Personalized job allocation method and platform based on data analysis

Abstract

The present invention discloses a personalized homework allocation method and platform based on data analysis, and relates to the field of data processing technology. The method includes: establishing a question database, performing dual association identification of knowledge points on the question database, and establishing knowledge point identification of the question; clustering student groups based on the completion of homework, and establishing a first group clustering result; locating the weak links of knowledge points based on student behavior data, adjusting the first group clustering result, and establishing a second group clustering result; matching the knowledge point identification of the question database, and establishing a matching result; constructing a first homework question bank based on the matching result, configuring the second homework question bank based on complexity constraints, and performing homework allocation of the second group clustering result according to the first homework question bank and the second homework question bank. The technical problem that the prior art cannot accurately allocate homework according to the personalized needs of students is solved, and the technical effect of personalized homework allocation is achieved through data analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a personalized job allocation method and platform based on data analysis. Background Art

[0002] In the field of education, with the rapid development of information technology and the increasing popularity of big data applications, personalized teaching has become an important way to improve the quality of education. Traditional homework assignment methods generally adopt a unified and standardized model, that is, all students receive homework of the same difficulty and content, and lack sufficient consideration of individual differences among students. This traditional method usually only focuses on the universality and generality of the questions, ignoring the individual differences between students and the imbalance of learning progress, resulting in students being unable to obtain targeted improvements during the homework process. Therefore, it is difficult to meet the personalized needs of students, especially for students with different learning foundations. The unified homework content may make it difficult for some students to keep up with the progress, while other students lack sufficient challenges. In addition, the existing technology lacks effective use of students' historical learning data, and cannot accurately identify students' knowledge weaknesses and learning preferences, resulting in the difficulty of homework arrangement to effectively consolidate and improve. Summary of the invention

[0003] This application provides a personalized homework assignment method and platform based on data analysis, which solves the technical problem in the prior art that it is impossible to accurately assign homework according to students' personalized needs.

[0004] In view of the above problems, embodiments of the present application provide a personalized job allocation method and platform based on data analysis.

[0005] A first aspect of an embodiment of the present application provides a personalized job allocation method based on data analysis, the method comprising:

[0006] Establish a question database, standardize the data in the question database, perform dual correlation identification of knowledge points on the question database, and establish knowledge point identification of the questions in the question database, wherein the dual correlation identification includes surface semantic correlation identification and implicit semantic analysis identification; configure a historical homework database of students, cluster student groups based on the homework completion status in the historical homework database, establish a first group clustering result, and configure a knowledge point weakness identification of the first clustering result; read student behavior data in the historical homework database, locate the weak links of knowledge points based on the student behavior data, adjust the first group clustering result based on the knowledge point weak link location result and the weakness identification, establish a second group clustering result, and the second group clustering result has a corrected weakness identification; use the corrected weakness identification and complexity constraints to match the knowledge point identification of the question database and establish a matching result; construct a first homework question bank based on the matching result, configure the second homework question bank based on the complexity constraint, the second homework question bank is a random test homework question bank, and the homework of the second group clustering result is assigned according to the first homework question bank and the second homework question bank.

[0007] A second aspect of the embodiment of the present application provides a personalized job allocation platform based on data analysis, the platform comprising:

