in particular to a test question tutoring method and related devices, equipment and storage media. Background Art

[0002] In traditional offline education scenarios, students often encounter some problems during the homework practice process. For example, teachers have limited time and it is difficult to ensure that all students can receive effective guidance. In addition, the guidance that students can receive varies due to the different distribution of teaching staff in different school regions.

[0003] At present, although there are some intelligent tutoring tools that can automatically evaluate individual answers in order to help tutor individuals to answer questions, the evaluation granularity of existing tutoring tools is usually relatively broad, and they can only provide an overall score, and it is difficult to provide more refined test question tutoring. In addition, existing tutoring tools perform well when processing simple test questions (such as multiple-choice questions, true or false questions, etc.), but they are often unable to cope with complex test questions (such as proof questions, application questions, calculation questions, etc.). In view of this, how to improve the refinement of test question tutoring, especially for complex test questions, 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 test question coaching method and related devices, equipment and storage media, which can improve the refinement of test question coaching, especially for complex test questions.

[0005] In order to solve the above-mentioned technical problems, the first aspect of the present application provides a test question tutoring method, including: obtaining the target object's personal answers to recommended test questions and reference answers to the recommended test questions; based on the personal answers, obtaining a first directed acyclic graph, and based on the reference answers, obtaining a second directed acyclic graph; wherein the directed acyclic graph contains a number of nodes connected in sequence, and the several nodes respectively represent each answering step in answering the test questions; based on comparing the first directed acyclic graph with the second directed acyclic graph, obtaining tutoring data that helps the target object to correctly answer the recommended test questions.

[0006] In order to solve the above-mentioned technical problems, the second aspect of the present application provides a test question tutoring device, including: an acquisition module, a graph output module and a comparison module, the acquisition module is used to obtain the target object's personal answers to recommended test questions and the reference answers to the recommended test questions; the graph output module is used to obtain a first directed acyclic graph based on the personal answers, and to obtain a second directed acyclic graph based on the reference answers; wherein the directed acyclic graph contains a number of nodes connected in sequence, and the several nodes respectively represent each answer step in answering the test questions; the comparison module is used to compare the first directed acyclic graph with the second directed acyclic graph to obtain tutoring data that helps the target object correctly answer the recommended test questions.

[0007] In order to solve the above technical problems, the third aspect of the present application provides an electronic device, which at least includes a memory and a processor coupled to each other, the memory at least stores program instructions, and the processor is used to execute the program instructions to implement the test question tutoring method in the above first aspect.

[0008] In order to solve the above technical problems, the fourth aspect of the present application provides a computer-readable storage medium storing program instructions that can be executed by a processor, and the program instructions are used to implement the test question tutoring method of the first aspect.

[0009] The above scheme obtains the target object's personal answers to the recommended test questions and the reference answers to the recommended test questions, and then obtains the first directed acyclic graph based on the personal answers, and obtains the second directed acyclic graph based on the reference answers, and the directed acyclic graph contains a number of nodes connected in sequence, and the nodes respectively represent the various answering steps of the test questions. On this basis, based on the comparison between the first directed acyclic graph and the second directed acyclic graph, the coaching data that helps the target object to correctly answer the recommended test questions is obtained. Therefore, a directed acyclic graph is constructed through each answering step in the test question answering and compared to generate coaching data. On the one hand, it can be applied to complex test questions, and on the other hand, the coaching data can be refined to the step level, which is helpful to improve the refinement of the test question coaching compared to only providing an overall score. Therefore, the refinement of the test question coaching can be improved, especially for complex test questions. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a flowchart of an embodiment of the test question tutoring method of the present application; Figure 2a is a schematic diagram of an embodiment of constructing a question-answer pair; Figure 2b It is a process diagram of an embodiment of the test question tutoring method of the present application; Figure 3 It is a schematic diagram of the framework of an embodiment of the test question tutoring device of the present application; Figure 4 It is a schematic diagram of the framework of an embodiment of the electronic device of the present application; Figure 5 It is a schematic diagram of a framework of an embodiment of a computer-readable storage medium of the present application. DETAILED DESCRIPTION

[0011] The scheme of the embodiment of the present application is described in detail below in conjunction with the drawings of the specification.

[0012] In the following description, for the purpose of explanation rather than limitation, specific details such as specific system structures, interfaces, and technologies are provided to facilitate a thorough understanding of the present application.

[0013] The terms "system" and "network" are often used interchangeably in this article. The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the fragment " / " in this article generally indicates that the associated objects before and after are in an "or" relationship. In addition, "many" in this article means two or more than two.

