A knowledge graph construction method, device, equipment and medium

Abstract

Embodiments of the present disclosure relate to a knowledge graph construction method, device and equipment and a medium, wherein the method comprises: obtaining an initial knowledge graph; wherein the initial knowledge graph is used to describe a plurality of knowledge points with dependency relationships; obtaining answer data of a plurality of questions by different user objects; wherein the questions are pre-labeled, and the labels include standard answers of the questions and knowledge points of the questions; determining a mastery level of a corresponding knowledge point by at least one user object according to the answer data and the standard answers of the questions corresponding to the knowledge point; and modifying the dependency relationships of the knowledge points in the initial knowledge graph according to the mastery level of the corresponding knowledge point by the user object, to obtain a target knowledge graph. The target knowledge graph determined by the present disclosure can more objectively and accurately display the dependency relationships between the knowledge points, and provide more accurate and effective learning guidance for users.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent teaching technology, and in particular to a method, apparatus, device and medium for constructing a knowledge graph. Background Technology

[0002] Adaptive learning is gradually becoming a hot technology in the online education field. Among them, knowledge graphs with difficulty level annotations for knowledge points are the core competitiveness of adaptive learning products. At present, knowledge graphs with difficulty level annotations for knowledge points are mainly generated manually. The generated knowledge graphs are highly dependent on the knowledge system of the teaching and research personnel and their understanding of the teaching materials. However, knowledge systems are often massive and complex, making manual generation of knowledge graphs not only extremely costly, but also resulting in significant differences and strong subjectivity among knowledge graphs generated by different teaching and research personnel. This makes it impossible to objectively and accurately provide users with guidance for learning behaviors such as knowledge diagnosis, learning, and practice. Summary of the Invention

[0003] To solve the above-mentioned technical problems, or at least partially solve them, this disclosure provides a method, apparatus, device, and medium for constructing a knowledge graph.

[0004] This disclosure provides a method for constructing a knowledge graph, including:

[0005] Obtain an initial knowledge graph; wherein the initial knowledge graph is used to describe multiple knowledge points with dependencies; obtain answer data from different user objects for multiple questions; wherein the questions are pre-labeled, and the labels include: the standard answer to the question and the knowledge point corresponding to the question; determine at least one user object's mastery level of the corresponding knowledge point based on the answer data of the question corresponding to the knowledge point and the standard answer; modify the dependencies of knowledge points in the initial knowledge graph based on the user object's mastery level of the corresponding knowledge point to obtain a target knowledge graph.

[0006] Optionally, determining the mastery level of at least one user object for the corresponding knowledge point based on the answer data of the question corresponding to the knowledge point and the standard answer includes:

[0007] Based on the knowledge points provided in the questions, the questions are divided into different combinations; wherein, the questions in the same combination have the same knowledge points; the answer data corresponding to each question in each combination is compared with the standard answer, and the mastery level of at least one user object for the corresponding knowledge point is determined based on the comparison results.

[0008] Optionally, the division of the problem into different combinations includes:

[0009] Each user object's question is divided into multiple thematic units with different knowledge points; all user objects' questions are divided into multiple unit sets with different knowledge points; wherein, the same unit set includes the same thematic unit of different user objects.

[0010] Optionally, for the current topic unit, comparing the answer data corresponding to each question in each of the combinations with the standard answer includes:

[0011] The answer data corresponding to each question in the current topic unit is compared with the standard answer to obtain the score value of each question; a first comparison result is obtained based on the first score ratio between the score value corresponding to each question and the preset full score value; wherein, the first comparison result is: the average value of the first score ratio of each question in the current topic unit.

[0012] Optionally, the step of modifying the dependencies of knowledge points in the initial knowledge graph according to the user's mastery level of the corresponding knowledge points to obtain the target knowledge graph includes:

[0013] Using the initial knowledge graph as a temporary knowledge graph, the following modification operations are repeatedly performed until the modification operations meet the preset iteration stopping rule:

[0014] The deviation of the temporary knowledge graph is determined based on preset node dependency conditions and the user's mastery level of the corresponding knowledge points. The node dependency conditions include relational conditions and influence conditions. The relational conditions indicate that the mastery level of the predecessor knowledge point is not less than the mastery level of the successor knowledge point. The influence conditions indicate the transfer effect of the mastery level of the predecessor knowledge point on the mastery level of the successor knowledge point, which is within a preset influence range. The predecessor knowledge point and the successor knowledge point are two adjacent knowledge points along the learning path of the temporary knowledge graph.

[0015] Determine whether the deviation reaches a preset deviation threshold; if it does, modify the dependency relationship of the knowledge points in the temporary knowledge graph according to the node dependency condition and the user object's mastery level of the corresponding knowledge point to obtain a new temporary knowledge graph; if it does not reach the threshold, determine that the modification operation satisfies the iteration stopping rule, and determine the temporary knowledge graph corresponding to the time when the modification operation stops as the target knowledge graph.

[0016] Optionally, determining the deviation of the temporary knowledge graph based on preset node dependency conditions and the user object's mastery level of the corresponding knowledge points includes:

[0017] Starting from any of the user objects, the traversal operation is performed iteratively until all user objects have completed the traversal operation, at which point the iteration terminates; wherein, the traversal operation is: traversing the temporary knowledge graph, checking the user object's mastery level of the knowledge points in the temporary knowledge graph, and obtaining traversal data;

[0018] From all the traversal data obtained at the end of the iteration, target deviation data that does not satisfy the node dependency condition is determined; wherein, the target deviation data includes first deviation data that does not satisfy the relationship condition and / or second deviation data that does not satisfy the influence condition; the deviation degree of the temporary knowledge graph is determined based on the target deviation data.

[0019] Optionally, determining the target deviation data that does not satisfy the node dependency condition from all the traversed data obtained at the time of termination of iteration includes:

[0020] For each set of traversed data, each knowledge point in the traversed data is taken as a predecessor knowledge point, and the dependency relationship between the predecessor knowledge point and the successor knowledge point is checked to see if it satisfies the relationship condition. Based on the detection result, the first knowledge point in each set of traversed data that does not satisfy the relationship condition is determined. Data noise reduction is performed on the first knowledge point and the mastery level corresponding to the first knowledge point in all the traversed data to determine the first deviation data.