[0008] An association recognition module is used to establish a question database, standardize the data in the question database, perform dual association recognition of knowledge points on the question database, and establish knowledge point identification of the questions in the question database, wherein the dual association recognition includes surface semantic association recognition and implicit semantic analysis recognition; a clustering module is used to configure a historical homework database of students, cluster student groups based on the homework completion status in the historical homework database, establish a first group clustering result, and configure a knowledge point weakness identification of the first clustering result; a clustering adjustment module is used to read student behavior data in the historical homework database, based on The student behavior data is used to locate the weak links of knowledge points, and the first group clustering results are adjusted based on the knowledge point weak link location results and weak identifiers to establish a second group clustering result, and the second group clustering result has a corrected weak identifier; a matching module, the matching module is used to use the corrected weak identifier and complexity constraints to match the knowledge point identifiers of the question database to establish a matching result; a homework allocation module, the homework allocation module is used to construct a first homework question bank based on the matching results, configure the second homework question bank based on complexity constraints, the second homework question bank is a random test homework question bank, and the homework allocation of the second group clustering results is performed according to the first homework question bank and the second homework question bank.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] First, a question database is established. After the data in the question database is standardized, the question database is subjected to dual association recognition of knowledge points, and the knowledge point identification of the questions in the question database is established. The dual association recognition includes surface semantic association recognition and implicit semantic analysis recognition. Then, the student's historical homework database is configured, and the student group is clustered based on the homework completion status in the historical homework database, the first group clustering result is established, and the knowledge point weak identification of the first clustering result is configured. Further, the student behavior data in the historical homework database is read, the weak link of the knowledge point is located based on the student behavior data, and the first group clustering result is adjusted based on the weak link location result and the weak identification of the knowledge point, and the second group clustering result is established, and the second group clustering result has a corrected weak identification. Then, the knowledge point identification of the question database is matched using the corrected weak identification and complexity constraints to establish a matching result. Finally, the first homework question bank is constructed based on the matching results, and the second homework question bank is configured based on the complexity constraints. The second homework question bank is a random test homework question bank, and the homework of the second group clustering result is assigned according to the first homework question bank and the second homework question bank. The technical problem that the existing technology cannot accurately assign homework according to the personalized needs of students is solved, and the technical effect of personalized homework assignment is achieved through data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A flowchart of a personalized job allocation method based on data analysis provided in an embodiment of the present application;

[0013] Figure 2 A schematic diagram of the structure of a personalized job allocation platform based on data analysis provided in an embodiment of the present application.

[0014] Explanation of the reference numerals: association identification module 11 , clustering module 12 , clustering adjustment module 13 , matching module 14 , job allocation module 15 . DETAILED DESCRIPTION

[0015] The embodiments of the present application solve the technical problem in the prior art that it is impossible to accurately assign homework according to the personalized needs of students by providing a personalized homework assignment method and platform based on data analysis.

[0016] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0017] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or are inherent to these processes, methods, products or devices.

[0018] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a personalized job allocation method based on data analysis, wherein the method includes:

[0019] A question database is established, and after the data in the question database is standardized, dual association recognition of knowledge points is performed on the question database to establish knowledge point identifiers of the questions in the question database, wherein the dual association recognition includes surface semantic association recognition and implicit semantic analysis recognition.

[0020] Questions are collected extensively from textbooks, supplementary teaching materials, past exam questions, online education resources and other channels to ensure that the questions cover the main knowledge points and difficulty levels of each subject. The collected questions are entered into the database in a unified format to obtain a question database. The data in the question database includes multiple dimensions such as the text content of the question, question type, difficulty, and knowledge points involved. After the data is stored, the data in the question database is standardized to ensure that the format, coding, punctuation, etc. of all questions meet the specifications.

[0021] After the data in the question database is standardized, dual association recognition of knowledge points is performed on each question in the question database. Dual association recognition includes surface semantic association recognition and implicit semantic analysis recognition; surface semantic association recognition is a direct semantic analysis of the question to identify the knowledge points explicitly mentioned in the question; implicit semantic analysis recognition is more in-depth, aiming to identify the implicit and deeper knowledge points in the question. Based on the results of dual association recognition, a corresponding knowledge point identifier is established for each question. These knowledge point identifiers not only reflect the knowledge content involved on the surface of the question, but also include the deep knowledge points that the question may involve. Through surface and implicit dual semantic recognition, each question in the question database can be accurately mapped to the knowledge points it involves, providing data support for subsequent personalized homework allocation.