[0014] See also Figure 1 , Figure 1 This is a flow chart of an embodiment of the test question tutoring method of the present application. Specifically, it may include the following steps: Step S11: Obtain the target subject's personal answers to the recommended test questions and the reference answers to the recommended test questions.

[0015] In an implementation scenario, the target object may be different depending on the application scenario. For example, in school education scenarios such as elementary schools, junior high schools, high schools, and universities, the target object may be a student; or in vocational education scenarios, the target object may be an employee. The specific category of the target object is not limited here.

[0016] In one implementation scenario, as a possible example, test questions can be recommended to the target object based on the portrait data of the target object to obtain recommended test questions for the target object. Exemplarily, if the portrait data of the target object includes "poor foundation", more basic related test questions can be recommended to the target object; or, if the portrait data of the target object includes "solid foundation", more advanced related test questions can be recommended to the target object. As another possible example, test questions can also be randomly recommended to the target object, such as randomly selecting test questions from the test question library and recommending them to the target object as recommended test questions for the target object. Of course, the above examples are only a few possible examples of recommending test questions to the target object, and other possible methods will not be given one by one here.

[0017] In one implementation scenario, the target object can input his or her personal answers to the recommended questions by text input, handwriting input, etc., and the specific method for the target object to input personal answers is not limited here. In addition, the reference answers to the recommended questions can be the answer data that comes with the recommended questions, and the reference answers to the recommended questions can be considered as the standard answers to the recommended questions. It should be noted that each recommended question can be preset with at least one reference answer. For example, in some cases, the standard answers to the recommended questions may not be unique, so the recommended questions can be preset with multiple reference answers. Of course, the above example is only a possible example in the actual application process, and the specific number of reference answers to the recommended questions is not limited here.

[0018] Step S12: Based on the personal answers, a first directed acyclic graph is obtained, and based on the reference answers, a second directed acyclic graph is obtained.

[0019] In the disclosed embodiment, a directed acyclic graph includes several nodes connected in sequence, and the several nodes respectively represent various answering steps in answering the test questions. It should be noted that a directed acyclic graph (DAG) is a special directed graph that does not contain any loops. In a DAG, each edge has a direction, and there cannot be a path that starts from a vertex and passes through several edges and then returns to the vertex. DAG is often used to describe some tasks or events with dependencies, where vertices represent tasks or events, and edges represent dependencies. Exemplarily, in the disclosed embodiment, answering a test question may include several answering steps that are executed in sequence, then each answering step can be transformed into a node in the DAG, and for any node, the answering step represented by its previous node is a pre-step of the answering step represented by the node, that is, the answering step represented by the node depends on the answering step represented by the previous node.

[0020] In one implementation scenario, as a possible example, a prompt instruction can be constructed based on the test answer, and the prompt instruction is used to instruct the large language model to construct a directed acyclic graph by decomposing the answer steps obtained by the step, and then input the prompt instruction to the large language model to obtain the output content of the large language model as a directed acyclic graph. It should be noted that the large language model may include but is not limited to open source large models such as Llama and Bloom, or the large language model can be obtained by fine-tuning the parameters of the open source large model based on a specific corpus, or the large language model can also be a custom large model, and the specific source of the large language model is not limited here. In addition, when the test answer is a personal answer, the directed acyclic graph is the first directed acyclic graph, and when the test answer is a reference answer, the directed acyclic graph is the second directed acyclic graph. The above method constructs a prompt instruction based on the test answer, and inputs the prompt instruction to the large language model to obtain the output content of the large language model as a directed acyclic graph, which can improve the construction efficiency of the directed acyclic graph by utilizing the general understanding ability of the large language model.

[0021] In a specific implementation scenario, before using the large language model to construct a directed acyclic graph, the large language model can also be pre-trained on test data (including several sample question-answer pairs). Figure 2a , Figure 2aIt is a schematic diagram of an embodiment of constructing a question-answer pair. Exemplarily, corpus data such as the syllabus, textbooks, handouts, supplementary teaching materials, and question banks in the relevant field can be obtained first. On this basis, the above-mentioned relevant data can be processed based on the general large model to process and disassemble the above-mentioned relevant data into a number of sample question-answer pairs (including sample questions and sample answers to sample questions), so that the sample questions in the sample question-answer pairs can be input into the large language model and the large language model can be instructed to answer, so as to obtain the output answer of the large language model as the predicted answer of the sample question, and then the network parameters of the large language model can be adjusted based on the difference between the sample answer and the predicted answer of the sample question, so that the general understanding ability of the general large model can be fully utilized to automatically form a corpus without artificially constructing the corpus, so that the large language model can fully learn the knowledge in the relevant field after pre-training on the question data, which is helpful for the accuracy of the subsequent use of the large language model to disassemble the answer and form a directed acyclic graph.