[0021] Optionally, determining the target deviation data that does not satisfy the node dependency condition from all the traversed data obtained at the time of termination of iteration includes:

[0022] For each set of traversed data, each knowledge point in the traversed data is taken as a predecessor knowledge point. Based on the user's mastery level of the predecessor knowledge point and their mastery level of the subsequent knowledge point, a migration influence parameter is obtained. The migration influence parameter represents the degree of change in the mastery level of the subsequent knowledge point compared to the predecessor knowledge point. A first proportion of migration influence parameters within a preset range and a second proportion of migration influence parameters outside the preset range are determined in all the traversed data. When the first proportion is less than the second proportion, the predecessor knowledge point and the subsequent knowledge point are determined as second knowledge points that do not meet the influence condition. Data noise reduction is performed on the second knowledge point and its corresponding mastery level to determine the second deviation data.

[0023] Optionally, for the current set of units, comparing the answer data corresponding to each question in each combination with the standard answer includes:

[0024] When the questions in the current unit set include objective questions, the answer data corresponding to each objective question in the current unit set is compared with the standard answer to obtain a second comparison result;

[0025] And / or,

[0026] When the current unit set includes subjective questions, the subjective questions in the current unit set are divided into different question sets; wherein, the subjective questions included in the same question set are identical; for any question set, the answer data of each user object for the subjective question is compared with the standard answer to obtain the average score of the subjective question; based on the second score ratio between the average score and the preset full score, a third comparison result is obtained; wherein, the third comparison result is the average of the second score ratios of each question set in the current unit set.

[0027] Optionally, the method further includes:

[0028] Based on the second comparison result and / or the third comparison result, the initial difficulty of each knowledge point in the target knowledge graph is determined; the initial difficulty of each knowledge point is normalized to obtain the target difficulty; and the target difficulty is marked at the corresponding knowledge point in the target knowledge graph.

[0029] This disclosure also provides an apparatus for constructing a knowledge graph, including:

[0030] The knowledge graph acquisition module is used to acquire an initial knowledge graph; wherein, the initial knowledge graph is used to describe multiple knowledge points with dependencies.

[0031] The answer data acquisition module is used to acquire answer data from different users for multiple questions; wherein, the questions are pre-labeled, and the labels include: the standard answer to the question and the knowledge point corresponding to the question;

[0032] The mastery level determination module is used to determine the mastery level of at least one user object for the corresponding knowledge point based on the answer data of the question corresponding to the knowledge point and the standard answer;

[0033] The knowledge graph modification module is used to modify the dependencies of knowledge points in the initial knowledge graph according to the user's mastery level of the corresponding knowledge points, so as to obtain the target knowledge graph.

[0034] This disclosure also provides an electronic device, the electronic device comprising: a processor; a memory for storing executable instructions of the processor; the processor being configured to read the executable instructions from the memory and execute the instructions to implement the knowledge graph construction method provided in this disclosure.

[0035] This disclosure also provides a computer-readable storage medium storing a computer program that, when executed by a computing device, enables the computer device to implement the knowledge graph construction method provided in this disclosure.

[0036] The technical solution provided in this disclosure has the following advantages compared with the prior art:

[0037] This disclosure provides a method, apparatus, device, and medium for constructing a knowledge graph. The method includes: first, acquiring an initial knowledge graph and answer data for questions from different user objects. The questions are pre-labeled with the following tags: the standard answer to the question and the corresponding knowledge point. Then, based on the answer data and standard answer for each knowledge point, determining the mastery level of at least one user object for that knowledge point. Finally, based on the user object's mastery level, modifying the dependencies between knowledge points in the initial knowledge graph to obtain a target knowledge graph. This solution automatically modifies the initial knowledge graph based on the user object's mastery level of the corresponding knowledge point, reducing labor costs and increasing the construction speed of the target knowledge graph. Because the user object's mastery level of the corresponding knowledge point has high accuracy, the target knowledge graph modified based on this mastery level can objectively and accurately display the dependencies between knowledge points according to their mastery level, providing users with more accurate and effective learning guidance. Attached Figure Description

[0038] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0039] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0040] Figure 1 A flowchart illustrating a method for constructing a knowledge graph according to an embodiment of this disclosure;

[0041] Figure 2 A schematic diagram illustrating the division of a problem into combinations, provided as an embodiment of this disclosure;

[0042] Figure 3 A flowchart illustrating a method for modifying a knowledge graph provided in this embodiment of the disclosure;

[0043] Figure 4A schematic diagram of the structure of a knowledge graph construction apparatus provided in an embodiment of this disclosure;

[0044] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0045] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0046] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0047] Considering the high cost, strong subjectivity, and low accuracy of manually generated knowledge graphs, this disclosure provides a method, apparatus, device, and medium for constructing knowledge graphs to address these issues. This technology can be applied to intelligent teaching scenarios. For ease of understanding, the embodiments of this invention are described in detail below.

[0048] Figure 1 This is a flowchart illustrating a method for constructing a knowledge graph according to an embodiment of this disclosure. This method can be executed by a knowledge graph construction device, which can be implemented in software and / or hardware, and is generally integrated into an electronic device. Figure 1 As shown, the method includes:

[0049] Step S102: Obtain an initial knowledge graph; wherein, the initial knowledge graph is used to describe multiple knowledge points with dependencies. The nodes of the initial knowledge graph are knowledge points, and the edges between knowledge points are directed edges, representing the dependencies between knowledge points; such dependencies include, for example, a reference relationship between one knowledge point and another, the transfer effect of a user's mastery level of one knowledge point on the mastery level of another knowledge point, and a user's mastery level of an old knowledge point being no less than their mastery level of a new knowledge point. In this embodiment, the initial knowledge graph does not require the accuracy and objectivity of the dependencies between knowledge points and can be a knowledge graph constructed manually.

[0050] Step S104: Obtain answer data for multiple questions from different user objects; wherein, the questions are pre-labeled, and the labels include: the standard answer of the question and the knowledge point corresponding to the question.