[0022] Furthermore, the dual association identification of knowledge points in the question database further includes:

[0023] A knowledge point database is established, and the knowledge points in the knowledge point database are classified based on subjects, and the knowledge points are sorted out through the education syllabus to update the knowledge point database; keywords are extracted from the question database after data standardization to establish high-frequency key phrases; high-frequency key phrases are used to match the updated knowledge point database to establish surface semantic association recognition; the question database after data standardization is used as input data and input into the deep learning model, and the question text is converted into a sentence vector using the deep learning model, and the semantic similarity between the sentence vector and the knowledge point database is calculated to establish a first analysis result; after calling the historical data of question answers, collaborative filtering analysis is performed on the historical data of question answers to establish a second analysis result; implicit semantic analysis and recognition are established based on the first and second analysis results, and surface semantic association recognition and implicit semantic analysis and recognition are integrated to complete the dual association recognition of knowledge points.

[0024] Specifically, a knowledge point library covering multiple subjects is established, and knowledge points are classified based on subjects, so that each knowledge point in the knowledge point library can be managed accordingly with the subject to which it belongs; at the same time, knowledge points are regularly sorted and updated by referring to the latest education syllabus to ensure that the content in the knowledge point library is up-to-date and meets teaching requirements. For the standardized question database, natural language processing technology (NLP) is used to extract keywords and phrases, and high-frequency key phrases closely related to knowledge points are screened out by counting word frequencies; the extracted high-frequency key phrases are matched with the updated knowledge point library to preliminarily establish surface semantic associations between questions and knowledge points; the standardized question database is used as input data and imported into a pre-trained deep learning model. The deep learning model can convert question text into sentence vectors, and then calculate the semantic similarity between the sentence vectors and the knowledge point library; based on the calculation results of the semantic similarity, the first analysis result between the question and the knowledge point is generated, that is, based on deep learning. The implicit semantic association of the learning model; when performing implicit semantic analysis, the historical data of question answers will also be called, and the historical data of question answers records the students' answers to the questions and the accuracy rate; collaborative filtering analysis is performed on the historical data of question answers, and based on the behavior and performance of other students when answering similar questions, the knowledge points that may be involved in the questions are deduced, and then the association between the questions is generated, that is, the second analysis result; the first analysis result based on the deep learning model and the second analysis result based on the collaborative filtering analysis are combined to complete the implicit semantic analysis and recognition; the results of the implicit semantic analysis and recognition are integrated with the results of the surface semantic association recognition to form a complete knowledge point dual association recognition.

[0025] A historical homework database of students is configured, student groups are clustered based on the homework completion status in the historical homework database, a first group clustering result is established, and a knowledge point weakness indicator of the first clustering result is configured.

[0026] Collect students' historical homework data from the school's teaching management system, online homework platform and other channels to build a historical homework database. The historical homework database contains all homework records completed by students in the past learning process. Each homework record should include the content of the question, the student's completion status (such as correct or incorrect, completion time), submission time of the homework, knowledge points involved, homework score, etc.

[0027] By analyzing the completion of homework in the historical homework database, it is possible to evaluate each student's mastery of each knowledge point. Specifically, homework completion can be evaluated through multiple dimensions such as students' accuracy and completion speed; based on these evaluation results, clustering algorithms (such as K-means clustering, hierarchical clustering, etc.) are used to divide students into multiple groups to form the first group clustering results. Among them, students in each group show similar characteristics in homework completion, such as similar knowledge point mastery or weak links. By comparing students' homework performance on different knowledge points, the weak areas of each group on specific knowledge points can be identified, and the corresponding knowledge point weak identification can be configured for the first group clustering results.