[0022] In a specific implementation scenario, further training can be continued on the relevant corpus after pre-training. It should be noted that the relevant corpus may include sample test questions, sample answer steps after the sample test questions are disassembled, and sample directed acyclic graphs. On this basis, the sample test questions can be input into the large language model and the large language model can be instructed to disassemble and form a directed acyclic graph to obtain the disassembled answer steps and the predicted directed acyclic graph output by the large language model, so that the network parameters of the large language model can be adjusted based on the differences between the sample answer steps after the sample test questions are disassembled and the disassembled answer steps output by the large language model, as well as the differences between the sample directed acyclic graph and the predicted directed acyclic graph, so that the large language model can learn step disassembly and form a directed acyclic graph through the above training.

[0023] In another implementation scenario, different from the aforementioned implementation mode, as another possible example, after obtaining the target subject's personal answers to the recommended test questions and the reference answers to the recommended test questions, and before obtaining the first directed acyclic graph based on the personal answers, the reference answers can also be adjusted based on the answer ideas of the personal answers, until the answer ideas of the adjusted reference answers are consistent with the answer ideas of the personal answers, and the second directed acyclic graph is obtained from the adjusted reference answers. In the above method, before forming a directed acyclic graph, the reference answers are first adjusted according to the answer ideas of the personal answers, until the answer ideas of the adjusted reference answers are consistent with the answer ideas of the personal answers, and then the second directed acyclic graph is formed according to the adjusted reference answers, which can make the subsequent comparison and analysis based on the first directed acyclic graph and the second directed acyclic graph more meaningful for reference.

[0024] In a specific implementation scenario, for example, taking proof questions as an example, the answering idea for personal answers may be to use the theorems used in each proof stage as prerequisite steps before continuing to answer, while the answering idea for reference answers may be to first list all theorems used in the proof questions as prerequisite steps and then continue the proof steps of each proof stage in sequence. In this case, the theorems placed in the reference answer can be adjusted to the corresponding proof stage as prerequisite steps, so that the answering idea for the reference answer is consistent with the answering idea for the personal answer. Of course, the above example is only a possible example of adjusting the reference answer, and does not limit other possible situations. Other possible situations will not be given examples one by one here.

[0025] In a specific implementation scenario, in order to adjust the reference answer, as a possible example, a first instruction can also be constructed based on the personal answer and the reference answer. It should be noted that the first instruction is used to instruct the first large model to adjust the reference answer to be consistent with the answer idea of ​​the personal answer. The first instruction is essentially still a prompt instruction, and the first large model is essentially still a large language model. For the sake of convenience, the aforementioned prompt instruction for indicating the step decomposition and forming a directed acyclic graph can be referred to as the second instruction, and the large language model used to process the second instruction can be referred to as the second large model. On this basis, the first instruction can be input to the first large model to obtain the output content of the first large model as the adjusted reference answer. The above method, by combining personal answers and reference answers, constructs relevant prompt instructions to adjust the reference answer through the general understanding ability of the large language model to make it consistent with the answer idea of ​​the personal answer, so that the adjustment efficiency and adjustment accuracy of the reference answer can be improved.

[0026] In an implementation scenario, when the recommended test questions are preset with multiple reference answers, you can first select the reference answer whose answer ideas are closest to the answer ideas of the personal answer, and generate a second directed acyclic graph based on the selected reference answer. Of course, after selecting the reference answer and before generating the second directed acyclic graph, you can also adjust the selected reference answer based on the answer ideas of the personal answer until the adjusted answer ideas are consistent with the answer ideas of the personal answer. The specific adjustment process can refer to the aforementioned related descriptions and will not be repeated here. It should be noted that in order to select the reference answer whose answer ideas are closest to the answer ideas of the personal answer, a prompt instruction can be constructed based on the personal answer of the recommended test question and each reference answer, and the prompt instruction is used to instruct the large language model to analyze the reference answer whose answer ideas are closest to the answer ideas of the personal answer, and then input the prompt instruction to the large language model to obtain the reference answer selected by the large language model. Of course, in addition to this, the semantic features of the individual answer and the semantic features of each reference answer can also be extracted based on language models such as BERT, and then based on the similarity between the semantic features of the individual answer and the semantic features of each reference answer, the reference answer whose answer ideas are closest to the individual answer can be selected. The above examples are just a few possible examples of selecting reference answers whose answer ideas are closest to the individual answer, and other possible ways of selecting reference answers will not be given one by one here.