[0051] In this embodiment, a question database can be established in advance by collecting questions and tagging them. The tags for questions may include, but are not limited to: the standard answer to the question and the corresponding knowledge point; the knowledge point includes the core knowledge point tested by the question and other related knowledge points, such as the knowledge point corresponding to the incorrect answer to the question, and related knowledge points similar to the core knowledge point. Then, questions are selected from the question database and sent to multiple user objects, and a large amount of user answer data is collected; the answer data includes common options, correct / incorrect judgments, and the text content entered.

[0052] Step S106: Based on the answer data and standard answers to the questions corresponding to the knowledge points, determine the mastery level of at least one user object for the corresponding knowledge point.

[0053] In practice, the same knowledge point may correspond to multiple different questions, and the same question may contain multiple knowledge points, meaning a question may have tags for multiple knowledge points. Based on this, this embodiment can divide questions into different combinations according to the knowledge points they contain; among them, questions in the same combination have the same knowledge points; the answer data corresponding to each question in each combination is compared with the standard answer, and the mastery level of at least one user object for the corresponding knowledge point is determined based on the comparison results.

[0054] Step S108: Based on the user's mastery level of the corresponding knowledge points, modify the dependency relationships of the knowledge points in the initial knowledge graph to obtain the target knowledge graph.

[0055] In one embodiment, the initial knowledge graph can be traversed, and based on the user's mastery level of the knowledge points, it can be determined whether the knowledge points along the learning path of the initial knowledge graph meet preset node dependency conditions. The node dependency conditions include: relationship conditions and influence conditions. The relationship condition indicates that the mastery level of the predecessor knowledge point is not less than the mastery level of the successor knowledge point; this can be understood as the user's mastery level of the knowledge point gradually decreasing along the learning path of the temporary knowledge graph. The influence condition indicates that the transfer effect of the mastery level of the predecessor knowledge point on the mastery level of the successor knowledge point is within a preset influence range. The aforementioned predecessor and successor knowledge points are two adjacent knowledge points along the learning path of the temporary knowledge graph. For dependency relationships on the initial knowledge graph that do not meet the node dependency conditions, this embodiment can modify the dependency relationship according to the user's mastery level of the corresponding knowledge point.

[0056] The knowledge graph construction method provided in this disclosure determines the user's mastery level of corresponding knowledge points based on user answer data and standard answers to questions, and then modifies the dependencies between knowledge points in the initial knowledge graph according to the mastery level. This solution automatically modifies the initial knowledge graph based on the user's mastery level of corresponding knowledge points, reducing labor costs and improving the construction speed of the target knowledge graph. Because the user's mastery level of corresponding knowledge points has high authenticity, the target knowledge graph modified based on this mastery level can more objectively and accurately display the dependencies between knowledge points according to their mastery level, providing users with more accurate and effective learning guidance.

[0057] like Figure 2 As shown, in an embodiment for determining the user's mastery level of the corresponding knowledge points, the problem can first be divided into different combinations of knowledge points, referring to the following implementation method.

[0058] In implementation method one, based on the dimension of user objects, the question corresponding to each user object is divided into multiple thematic units with different knowledge points. Specifically, for any user object among multiple user objects, such as... Figure 2 Taking user object 1 as an example, based on the knowledge points set in the questions, the questions corresponding to user object 1 are divided into thematic units corresponding to knowledge points 1, 2, and 3, respectively. Multiple questions in the same thematic unit not only have the same knowledge points, but also correspond to the same user object.

[0059] In implementation method two, based on the dimension of the question (i.e., regardless of the user object to which the question belongs), all questions corresponding to all user objects are divided into multiple unit sets with different knowledge points; where each unit set includes the same topic unit from multiple different user objects. In specific implementation, all questions can be collected first, and then, based on the knowledge points defined in the questions, the large number of collected questions can be divided into multiple unit sets with different knowledge points. Alternatively, refer to... Figure 2 First, following implementation method one, obtain multiple thematic units with different knowledge points corresponding to each user object. Then, divide the thematic units with the same knowledge points into unit sets. The questions in the same unit set may correspond to different user objects, but they all have the same knowledge points.

[0060] This embodiment provides a process for comparing the answer data and standard answers for each question in each unit, specifically for the current topic unit. The process includes:

[0061] The answer data and standard answers for each question in the current topic unit are compared to obtain the score for each question; the first comparison result is obtained based on the first score ratio between the score for each question and the preset full score; wherein, the first comparison result is the average of the first score ratios for each question in the current topic unit.

[0062] In practical applications, questions are generally categorized into objective questions and subjective questions. Based on this, and assuming that the questions in the combination are objective questions and / or subjective questions, the above-mentioned implementation of comparing answer data with standard answers will be described in detail.

[0063] When the current topic unit includes objective questions, the answer data for each objective question in the current topic unit is compared with the standard answer. If the answer data matches the standard answer (i.e., the answer is correct), the score is the preset full score for that question; if the answer data does not match the standard answer (i.e., the answer is incorrect), the score is zero. Then, the first score ratio between the score for each objective question in the current topic unit and the preset full score is calculated, and the average of these first score ratios is determined as the first comparison result. For ease of distinction, the first comparison result for objective questions can be represented as P. N .

[0064] In addition, for objective questions, another comparison method can be provided, namely, referring to the following formula (1): compare the answer data and standard answer of each objective question in the current topic unit, and determine the first comparison result P by the ratio between the number of objective questions n whose answer data and standard answer match and the total number N of all objective questions in the current topic unit. N .

[0065]

[0066] Given that all questions in the current thematic unit are objective questions, the first comparison result P can be used. N Determine the user's level of mastery of the corresponding knowledge points.

[0067] When the current topic unit includes subjective questions, the answer data for each subjective question in the current topic unit is compared with the standard answer to obtain the score for each subjective question. Then, the first score ratio between the score for each subjective question and the preset full score is calculated, and the average of the first score ratios is determined as the first comparison result (the first comparison result for subjective questions can be represented as P). M ).