[0028] Furthermore, clustering the student groups based on the homework completion status in the historical homework database to establish a first group clustering result also includes:

[0029] Based on the completion status of the homework, feature extraction is performed to establish feature extraction results, which include homework performance features and knowledge point performance features; after the feature extraction results are normalized, hierarchical clustering is used to perform hierarchical clustering on the normalized results to establish a first group clustering result; a knowledge point weakness analysis is performed on the first group clustering result, and the analysis features of the knowledge point weakness analysis include knowledge point accuracy analysis features, knowledge point mastery curves, and knowledge point difficulty and accuracy relationship features; a knowledge point weakness identification is established based on the knowledge point weakness analysis results.

[0030] Specifically, multi-dimensional feature extraction is performed based on the homework completion status in the historical homework database. The goal of feature extraction is to extract key information from students' homework performance that is helpful for group division. The extracted features are mainly divided into homework performance features and knowledge point performance features. Homework performance features include students' accuracy, error rate, completion time, submission time, homework scores, etc. Knowledge point performance features mainly reflect students' mastery of each knowledge point, including the accuracy of each knowledge point (the accuracy of students' answers on different knowledge points) and the degree of mastery of knowledge points (the difficulty coefficient and accuracy of knowledge points are used to evaluate the depth of students' mastery). After completing feature extraction, in order to ensure that the weights between different features are not affected by the dimension, the extracted features need to be normalized. Normalization can adjust the feature values ​​to the same numerical range, thereby ensuring that the clustering algorithm can more accurately measure the impact of different features on the division of student groups. Using the normalized feature values, the hierarchical clustering method is used to divide the student groups, and the first group clustering results are formed by gradually merging individual students or small groups into larger groups.

[0031] For the generated first group clustering results, further knowledge point weakness analysis is conducted, including knowledge point accuracy analysis features, knowledge point mastery curve, and knowledge point difficulty and accuracy relationship features. Knowledge point accuracy analysis features: Analyze the students' accuracy on each knowledge point. By comparing the accuracy of different knowledge points in the group, it is possible to identify which knowledge points the student group has weaker performance on. Knowledge point mastery curve: Analyze the students' mastery progress on multiple knowledge points and draw a learning mastery curve. The learning mastery curve shows the changes in students' mastery of different knowledge points during the learning process, thereby helping to identify long-term weaknesses or recently emerged weaknesses. Knowledge point difficulty and accuracy relationship features: By analyzing the relationship between the difficulty coefficient of the knowledge point and the student's accuracy, it is determined whether students have insufficient mastery of knowledge points with higher difficulty. Based on the results of the above knowledge point weakness analysis, a knowledge point weakness mark is established for the first group clustering results. The mark clearly marks the deficiencies of the student group on certain knowledge points and reflects the knowledge areas that each group needs to focus on during the learning process.

[0032] Read the student behavior data in the historical homework database, locate the weak links of knowledge points based on the student behavior data, adjust the first group clustering results based on the weak link location results of knowledge points and weak identification, and establish the second group clustering results, and the second group clustering results have corrected weak identifications.

[0033] Students' behavioral data are extracted from the historical homework database. The behavioral data include but are not limited to the completion time of the homework (the time it takes students to complete each homework), the number of pauses or revisions in the problem-solving process (reflecting students' thinking or correction behavior in the problem-solving process), the error rate and error correction process in the homework (analyzing the distribution of students' errors and the number of corrections in the homework), and the time pattern of homework submission (whether there is any delayed submission or early completion). These behavioral data can reflect the difficulties students encounter in the process of completing homework and their problem-solving habits, and serve as an important basis for identifying weak links in knowledge points.

[0034] Using student behavior data, further analyze the weak links of each student in the problem-solving process in order to locate specific knowledge points. Optionally, by analyzing the time it takes for students to complete their homework and the number of errors or revisions that occur during the homework process, infer which knowledge points the students have insufficiently mastered. For example, if students spend too much time on certain knowledge points, have a high error rate, or frequently modify their answers, these may indicate that the knowledge point is a weak link.