[0027] In an implementation scenario, the nodes in a directed acyclic graph can be labeled not only with the answer steps corresponding to the nodes, but also with the knowledge and skills required for the answer steps represented by the nodes, and the knowledge and skills may include but are not limited to at least one of knowledge points and abilities. Exemplarily, the knowledge points required for a certain answer step may include but are not limited to the "Pythagorean Theorem", and the abilities required for a certain answer step may include but are not limited to "spatial imagination ability", etc. Other possible situations will not be given one by one here. It should be noted that, still taking the example of decomposing the steps through a large language model (such as the aforementioned second large model) and forming a directed acyclic graph, the second instruction constructed based on the test question answering can further instruct the second large model to label the nodes in the constructed directed acyclic graph with the answer steps represented and the required knowledge and skills.

[0028] Step S13: Based on the comparison between the first directed acyclic graph and the second directed acyclic graph, coaching data for helping the target object to correctly answer the recommended test questions is obtained.

[0029] In an implementation scenario, as a possible example, after obtaining the first directed acyclic graph and the second directed acyclic graph, a node-by-node comparison can be performed based on the first directed acyclic graph and the second directed acyclic graph to obtain the target steps in the individual answer, and the target steps include answer steps with flawed answers, answer steps with incorrect answers, and answer steps with missing answers. It should be noted that the answer steps with flawed answers indicate that there are no errors in essence but there are flaws (such as typos, etc.), the answer steps with missing answers indicate such as missing key steps (such as missing key steps that connect the previous and the next in the proof question), and the answer steps with incorrect answers indicate wrong answer steps such as premises and conclusions. On this basis, analysis can be performed based on the target steps to obtain the causes of the target steps, such as typos, calculation errors, wrong ideas, irregular answers, etc. After obtaining the target steps and the causes of the target steps, tutoring data that helps the target object to correctly answer the recommended test questions can be obtained accordingly. In addition, based on the annotation information of the nodes corresponding to the target steps in the directed acyclic graph (such as the aforementioned knowledge points, abilities and other knowledge and skills), the knowledge and skills to be strengthened of the target object can be obtained (for example, due to a lack of firm grasp of a certain knowledge point, the knowledge and skills to be strengthened include the knowledge point, or due to a relatively weak ability, the wrong answer is caused, so the knowledge and skills to be strengthened also include the ability). In other words, the tutoring data can further include the knowledge and skills to be strengthened of the target object.

[0030] In another implementation scenario, different from the aforementioned implementation mode, as another possible example, a third instruction can also be constructed based on the first directed acyclic graph and the second directed acyclic graph, and the third instruction is used to instruct the third largest model to output coaching data that helps the target object to correctly answer the recommended test questions by comparing the first directed acyclic graph and the second directed acyclic graph, such as including but not limited to answering steps for flawed answers, answering steps for incorrect answers, answering steps for missing answers, and other target steps and causes, as well as the knowledge and skills to be strengthened by the target object. It should be noted that, similar to the aforementioned second instruction, the third instruction is essentially a prompt instruction, and similar to the aforementioned second largest model, the third largest model is essentially a large language model. On this basis, the third instruction can be input into the third largest model to obtain the output content of the third largest model as coaching data. The above method constructs relevant prompt instructions through the first directed acyclic graph and the second directed acyclic graph, and the large language model performs comparison analysis based on the prompt instructions to output tutoring data that helps the target object to correctly answer the recommended test questions. The comparison analysis can be performed with the help of the general understanding ability of the large language model, which helps to improve the efficiency and accuracy of the comparison analysis.

[0031] In an implementation scenario, after obtaining the tutoring data, the target object's portrait data can be updated based on the tutoring data, and test questions can be recommended for the target object based on the new portrait data to obtain new recommended test questions, thereby returning to the aforementioned steps of obtaining the target object's personal answers to the recommended test questions and the reference answers to the recommended test questions to perform iterations, so as to provide test question tutoring for the target object in a long-term, targeted and stable manner.

[0032] In a specific implementation scenario, the target object's portrait data can be updated in combination with the causes of the target steps in the tutoring data and the target object's knowledge and skills to be strengthened. For example, when the causes include typos and calculation errors, the target object's portrait data can be updated with a portrait label of "careless"; or, when the knowledge and skills to be strengthened include "Pythagorean theorem", the target object's portrait data can be updated with a portrait label of "strengthen practice of Pythagorean theorem". Of course, the above examples are only a few possible examples of updating portrait data in actual applications, and other possible situations will not be given one by one here.