[0068] Different subjective questions typically have different preset full marks. In this case, to ensure the accuracy of the first comparison result P... MThe accuracy and objectivity of the result can be referenced to the following formula (2):

[0069]

[0070] For any subjective question m, calculate the first score ratio between its score x and the preset full score X, and then average the first score ratios P ​​of the M subjective questions. M This serves as the first comparison result.

[0071] Given that all questions in the current thematic unit are subjective, the first comparison result P can be used. M Determine the user's level of mastery of the corresponding knowledge points.

[0072] Furthermore, when the questions in the current thematic unit include both objective and subjective questions, the results of the two first comparisons mentioned above can be used as a basis for further comparison. N and P M Determine the user's level of mastery of the corresponding knowledge points; for example, calculate the first comparison result P. N and P M The sum or weighted average of the results is used as the user's level of mastery of the corresponding knowledge point.

[0073] After obtaining the user's mastery level of the corresponding knowledge point according to the above embodiments, this embodiment provides a method for modifying the dependency relationship of knowledge points in the initial knowledge graph based on the mastery level, referring to... Figure 3 As shown, the modification method includes:

[0074] Using the initial knowledge graph as a temporary knowledge graph, the modification operations shown in steps S302 to S306 are repeated until the modification operations meet the preset iteration stopping rules:

[0075] Step S302: Determine the deviation of the temporary knowledge graph based on the preset node dependency conditions and the user object's mastery level of the corresponding knowledge points.

[0076] The method for determining the deviation of the temporary knowledge graph in this embodiment is as follows:

[0077] (I) Starting from any user object, iteratively execute the traversal operation until all user objects have completed the traversal operation, at which point the iteration terminates. The traversal operation involves: traversing the temporary knowledge graph, checking the current user object's mastery level of the knowledge points in the temporary knowledge graph, and obtaining traversal data. Specifically, based on the user object's mastery level of the corresponding knowledge point, along the learning path of the temporary knowledge graph, sequentially check the mastery level corresponding to each knowledge point in the temporary knowledge graph to obtain traversal data. That is, the traversal data includes: each knowledge point along the learning path of the temporary knowledge graph, and the user object's mastery level of each knowledge point.

[0078] (II) When all user objects have completed the traversal operation, the iteration is terminated. From all the traversal data obtained when the iteration is terminated, the target deviation data that does not meet the node dependency condition is determined.

[0079] In this embodiment, the node dependency conditions include relationship conditions and influence conditions. Accordingly, the target deviation data may include: first deviation data that does not meet the relationship conditions and second deviation data that does not meet the influence conditions. This embodiment describes the methods for determining the first deviation data and the second deviation data respectively.

[0080] An embodiment for determining the first deviation data is provided herein.

[0081] For each piece of traversed data, each knowledge point in the traversed data is taken as a predecessor knowledge point, and the dependency relationship between the predecessor knowledge point and the successor knowledge point is checked to see if the relationship condition is met. Based on the detection results, the first knowledge point in each piece of traversed data that does not meet the relationship condition is determined.

[0082] Specifically, the system checks whether the user's mastery level of the predecessor knowledge point is not less than their mastery level of the successor knowledge point. If it is not less, the dependency relationship between the predecessor and successor knowledge points satisfies the relation condition; otherwise, if it is less, the dependency relationship between the predecessor and successor knowledge points does not satisfy the relation condition. Based on the above detection results, the first knowledge point in each piece of traversed data that does not satisfy the relation condition is determined.

[0083] Then, data denoising is performed on the first knowledge point and its corresponding mastery level in all traversed data to determine the first deviation data. For ease of description, this embodiment refers to the first knowledge point and its corresponding mastery level in all traversed data as the full data. In specific implementation, data denoising is performed on the full data to filter out random errors or deviations, and the denoised data is determined as the first deviation data. For example, if the proportion of the first knowledge point 1 and the first knowledge point 2 and their corresponding mastery levels in the full data is higher than a preset proportion (e.g., 70%), this means that in more than 70% of the traversed data of user objects, there are cases where the dependency relationship between the first knowledge point 1 and the first knowledge point 2 does not meet the relationship condition. In contrast, the proportion of the first knowledge point 3 and the first knowledge point 4 and their corresponding mastery levels in the full data is lower than a preset proportion (e.g., only 5%). The above two cases indicate that the first knowledge point 1 and the second knowledge point 2 that do not satisfy the conditional relationship are universal and can truly reflect the dependency relationship between the first knowledge point 1 and the second knowledge point 2; correspondingly, the first knowledge point 3 and the second knowledge point 4 that do not satisfy the conditional relationship are random noise data, so the noise data is filtered out to obtain the first deviation data.

[0084] An embodiment for determining the second deviation data is provided herein.

[0085] For each set of traversed data, each knowledge point in the data is treated as a predecessor knowledge point. A transfer impact parameter is obtained based on the user's mastery level of the predecessor knowledge point and their mastery level of the subsequent knowledge point. This transfer impact parameter represents the degree of change in the user's mastery level of the subsequent knowledge point compared to its predecessor. For example, in a set of traversed data, if the user's mastery level of the predecessor knowledge point is 1 and their mastery level of the subsequent knowledge point is 0.95, then the 5% change in mastery level between the two knowledge points is used as the transfer impact parameter.

[0086] Determine the first proportion of migration influence parameters within a preset range and the second proportion of migration influence parameters outside the preset range in all traversed data.

[0087] The preset range can be the initial range of migration impact parameters set manually during the first modification of the temporary knowledge graph, or the range of migration impact parameters determined after the previous (i.e., the n-1th) modification of the knowledge graph during the nth modification of the temporary knowledge graph.