[0035] According to the results of knowledge point weak link location, combined with the knowledge point weak link identification in the previously generated first group clustering results, the existing group division is optimized and adjusted. Based on the new weak link location results, the students in each group are re-analyzed, and students with similar weak links are re-divided into the same group, so as to ensure that the students in the group have more consistent performance in the knowledge point weak links. With the adjustment of group division, it is necessary to update the knowledge point weak link identification of each group; based on the new knowledge point weak link location results, the knowledge point weak link identification of each group is updated to reflect the latest student performance. After completing the optimization and adjustment of the first group clustering results, a new second group clustering result is established. The second group clustering result more accurately reflects the students' weak situation in knowledge points than the first group clustering result. The second group clustering result not only reflects the students' performance in homework completion, but also combines the students' behavioral data with the revised knowledge point weak link identification. The revised identification more accurately reflects the deficiencies of each student group in specific knowledge points, which can provide a more targeted basis for subsequent personalized homework allocation.

[0036] Furthermore, the locating of weak links in knowledge points based on the student behavior data also includes:

[0037] Analyze student behavior data to obtain answering time, number of redos, answering order, and prompt dependency features; use question level and homework level to perform feature analysis on answering time and establish a first analysis result; evaluate knowledge point mastery based on the number of redos and establish a second analysis result; use answering order to perform confidence analysis on knowledge points and establish a third analysis result; calculate the dependency degree of prompt dependency features and establish a fourth analysis result; use the first analysis result, second analysis result, third analysis result, and fourth analysis result to locate weak links in knowledge points.

[0038] Preferably, the students' answering time, number of redos, answering order and prompt dependency features are extracted from the historical homework database. The answering time refers to the time the students spend to complete each question. The number of redos refers to the number of attempts the students make on a question. The more redos, the more likely it is that the students have difficulty with the knowledge points involved in the question. The answering order refers to the order in which the students choose to answer the questions when completing the homework. The prompt dependency feature refers to the number of times the students rely on prompts in the answering process, so as to judge the degree of students' dependence on the knowledge point.

[0039] Questions and assignments are classified according to difficulty levels, and then the answering time at different difficulty levels is analyzed; for relatively simple questions, a long answering time may indicate that students have weak points in basic knowledge; for difficult questions, a faster answering time may indicate that students have a good grasp of the knowledge point; based on these analyses, a first parsing result is generated, reflecting students' mastery of knowledge points at different difficulty levels.

[0040] If students repeat certain questions many times, it means that they have not mastered the knowledge points involved in these questions sufficiently. By analyzing the number of times students repeat, combined with the knowledge points to evaluate the mastery, the second analysis result is generated. Among them, the knowledge points with more repeat times may be the weak links of students and need further consolidation; fewer repeat times indicate that students have a solid grasp of the knowledge points.

[0041] Analyze the order in which students answer questions in their homework, especially whether they tend to skip certain knowledge points and then come back to answer them. This behavior can reflect the students' confidence in certain questions. If students tend to skip questions on specific knowledge points, it indicates that they lack confidence in the knowledge points, and these knowledge points can be identified as potential weak links; if students answer questions from beginning to end with confidence, it indicates that they have a strong grasp of these knowledge points; based on this, generate the third analysis result.

[0042] Analyze the frequency of students using hints in the problem-solving process. Hint dependence refers to the degree to which students rely on hints in the problem-solving process. Knowledge points with a high hint dependence frequency may be students' weak points because students cannot solve related problems independently. Low hint dependence indicates that students are relatively confident in this knowledge point. Based on this analysis, generate the fourth analysis result.

[0043] The first, second, third and fourth analysis results are combined to locate the weak links in knowledge points; the first analysis result provides information about students' performance on knowledge points of different difficulty levels through answering time feature analysis; the second analysis result evaluates students' mastery of knowledge points through redo number analysis; the third analysis result evaluates students' confidence in knowledge points through answering order analysis; the fourth analysis result further determines whether students have a high dependence on certain knowledge points through prompt dependency calculation; combining these results can more comprehensively locate students' weak links in knowledge, providing a more accurate basis for subsequent personalized homework assignments and learning guidance.