[0033] In a specific implementation scenario, in addition to recommending test questions for the target object based on the new portrait data, a consolidation plan suitable for the target object can also be generated for the target object based on the new portrait data. For example, if the thinking is wrong, it may be that the target object is not proficient in the application of the overall knowledge points, and needs to explain the knowledge points in detail before consolidation exercises; if a certain knowledge point is not firmly grasped, it is necessary to provide knowledge points from easy to difficult to strengthen the exercises step by step; if the answers are not standardized, it is necessary to cultivate the learning, thinking, practice and other habits of the target object in a more systematic and in-depth manner, which is a long-term tutoring task.

[0034] In one implementation scenario, see Figure 2b , Figure 2b Schematic diagram of the process of an embodiment of the test question tutoring method of the present application. Figure 2bAs shown, the general large model can be used to fine-tune in specific subject areas. The general large model can answer common questions in various fields. For example, how to keep warm when the weather is cold, how to do simple elementary school math problems, write a short essay, etc. However, in specific fields, it performs poorly on questions that require professional knowledge to answer. For example, it is unable to answer how to solve high school or even college math problems. Therefore, in order to enable the model to score in a specific subject, it is necessary to fine-tune the corpus of the subject so that the large model has the level of a professional teacher in the subject, that is, to enable the large model to answer questions, analyze the test points of the questions, comment on the answers, and even give requirements for the large model to design test questions. As a possible way of fine-tuning, a large amount of field-related syllabus, textbooks, handouts, supplementary teaching materials, question banks, etc. can be collected as inputs, so that the big model can try to understand; as another possible way of fine-tuning, question-answer pairs can be constructed artificially, so that the model can learn how to answer various professional questions from one question and one answer; as another possible way of fine-tuning, the aforementioned syllabus, textbooks, handouts, supplementary teaching materials, question banks, etc. can be disassembled into question-answer pairs, and then the constructed question-answer pairs can be used to fine-tune the general big model. On this basis, the answer logic understanding can be performed and the topological structure can be attempted to be constructed. Specifically, the test questions can be examined first, and the reference answers can be adjusted in combination with the personal answers of the target objects. Make the reference answer consistent with the personal answer ideas of the target object. This step is used to solve the problem that the reference answer is inconsistent with the personal answer ideas of the target object or there is a partial inconsistency, which leads to the problem that there is no basis for correcting the personal answer of the target object; then the reference answer can be examined and the answer can be logically split, that is, what the reference answer did in the first step and what it did in the second step. According to the logic of the split, a topological diagram is constructed. Specifically, the logical decomposition is sequential and has a sequence. In fact, some steps have no obvious order. There is a directed acyclic graph (DAG) between the steps, and each node on the DAG diagram is Label the knowledge points and abilities required for this step on the nodes; then you can review the personal answers of the target object and perform the same operations on the personal answers of the target object as on the reference answers. The difference is that the personal answers of the target object may have a series of problems such as missing steps, wrong steps, and irregular steps. It is also necessary to label each node accordingly. Finally, you can compare the topological structures and compare the differences between the two DAG graphs mentioned above, such as pointing out the similarities and differences between the two, and pointing out the flaws in the DAG graph of the personal answers of the target object relative to the DAG graph of the reference answer and the nodes where the first error leads to errors in subsequent nodes. Based on the above steps, a scoring prompt is used to fine-tune the model so that the model has the same scoring ability as a professional teacher. First, the model is prompted to immerse itself in the role of a teacher with scoring ability and senior ability.Provide test questions, reference answers corresponding to the test questions, and personal answers of the target object; then let the model adjust the reference answers according to the ideas of the target object's personal answers so that the reference answers and the target object's personal answers follow the same ideas and use the same methods to complete the test questions; on this basis, logically disassemble the reference answers, form the disassembled logic into a DAG graph and make corresponding annotations for each node in the graph; according to the same method, convert the personal answers into a DAG graph and indicate the flaws, errors, unfinished points, etc. Finally, compare the two DAG graphs to point out the problems with the target object's personal answers. Furthermore, the target object's personal answers can be scored and necessary comments can be made, such as analyzing whether the target object's personal answers at each node have flaws or errors, and further analyzing the causes of the flaws and errors (such as typos, calculation errors, errors in thinking, irregular answers, weak grasp of a certain knowledge point, etc.). Finally, personalized tutoring can be provided based on the scoring and review results. For example, the causes of mistakes can be fully utilized to intervene in the target object's consolidation exercises, so that each target object has an exercise consolidation plan that suits him or her. Based on the exercises in this stage, the target object's portrait data can be updated to prepare for the next round of personalized recommendations and tutoring.