[0088] In this embodiment, the distribution of migration influence parameters between the same predecessor knowledge point and its corresponding successor knowledge point in all traversed data can be statistically analyzed first. For example, assuming that second knowledge point 1 is a predecessor knowledge point and second knowledge point 2 is its corresponding successor knowledge point, the proportion P1 of migration influence parameters distributed within the preset range S1 in all traversed data is 20%, and the distribution of migration influence parameters not within the preset range S1 is as follows: distributed within the range... The second proportion of the migration effect parameters within the range is Distributed in range The second proportion of the migration effect parameters within the range is

[0089] When the first proportion is less than the second proportion, the predecessor and successor knowledge points are identified as second knowledge points that do not meet the influence conditions. Specifically, the first proportion is less than at least one of the second proportions. In this case, across all traversed data, the proportion of cases where the migration influence parameter between the predecessor and successor knowledge points does not conform to the preset range is larger, indicating a significant data deviation. Therefore, the predecessor and successor knowledge points are identified as second knowledge points that do not meet the influence conditions. Conversely, if the first proportion is greater than every second proportion, it indicates that the current preset range of the migration influence parameter is reasonable and accurate, and no modification to the preset range in the temporary knowledge graph is necessary.

[0090] Considering the potential data noise in the second knowledge point, we can perform data noise reduction on the second knowledge point and the corresponding mastery level to determine the second deviation data.

[0091] In practical implementation, the second knowledge point and its mastery level corresponding to the largest second proportion can be considered as valid data, while the second knowledge points and their mastery levels corresponding to other second proportions can be considered as noise data. For example, in the aforementioned example, the second knowledge point with the largest proportion is... This indicates that in 70% of the user object traversal data, the migration impact parameters of Knowledge Point 1 and Knowledge Point 2 are distributed within the range Within this scope, the data is universal and can reflect the true situation of the transfer effect between the second knowledge point 1 and the second knowledge point 2; while for other second proportions This indicates that only in the traversal data of 10% of user objects, the migration impact parameters of both Knowledge Point 1 and Knowledge Point 2 are distributed within the range. Within this range, the data is noise data that appears by chance, so the noise data is filtered out to obtain the second deviation data.

[0092] The first deviation data and / or the second deviation data are determined as the target deviation data. Specifically, when only the first deviation data that does not meet the relationship conditions can be obtained according to the above embodiments, the first deviation data is determined as the target deviation data; when only the second deviation data that does not meet the influence conditions can be obtained according to the above embodiments, the second deviation data is determined as the target deviation data; when both the first deviation data and the second deviation data are obtained according to the above embodiments, the first deviation data and the second deviation data are jointly determined as the target deviation data.

[0093] (III) Determine the deviation degree of the temporary knowledge graph based on the target deviation data.

[0094] In one implementation, the mean squared error loss between the target deviation data and the temporary knowledge graph can be calculated, and the obtained mean squared error loss can be determined as the deviation of the temporary knowledge graph.

[0095] Having determined the deviation of the temporary knowledge graph through the above steps (I)-(III), proceed to step S304.

[0096] Step S304: Determine whether the deviation reaches a preset deviation threshold. To improve the accuracy of the final knowledge graph, this deviation threshold is usually small, for example, 3%. If it is reached, it means that the temporary knowledge graph does not meet the node dependency conditions and has a significant deviation from the user's actual level of mastery of the knowledge points. In this case, proceed to step S306, which requires modifying the temporary knowledge graph. Conversely, if it is not reached, the deviation of the temporary knowledge graph is considered to be statistical noise, and the modification operation can be stopped. In this case, proceed to step S308.

[0097] Step S306: Modify the dependency relationships of knowledge points in the temporary knowledge graph according to the node dependency conditions and the user object's mastery level of the corresponding knowledge points to obtain a new temporary knowledge graph.

[0098] To reduce the workload and improve the efficiency of modification, this embodiment can directly modify the dependency relationship between the first and second knowledge points in the temporary knowledge graph according to the user's mastery level of the corresponding knowledge points, so that the modified dependency relationship satisfies the node dependency condition; thereby obtaining a new temporary knowledge graph.

[0099] In practical implementation, the order of knowledge points on the temporary knowledge graph can be modified based on the mastery level relationship between knowledge points represented by the first deviation data, ensuring that the mastery level of predecessor knowledge points along the learning path of the temporary knowledge graph is not lower than the mastery level of successor knowledge points. Furthermore, the preset range originally labeled on the temporary knowledge graph can be modified based on the migration influence parameters of predecessor and successor knowledge points represented by the second deviation data, ensuring that the new preset range represents the range of the migration influence parameters.

[0100] After obtaining the new temporary knowledge graph, the modification operation is re-executed using the new temporary knowledge graph.

[0101] Step S308: Determine that the modification operation satisfies the iteration stopping rule, and determine the temporary knowledge graph corresponding to the time when the modification operation stops as the target knowledge graph.

[0102] If the deviation of the temporary knowledge graph does not reach the deviation threshold, the modification operation is stopped, and the temporary knowledge graph is determined as the final target knowledge graph that meets the actual use.

[0103] In the above embodiments, the user's mastery level of the corresponding knowledge points has high accuracy, authenticity and objectivity. The temporary knowledge graph is continuously modified using this mastery level so that the final target knowledge graph satisfies the node dependency conditions. It can objectively and accurately display the dependency relationship between knowledge points, thereby providing users with more accurate and effective learning guidance.

[0104] Based on the target knowledge graph obtained in the above embodiments, this embodiment can further provide a method for marking the difficulty of knowledge points in the target knowledge graph. This method first uses the comparison results between the answer data corresponding to the questions in the unit set and the standard answers to calculate the initial difficulty of each knowledge point in the target knowledge graph. Then, it normalizes the initial difficulty and marks the normalized target difficulty at the corresponding knowledge point in the target knowledge graph.

[0105] This embodiment provides a method for calculating the initial difficulty of a knowledge point, as shown below.

[0106] For the current set of units, the answer data corresponding to the question can be compared with the standard answer according to the following example:

[0107] When the current unit set includes objective questions, the answer data corresponding to each objective question in the current unit set is compared with the standard answer to obtain a second comparison result. The specific implementation process can refer to the method of comparing the answer data corresponding to each objective question in the topic unit with the standard answer in the aforementioned embodiment, which will not be described in detail here.

[0108] Since each unit set uniquely corresponds to a knowledge point, the second comparison result for the current unit set is also the second comparison result for the corresponding knowledge point. If all questions in the current unit set are objective questions, the second comparison result can be used to determine the initial difficulty level of the corresponding knowledge point in the target knowledge graph.