[0044] Use the correction of weak identification and complexity constraints to match the knowledge point identifications in the question database and establish the matching results.

[0045] The corrected weak markers reflect the weak links of the student group in specific knowledge points. These corrected weak markers will serve as the core basis for subsequent matching questions to ensure that the assigned questions are targeted at the learning weaknesses of the student group. When matching questions, in addition to considering the weak links of knowledge points, it is also necessary to make reasonable choices based on the complexity of the questions. Complexity constraints can include question difficulty, knowledge point level, and question type. Specifically, according to the difficulty coefficient of the question itself, the questions are divided into basic questions, medium-difficulty questions, and high-difficulty questions to ensure that the questions assigned to students are of moderate difficulty, which can not only consolidate basic knowledge but also appropriately challenge students' learning ability; according to the hierarchy of knowledge points, ensure that the questions cover basic knowledge and higher-level knowledge; question types include multiple-choice questions, fill-in-the-blank questions, short-answer questions, etc., to ensure that the types of questions assigned to students are diverse, covering different ways of thinking and problem-solving strategies. The introduction of complexity constraints is to avoid an imbalance in the difficulty of student assignments and ensure that the questions can meet the learning objectives without affecting the learning effect due to excessive or low difficulty.

[0046] Each question in the question database is set with a corresponding knowledge point identifier. Based on the corrected knowledge point weakness identifier, the question database is matched with the knowledge point identifier. Specifically, the question knowledge point identifiers in the question database are matched one by one with the corrected weakness identifiers of the student group to ensure that the selected questions cover the students' weak knowledge points; based on the aforementioned complexity constraints, further select questions suitable for the students' current learning stage to ensure that the difficulty of the questions is consistent with the students' mastery. After completing the matching of knowledge point identifiers and complexity constraints, a matching result is generated. The matching result is a set of questions selected from the question database based on the knowledge point weakness identifiers and complexity requirements of the student group. The matching results will serve as the basis for the personalized homework question bank for subsequent homework assignments.

[0047] Furthermore, the method further comprises:

[0048] The student's current level score is established based on the historical homework database; the current level score is used to establish the complexity ratio of the question, and the complexity ratio of the question is used as the complexity constraint. The complexity constraint and the correction of the weak mark are used to match the knowledge point mark of the question database to establish the matching result.

[0049] The current level score of each student is calculated from the student's homework completion and behavior data in the historical homework database. This score combines the student's homework accuracy, answer time, number of redos, and other related performance to form a comprehensive evaluation of the student's current learning level. Specifically, the weight is assigned based on the student's accuracy on each knowledge point to generate the score result; the student's problem-solving efficiency is evaluated based on the answer time and the score is adjusted; the score is further revised based on the number of redos and the reliance on prompts to reflect the student's performance on questions of different difficulty levels.

[0050] Based on the student's current level score, determine the distribution of question complexity suitable for that level, that is, the question complexity ratio. The question complexity ratio refers to the proportion of questions of different difficulty levels in the assignments assigned to students, so as to ensure that there are both basic questions to consolidate students' knowledge and appropriate high-difficulty questions to challenge students' thinking. Specifically, according to the student's current level score, the students are divided into different level intervals (such as low level, medium level and high level), and the corresponding question complexity ratio is set for each interval. For example, for low-level students, the proportion of basic questions may be large, while for high-level students, the proportion of more difficult questions is larger; according to different scoring intervals, different question difficulty ratios are assigned. For example, the complexity ratio may be defined as: 60% for basic questions, 30% for medium-difficulty questions, and 10% for high-difficulty questions. After establishing the question complexity ratio, the question complexity ratio is used as a complexity constraint to ensure that the questions assigned to students cover different difficulty levels and meet their current learning level. The complexity constraint stipulates the proportion of questions of various difficulty levels in the question selection, making the question allocation more reasonable and preventing students from only doing basic questions or questions of too high difficulty, which affects their learning progress.