[0035] The above scheme obtains the target object's personal answers to the recommended test questions and the reference answers to the recommended test questions, and then obtains the first directed acyclic graph based on the personal answers, and obtains the second directed acyclic graph based on the reference answers, and the directed acyclic graph contains a number of nodes connected in sequence, and the nodes respectively represent the various answering steps of the test questions. On this basis, based on the comparison between the first directed acyclic graph and the second directed acyclic graph, the coaching data that helps the target object to correctly answer the recommended test questions is obtained. Therefore, a directed acyclic graph is constructed through each answering step in the test question answering and compared to generate coaching data. On the one hand, it can be applied to complex test questions, and on the other hand, the coaching data can be refined to the step level, which is helpful to improve the refinement of the test question coaching compared to only providing an overall score. Therefore, the refinement of the test question coaching can be improved, especially for complex test questions.

[0036] See also Figure 3 , Figure 3It is a schematic diagram of the framework of an embodiment of the test question coaching device of the present application. The test question coaching device 30 includes: an acquisition module 31, a graph output module 32 and a comparison module 33. The acquisition module 31 is used to obtain the personal answers of the target object to the recommended test questions and the reference answers of the recommended test questions; the graph output module 32 is used to obtain a first directed acyclic graph based on the personal answers, and to obtain a second directed acyclic graph based on the reference answers; wherein the directed acyclic graph contains a number of nodes connected in sequence, and the nodes respectively represent each answering step in the test question answering; the comparison module 33 is used to compare the first directed acyclic graph with the second directed acyclic graph to obtain coaching data that helps the target object to correctly answer the recommended test questions.

[0037] In the above scheme, the test question coaching device 30 obtains the target object's personal answers to the recommended test questions and the reference answers to the recommended test questions, and then obtains a first directed acyclic graph based on the personal answers, and obtains a second directed acyclic graph based on the reference answers, and the directed acyclic graph contains a number of nodes connected in sequence, and the nodes respectively represent the various answering steps of the test question answering. On this basis, based on the comparison between the first directed acyclic graph and the second directed acyclic graph, coaching data that helps the target object to correctly answer the recommended test questions is obtained. Therefore, a directed acyclic graph is constructed through each answering step in the test question answering and compared to generate coaching data, which can be applied to complex test questions on the one hand, and can make the coaching data refined to the step level on the other hand, which is helpful to improve the refinement of the test question coaching compared to only providing an overall score. Therefore, the refinement of the test question coaching can be improved, especially for complex test questions.

[0038] In some public embodiments, the test question tutoring device 30 also includes an answer adjustment module, which is used to adjust the reference answer based on the answer ideas of the personal answer, until the answer ideas of the adjusted reference answer are consistent with the answer ideas of the personal answer; wherein, the second directed acyclic graph is obtained by the adjusted reference answer.

[0039] In some disclosed embodiments, the answer adjustment module includes a first construction submodule, which is used to construct a first instruction based on the personal answer and the reference answer; wherein the first instruction is used to instruct the first large model to adjust the reference answer to make it consistent with the answer idea of ​​the personal answer; the answer adjustment module includes a first acquisition submodule, which is used to input the first instruction to the first large model to obtain the output content of the first large model as the adjusted reference answer.

[0040] In some disclosed embodiments, the output module 32 includes a second construction submodule, which is used to construct a second instruction based on the test answer; wherein the second instruction is used to instruct the second large model to decompose the test answer into steps and construct a directed acyclic graph based on the decomposed answer steps; the output module 32 includes a second acquisition submodule, which is used to input the second instruction into the second large model to obtain the output content of the second large model as a directed acyclic graph.

[0041] In some disclosed embodiments, the comparison module 33 includes a third construction submodule, which is used to construct a third instruction based on the first directed acyclic graph and the second directed acyclic graph; wherein the third instruction is used to instruct the third largest model to output tutoring data that helps the target object to correctly answer the recommended test questions by comparing the first directed acyclic graph and the second directed acyclic graph; the comparison module 33 includes a third acquisition submodule, which is used to input the third instruction into the third largest model to obtain the output content of the third largest model as tutoring data.

[0042] In some disclosed embodiments, the test question coaching device 30 also includes a portrait updating module, which is used to update the portrait data of the target object based on the coaching data; the test question coaching device 30 also includes a loop iteration module, which is used to recommend test questions for the target object based on the new portrait data of the target object, obtain new recommended test questions, and return to execute the steps of obtaining the target object's personal answers to the recommended test questions and the reference answers to the recommended test questions for loop iteration.