[0109] When the current unit set includes subjective questions, the subjective questions in the current unit set are divided into different question sets; within the same question set, the subjective questions are identical. For any given question set, each user's answer data for the subjective questions is compared with the standard answers to obtain the average score for the subjective questions.

[0110] Referring to the following formula (4), based on the average score... The second score ratio between the second score and the preset maximum score G The third comparison result was obtained:

[0111]

[0112] Among them, the third comparison result P I Let be the average of the second score ratios of all question sets in the current unit set; i represents the i-th question set, and I represents the number of question sets in the current unit set.

[0113] Similar to the second comparison result, the third comparison result for the current unit set is also the third comparison result for the corresponding knowledge point. If all questions in the current unit set are subjective questions, the third comparison result can be used to determine the initial difficulty level of the corresponding knowledge point in the target knowledge graph.

[0114] Furthermore, when the questions in the current unit set include both objective and subjective questions, the initial difficulty level of each knowledge point in the target knowledge graph can be determined based on the second and third comparison results.

[0115] Considering that different users may provide different answers for each knowledge point, it is impossible to accurately and systematically compare knowledge points horizontally based on answer data for each individual knowledge point. In this case, this embodiment can normalize the initial difficulty level of each knowledge point to obtain the target difficulty level. For example, this embodiment can use a percentage norm to normalize the initial difficulty level.

[0116] Finally, the difficulty level of the target is marked at the corresponding knowledge point in the target knowledge graph, resulting in a target knowledge graph containing the difficulty level markings of the knowledge points.

[0117] In the knowledge point difficulty labeling method provided in the above embodiments, the second and / or third comparison results used to determine the initial difficulty are obtained based on the user's actual answer data, which matches the difficulty level when the user learns the corresponding knowledge point. Then, by normalizing the initial difficulty, the resulting target difficulty accurately reflects the actual situation. Therefore, this technical solution can improve the accuracy of knowledge point difficulty labeling, thereby improving the accuracy of the target knowledge graph containing knowledge point difficulty labeling.

[0118] Figure 4 This is a schematic diagram of a knowledge graph construction apparatus provided in an embodiment of the present disclosure. The apparatus can be implemented by software and / or hardware, and is generally integrated into an electronic device. It can obtain a target knowledge graph by executing a knowledge graph construction method. Figure 4 As shown, the device includes:

[0119] The graph acquisition module 402 is used to acquire an initial knowledge graph; wherein, the initial knowledge graph is used to describe multiple knowledge points with dependencies.

[0120] The answer data acquisition module 404 is used to acquire answer data for multiple questions from different user objects; wherein, the questions are pre-labeled, and the labels include: the standard answer to the question and the knowledge point corresponding to the question;

[0121] The mastery level determination module 406 is used to determine the mastery level of at least one user object for the corresponding knowledge point based on the answer data and standard answers to the questions corresponding to the knowledge point.

[0122] The knowledge graph modification module 408 is used to modify the dependencies of knowledge points in the initial knowledge graph according to the user's mastery level of the corresponding knowledge points, so as to obtain the target knowledge graph.

[0123] In one embodiment, the mastery level determination module 406 is specifically used for:

[0124] Based on the knowledge points provided in the questions, the questions are divided into different combinations; wherein, the questions in the same combination have the same knowledge points; the answer data corresponding to each question in each combination is compared with the standard answer, and the mastery level of at least one user object for the corresponding knowledge point is determined based on the comparison results.

[0125] In one embodiment, the mastery level determination module 406 is specifically used for:

[0126] The questions corresponding to each user object are divided into multiple thematic units with different knowledge points;

[0127] The questions corresponding to all the user objects are divided into multiple unit sets with different knowledge points; wherein, the same unit set includes the same topic unit of different user objects.

[0128] In one embodiment, for the current topic unit, the mastery level determination module 406 is specifically used for:

[0129] The answer data corresponding to each question in the current topic unit is compared with the standard answer to obtain the score value of each question; a first comparison result is obtained based on the first score ratio between the score value corresponding to each question and the preset full score value; wherein, the first comparison result is: the average value of the first score ratio of each question in the current topic unit.

[0130] In one embodiment, the map modification module 408 is specifically used for:

[0131] Using the initial knowledge graph as a temporary knowledge graph, the following modification operations are repeatedly performed until the modification operations meet the preset iteration stopping rule:

[0132] The deviation of the temporary knowledge graph is determined based on preset node dependency conditions and the user's mastery level of the corresponding knowledge points. The node dependency conditions include relationship conditions and influence conditions. The relationship condition indicates that the mastery level of the predecessor knowledge point is not less than the mastery level of the successor knowledge point. The influence condition indicates that the transfer effect of the mastery level of the predecessor knowledge point on the mastery level of the successor knowledge point is within a preset range. The predecessor knowledge point and the successor knowledge point are two adjacent knowledge points along the learning path of the temporary knowledge graph.

[0133] Determine whether the deviation reaches a preset deviation threshold;

[0134] If the condition is met, the dependency relationships of the knowledge points in the temporary knowledge graph are modified according to the node dependency conditions and the user object's mastery level of the corresponding knowledge points, to obtain a new temporary knowledge graph.

[0135] If the condition is not met, the modification operation is determined to satisfy the iteration stopping rule, and the temporary knowledge graph corresponding to the time when the modification operation stops is determined as the target knowledge graph.

[0136] In one embodiment, the map modification module 408 is specifically used for:

[0137] Starting from any of the user objects, the traversal operation is performed iteratively until all user objects have completed the traversal operation, at which point the iteration terminates; wherein, the traversal operation is: traversing the temporary knowledge graph, checking the user object's mastery level of the knowledge points in the temporary knowledge graph, and obtaining traversal data;

[0138] From all the traversed data obtained at the time of termination of the iteration, target deviation data that does not satisfy the node dependency condition is determined; wherein, the target deviation data includes first deviation data that does not satisfy the relationship condition and / or second deviation data that does not satisfy the influence condition;

[0139] The deviation degree of the temporary knowledge graph is determined based on the target deviation data.