[0043] In some public embodiments, the coaching data includes at least the target steps in an individual's answer and the reasons for the target steps, the target steps include: answer steps with flaws, answer steps with errors, and answer steps with missing answers; and / or, the nodes in the directed acyclic graph are also marked with the knowledge and skills required for the answer steps represented by the nodes, the knowledge and skills include at least one of knowledge points and abilities, and the coaching data also includes the knowledge and skills of the target object to be strengthened.

[0044] See also Figure 4 , Figure 4 It is a schematic diagram of the framework of an embodiment of an electronic device of the present application. The electronic device 40 includes at least a memory 41 and a processor 42 coupled to each other, the memory 41 stores at least program instructions, and the processor 42 is used to execute the program instructions to implement the steps in any of the above-mentioned test question tutoring method embodiments. For details, please refer to the aforementioned disclosed embodiments, which will not be repeated here. As a possible example, the electronic device 40 may include but is not limited to an office notebook, a smart phone, a tablet computer, a learning machine, etc., and the specific type of the electronic device 40 is not limited here.

[0045] Specifically, the processor 42 is used to control itself and the memory 41 to implement the steps in any of the above-mentioned intelligent dialogue method embodiments. The processor 42 can also be called a CPU (Central Processing Unit). The processor 42 may be an integrated circuit chip with signal processing capabilities. The processor 42 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 42 can be implemented by an integrated circuit chip.

[0046] In the above scheme, the electronic device 40 obtains the target object's personal answers to the recommended test questions and the reference answers to the recommended test questions, and then obtains a first directed acyclic graph based on the personal answers, and obtains a second directed acyclic graph based on the reference answers, and the directed acyclic graph contains a number of nodes connected in sequence, and the nodes respectively represent the various answering steps of the test questions. On this basis, the first directed acyclic graph is compared with the second directed acyclic graph to obtain the coaching data that helps the target object to correctly answer the recommended test questions. Therefore, a directed acyclic graph is constructed through each answering step in the test question answering and compared to generate coaching data. On the one hand, it can be applied to complex test questions, and on the other hand, the coaching data can be refined to the step level, which is helpful to improve the refinement of the test question coaching compared to only providing an overall score. Therefore, the refinement of the test question coaching can be improved, especially for complex test questions.

[0047] See also Figure 5 , Figure 5 The schematic diagram is a schematic diagram of a computer-readable storage medium 50 of the present application. The computer-readable storage medium 50 stores program instructions 51 that can be executed by a processor, and the program instructions 51 are used to implement the steps in any of the above-mentioned test question tutoring method embodiments.

[0048] In the above scheme, the computer-readable storage medium 50 obtains the target object's personal answers to the recommended test questions and the reference answers to the recommended test questions, and then obtains the first directed acyclic graph based on the personal answers, and obtains the second directed acyclic graph based on the reference answers, and the directed acyclic graph contains a number of nodes connected in sequence, and the several nodes respectively represent the various answering steps of the test questions. On this basis, the first directed acyclic graph is compared with the second directed acyclic graph to obtain the coaching data that helps the target object to correctly answer the recommended test questions. Therefore, a directed acyclic graph is constructed through each answering step in the test question answering and compared accordingly to generate coaching data. On the one hand, it can be applied to complex test questions, and on the other hand, the coaching data can be refined to the step level, which is helpful to improve the refinement of the test question coaching compared to only providing an overall score. Therefore, the refinement of the test question coaching can be improved, especially for complex test questions.

[0049] In some embodiments, the functions or modules included in the device provided by the embodiments of the present disclosure can be used to execute the method described in the above method embodiments. The specific implementation can refer to the description of the above method embodiments, and for the sake of brevity, it will not be repeated here.

[0050] The above description of various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced to each other, and for the sake of brevity, they will not be repeated herein.

[0051] 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.

[0052] 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.

[0053] 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.

[0054] 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. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute 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.

[0055] If the technical solution of this application involves personal information, the product using the technical solution of this application has clearly informed the personal information processing rules and obtained the individual's voluntary consent before processing the personal information. If the technical solution of this application involves sensitive personal information, the product using the technical solution of this application has obtained the individual's separate consent before processing the sensitive personal information, and at the same time meets the "explicit consent" requirement. For example, on personal information collection devices such as cameras, clear and prominent signs are set to inform that the personal information collection scope has been entered and personal information will be collected. If the individual voluntarily enters the collection scope, it is deemed that he or she agrees to the collection of his or her personal information; or on the device that processes personal information, the personal information processing rules are notified by obvious signs / information, and the individual's authorization is obtained through pop-up information or by asking the individual to upload his or her personal information; among them, the personal information processing rules may include information such as the personal information processor, the purpose of personal information processing, the processing method, and the type of personal information processed.