[0140] In one embodiment, the map modification module 408 is specifically used for:

[0141] For each piece of traversed data, each knowledge point in the traversed data is taken as a predecessor knowledge point, and the dependency relationship between the predecessor knowledge point and the successor knowledge point is checked to see if the relationship condition is satisfied.

[0142] Based on the detection results, determine the first knowledge point in each piece of traversed data that does not satisfy the relation condition;

[0143] Data noise reduction is performed on the first knowledge point and the mastery level corresponding to the first knowledge point in all the traversed data to determine the first deviation data.

[0144] In one embodiment, the map modification module 408 is specifically used for:

[0145] For each set of traversed data, each knowledge point in the traversed data is taken as a predecessor knowledge point. Based on the user's mastery level of the predecessor knowledge point and the mastery level of the successor knowledge point, a transfer influence parameter is obtained. The transfer influence parameter represents the degree of change in the mastery level of the successor knowledge point compared to the predecessor knowledge point.

[0146] Determine the first proportion of migration influence parameters within a preset range and the second proportion of migration influence parameters outside the preset range in all the traversed data.

[0147] When the first proportion is less than the second proportion, the predecessor knowledge point and the successor knowledge point are determined as the second knowledge point that does not meet the influence condition;

[0148] Data noise reduction is performed on the second knowledge point and the corresponding mastery level to determine the second deviation data.

[0149] In one embodiment, for the current set of units, the mastery level determination module 406 is specifically used for:

[0150] When the current set of questions includes objective questions, the answer data corresponding to each objective question in the current set of questions is compared with the standard answer to obtain a second comparison result; and / or,

[0151] When the current unit set includes subjective questions, the subjective questions in the current unit set are divided into different question sets; wherein, the subjective questions included in the same question set are identical; for any question set, the answer data of each user object for the subjective question is compared with the standard answer to obtain the average score of the subjective question; based on the second score ratio between the average score and the preset full score, a third comparison result is obtained; wherein, the third comparison result is the average of the second score ratios of each question set in the current unit set.

[0152] In one embodiment, the knowledge graph construction apparatus further includes a difficulty labeling module, which is used for:

[0153] Based on the second comparison result and / or the third comparison result, the initial difficulty of each knowledge point in the target knowledge graph is determined; the initial difficulty of each knowledge point is normalized to obtain the target difficulty; and the target difficulty is marked at the corresponding knowledge point in the target knowledge graph.

[0154] The knowledge graph construction apparatus provided in this disclosure can execute the knowledge graph construction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0155] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Figure 5 As shown, the electronic device 500 includes one or more processors 501 and memory 502.

[0156] The processor 501 may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 500 to perform desired functions.

[0157] The memory 502 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 501 may execute the program instructions to implement the knowledge graph construction method of the embodiments of this disclosure described above and / or other desired functions. Various contents such as input signals, signal components, and noise components may also be stored in the computer-readable storage medium.

[0158] In one example, the electronic device 500 may also include an input device 503 and an output device 504, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0159] In addition, the input device 503 may also include, for example, a keyboard, a mouse, etc.

[0160] The output device 504 can output various information to the outside, including determined distance information, direction information, etc. The output device 504 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0161] Of course, for the sake of simplicity, Figure 5 Only some of the components of the electronic device 500 relevant to this disclosure are shown, omitting components such as buses, input / output interfaces, etc. In addition, the electronic device 500 may include any other suitable components depending on the specific application.

[0162] In addition to the methods and devices described above, embodiments of this disclosure may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the knowledge graph construction method described in the embodiments of this disclosure.

[0163] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this disclosure. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on a user's computing device, partially on a user's computing device, as a standalone software package, partially on a user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0164] Furthermore, embodiments of this disclosure may also be computer-readable storage media storing a computer program that, when executed by a computing device, enables the computing device to implement the knowledge graph construction method provided in embodiments of this disclosure.

[0165] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0166] This disclosure also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the methods described in this disclosure.

[0167] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0168] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a knowledge graph, characterized in that, include: Obtain an initial knowledge graph; wherein the initial knowledge graph is used to describe multiple knowledge points with dependencies; Obtain answer data from different users for multiple questions; wherein, the questions are pre-labeled, and the labels include: the standard answer to the question and the corresponding knowledge point; Based on the answer data of the question corresponding to the knowledge point and the standard answer, determine the mastery level of at least one user object for the corresponding knowledge point; Based on the user's mastery level of the corresponding knowledge points, the dependencies of the knowledge points in the initial knowledge graph are modified to obtain the target knowledge graph; The step of modifying the dependencies of knowledge points in the initial knowledge graph based on the user's mastery level of the corresponding knowledge points to obtain the target knowledge graph includes: Using the initial knowledge graph as a temporary knowledge graph, the following modification operations are repeatedly performed until the modification operations meet the preset iteration stopping rule: The deviation of the temporary knowledge graph is determined based on preset node dependency conditions and the user's mastery level of the corresponding knowledge points. The node dependency conditions include relationship conditions and influence conditions. The relationship condition indicates that the mastery level of the predecessor knowledge point is not less than the mastery level of the successor knowledge point. The influence condition indicates that the transfer effect of the mastery level of the predecessor knowledge point on the mastery level of the successor knowledge point is within a preset range. The predecessor knowledge point and the successor knowledge point are two adjacent knowledge points along the learning path of the temporary knowledge graph. Determine whether the deviation reaches a preset deviation threshold; If the condition is met, the dependency relationships of the knowledge points in the temporary knowledge graph are modified according to the node dependency conditions and the user object's mastery level of the corresponding knowledge points, to obtain a new temporary knowledge graph. If the condition is not met, the modification operation is determined to satisfy the iteration stopping rule, and the temporary knowledge graph corresponding to the time when the modification operation stops is determined as the target knowledge graph.

2. The method according to claim 1, characterized in that, The step of determining at least one user's mastery level of a corresponding knowledge point based on the answer data of the question corresponding to the knowledge point and the standard answer includes: Based on the knowledge points provided in the questions, the questions are divided into different combinations; wherein, the questions in the same combination have the same knowledge points. The answer data corresponding to each question in each combination is compared with the standard answer, and the mastery level of at least one user object for the corresponding knowledge point is determined based on the comparison result.