Claims

1. A test question tutoring method, characterized in that: include: Obtaining the target subject's personal answers to the recommended test questions and reference answers to the recommended test questions; Based on the personal answer, a first directed acyclic graph is obtained, and based on the reference answer, a second directed acyclic graph is obtained; wherein the directed acyclic graph includes a plurality of nodes connected in sequence, and the plurality of nodes respectively represent various answering steps in answering the test question; Based on the comparison between the first directed acyclic graph and the second directed acyclic graph, coaching data for helping the target object to correctly answer the recommended test questions is obtained.

2. The method according to claim 1, characterized in that After obtaining the target subject's personal answers to the recommended test questions and the reference answers to the recommended test questions, and before obtaining the first directed acyclic graph based on the personal answers, the method further includes: Based on the answering ideas of the personal answer, the reference answer is adjusted until the answering ideas of the reference answer after adjustment are consistent with the answering ideas of the personal answer; wherein the second directed acyclic graph is obtained from the adjusted reference answer.

3. The method according to claim 2, characterized in that The adjusting the reference answer based on the answering ideas of the personal answer includes: Based on the personal answer and the reference answer, construct a first instruction; wherein the first instruction is used to instruct the first large model to adjust the reference answer to make it consistent with the answer idea of ​​the personal answer; The first instruction is input to the first large model to obtain output content of the first large model as the adjusted reference answer.

4. The method according to claim 1, characterized in that: The steps of constructing the directed acyclic graph include: Based on the test question answer, construct a second instruction; wherein the second instruction is used to instruct the second large model to decompose the test question answer into steps and construct the directed acyclic graph based on the decomposed answer steps; Input the second instruction to the second large model to obtain output content of the second large model as the directed acyclic graph.

5. The method according to claim 1, characterized in that The obtaining of the tutoring data for helping the target subject to correctly answer the recommended test questions based on the comparison between the first directed acyclic graph and the second directed acyclic graph includes: Based on the first directed acyclic graph and the second directed acyclic graph, construct a third instruction; wherein the third instruction is used to instruct the third large model to output tutoring data that helps the target object to correctly answer the recommended test questions by comparing the first directed acyclic graph and the second directed acyclic graph; The third instruction is input to the third large model to obtain output content of the third large model as the coaching data.

6. The method according to claim 1, characterized in that After obtaining the tutoring data for helping the target object to correctly answer the recommended test questions based on the comparison between the first directed acyclic graph and the second directed acyclic graph, the method further includes: Based on the coaching data, updating the portrait data of the target object; Based on the new portrait data of the target object, test questions are recommended for the target object to obtain new recommended test questions, and the step of obtaining the target object's personal answers to the recommended test questions and the reference answers to the recommended test questions is returned to perform a loop iteration.

7. The method according to any one of claims 1 to 6, characterized in that: The coaching data at least includes the target steps in the individual's answer and the reasons for the target steps, wherein the target steps include: answer steps with defective answers, answer steps with incorrect answers, and answer steps with omitted answers; And / or, the nodes in the directed acyclic graph are also marked with the knowledge and skills required for the answering steps represented by the nodes, and the knowledge and skills include at least one of knowledge points and abilities, and the tutoring data also includes the knowledge and skills of the target object to be strengthened.

8. A test question tutoring device, characterized in that: include: An acquisition module, used to acquire the target subject's personal answers to the recommended test questions and reference answers to the recommended test questions; A graph output module, used to obtain a first directed acyclic graph based on the personal answer, and to obtain a second directed acyclic graph based on the reference answer; wherein the directed acyclic graph includes a plurality of nodes connected in sequence, and the plurality of nodes respectively represent various answering steps in answering the test question; A comparison module is used to compare the first directed acyclic graph with the second directed acyclic graph to obtain coaching data that helps the target object to correctly answer the recommended test questions.

9. An electronic device, characterized in that: The system at least comprises a memory and a processor coupled to each other, wherein the memory at least stores program instructions, and the processor is used to execute the program instructions to implement the test question tutoring method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Program instructions that can be executed by a processor are stored, and the program instructions are used to implement the test question tutoring method described in any one of claims 1 to 7.

Images