3. The method according to claim 2, characterized in that, The problem is divided into different combinations, including: The questions corresponding to each user object are divided into multiple thematic units with different knowledge points; The questions corresponding to all the user objects are divided into multiple unit sets with different knowledge points; wherein, the same unit set includes the same topic unit of different user objects.

4. The method according to claim 3, characterized in that, For the current thematic unit, comparing the answer data corresponding to each question in each of the combinations with the standard answer includes: The answer data corresponding to each question in the current topic unit is compared with the standard answer to obtain the score of each question; A first comparison result is obtained based on the first score ratio between the score value corresponding to each question and the preset full score value; wherein, the first comparison result is the average value of each of the first score ratios in the current topic unit.

5. The method according to claim 1, characterized in that, The step of determining the deviation of the temporary knowledge graph based on preset node dependency conditions and the user's mastery level of the corresponding knowledge points includes: Starting from any of the user objects, the traversal operation is performed iteratively until all user objects have completed the traversal operation, at which point the iteration terminates; wherein, the traversal operation is: traversing the temporary knowledge graph, checking the user object's mastery level of the knowledge points in the temporary knowledge graph, and obtaining traversal data; From all the traversed data obtained at the time of termination of the iteration, target deviation data that does not satisfy the node dependency condition is determined; wherein, the target deviation data includes first deviation data that does not satisfy the relationship condition and / or second deviation data that does not satisfy the influence condition; The deviation degree of the temporary knowledge graph is determined based on the target deviation data.

6. The method according to claim 5, characterized in that, From all the traversed data obtained at the time of termination of iteration, the target deviation data that does not satisfy the node dependency condition is determined, including: For each piece of traversed data, each knowledge point in the traversed data is taken as a predecessor knowledge point, and the dependency relationship between the predecessor knowledge point and the successor knowledge point is checked to see if the relationship condition is satisfied. Based on the detection results, determine the first knowledge point in each piece of traversed data that does not satisfy the relation condition; Data noise reduction is performed on the first knowledge point and the mastery level corresponding to the first knowledge point in all the traversed data to determine the first deviation data.

7. The method according to claim 5, characterized in that, From all the traversed data obtained at the time of termination of iteration, the target deviation data that does not satisfy the node dependency condition is determined, including: For each set of traversed data, each knowledge point in the traversed data is taken as a predecessor knowledge point. Based on the user's mastery level of the predecessor knowledge point and the mastery level of the successor knowledge point, a transfer influence parameter is obtained. The transfer influence parameter represents the degree of change in the mastery level of the successor knowledge point compared to the predecessor knowledge point. Determine the first proportion of migration influence parameters within a preset range and the second proportion of migration influence parameters outside the preset range in all the traversed data. When the first proportion is less than the second proportion, the predecessor knowledge point and the successor knowledge point are determined as the second knowledge point that does not meet the influence condition; Data noise reduction is performed on the second knowledge point and the corresponding mastery level to determine the second deviation data.

8. The method according to claim 3, characterized in that, For the current set of units, comparing the answer data corresponding to each question in each combination with the standard answer includes: When the questions in the current unit set include objective questions, the answer data corresponding to each objective question in the current unit set is compared with the standard answer to obtain a second comparison result; And / or, When the questions in the current unit set include subjective questions, the subjective questions in the current unit set are divided into different question sets; wherein, the subjective questions included in the same question set are the same. For any of the aforementioned question sets, the answer data of each user object to the subjective question is compared with the standard answer to obtain the average score of the subjective question; A third comparison result is obtained based on the second score ratio between the average score and the preset full score; wherein, the third comparison result is the average of the second score ratios of each question set in the current unit set.

9. The method according to claim 8, characterized in that, The method further includes: Based on the second comparison result and / or the third comparison result, determine the initial difficulty of each knowledge point in the target knowledge graph; The initial difficulty level of each knowledge point is normalized to obtain the target difficulty level. The difficulty level of the target is marked at the corresponding knowledge point in the target knowledge graph.

10. A knowledge graph construction apparatus, characterized in that, include: The knowledge graph acquisition module is used to acquire an initial knowledge graph; wherein, the initial knowledge graph is used to describe multiple knowledge points with dependencies. The answer data acquisition module is used to acquire answer data from different users for multiple questions; wherein, the questions are pre-labeled, and the labels include: the standard answer to the question and the knowledge point corresponding to the question; The mastery level determination module is used to determine the mastery level of at least one user object for the corresponding knowledge point based on the answer data of the question corresponding to the knowledge point and the standard answer; The knowledge graph modification module is used to modify the dependency relationships of knowledge points in the initial knowledge graph according to the user's mastery level of the corresponding knowledge points, so as to obtain the target knowledge graph. The map modification module is also used for: Using the initial knowledge graph as a temporary knowledge graph, the following modification operations are repeatedly performed until the modification operations meet the preset iteration stopping rule: The deviation of the temporary knowledge graph is determined based on preset node dependency conditions and the user's mastery level of the corresponding knowledge points. The node dependency conditions include relationship conditions and influence conditions. The relationship condition indicates that the mastery level of the predecessor knowledge point is not less than the mastery level of the successor knowledge point. The influence condition indicates that the transfer effect of the mastery level of the predecessor knowledge point on the mastery level of the successor knowledge point is within a preset range. The predecessor knowledge point and the successor knowledge point are two adjacent knowledge points along the learning path of the temporary knowledge graph. Determine whether the deviation reaches a preset deviation threshold; If the condition is met, the dependency relationships of the knowledge points in the temporary knowledge graph are modified according to the node dependency conditions and the user object's mastery level of the corresponding knowledge points, to obtain a new temporary knowledge graph. If the condition is not met, the modification operation is determined to satisfy the iteration stopping rule, and the temporary knowledge graph corresponding to the time when the modification operation stops is determined as the target knowledge graph.

11. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store the processor's executable instructions; The processor is configured to read the executable instructions from the memory and execute the instructions to implement the knowledge graph construction method according to any one of claims 1-9.

12. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a computing device, enables the computer device to implement the knowledge graph construction method described in any one of claims 1-9.

Images