[0051] After completing the setting of complexity constraints, the knowledge point identifiers of the question database are matched in combination with the revised knowledge point weak identifiers. Specifically, based on the complexity constraints, a set of questions that meet the complexity ratio requirements are screened out from the question database to ensure that the assigned questions cover different difficulty levels; in the screened question set, the revised knowledge point weak identifiers are further used to select questions related to the students' weak knowledge points to ensure that the assigned questions can effectively train and consolidate the students' weak knowledge points; and finally a matching result is generated, that is, a set of questions that meet the complexity constraints of the students' current level scores and can effectively cover the weak links of knowledge points.

[0052] A first homework question bank is constructed based on the matching results, and a second homework question bank is configured based on complexity constraints. The second homework question bank is a random test homework question bank. The homework allocation of the second group clustering result is performed according to the first homework question bank and the second homework question bank.

[0053] The first homework question bank is constructed based on the matching results. The questions in the first homework question bank are highly relevant to the students' weak knowledge points, helping students to practice and consolidate their weak knowledge points in a targeted manner. In addition to the highly targeted first homework question bank, it is also necessary to provide students with additional question challenges to test their comprehensive performance on random knowledge points. Therefore, based on complexity constraints, the second homework question bank is configured, also known as the random test homework question bank. The questions in the second homework question bank not only include students' weak knowledge points, but also randomly cover other knowledge points to ensure a comprehensive test of students' knowledge mastery. Through the test of random knowledge points, students' learning progress can be more comprehensively evaluated and possible new weaknesses can be discovered.

[0054] After the first and second homework question banks are constructed, personalized homework assignments will be made based on the clustering results of the second group. Specifically, questions in the two question banks will be reasonably allocated according to the actual needs of the student group; questions in the first homework question bank will account for the majority, while questions in the second homework question bank will play a supplementary role to ensure that students receive sufficient comprehensive tests; the proportion of the two question banks in the homework will be controlled according to the learning goals and levels of the groups. For example, groups with more weak links may be assigned more questions in the first homework question bank, while groups with higher learning levels may increase the proportion of questions in the second homework question bank to enhance the assessment of comprehensive abilities; the allocation ratio will be dynamically adjusted according to the student's performance to ensure that the homework assignment can not only solve the current weak links, but also introduce new challenges in a timely manner to help students improve comprehensively.

[0055] Furthermore, the method further comprises:

[0056] Reading the homework assignment feedback of the students, wherein the homework assignment feedback includes the students' follow-up feedback and homework supervision feedback; generating reconstruction constraint instructions based on the homework assignment feedback, and reconstructing the complexity constraint ladder based on the reconstruction constraint instructions; and constructing the subsequent first homework question bank based on the ladder reconstruction result.

[0057] Preferably, after the students complete the homework, feedback data related to the homework assignment is collected and read. The homework assignment feedback includes student return feedback and homework supervision feedback. Student return feedback refers to the feedback information actively provided by students after completing the homework, which may include the difficulty of the questions, the feelings about the homework content, and the opinions on the homework amount. Homework supervision feedback comes from supervisors such as teachers or parents, and the feedback content includes the students' homework completion, attitude, and adaptation to the homework difficulty. Through the analysis of student return feedback and homework supervision feedback, reconstruction constraint instructions are generated to adaptively adjust the current complexity constraints according to the feedback information. Specifically, through natural language processing or rule analysis, student and supervision feedback are parsed to extract key information that affects the homework effect, such as whether the questions are too difficult or too easy, and whether the homework amount is too large or too small. According to the parsed feedback information, corresponding reconstruction constraint instructions are generated, including increasing or decreasing the difficulty of the questions, adjusting the number of questions, or further adjusting the homework assignment strategy for specific knowledge points. For example, if students feedback that some questions are too difficult, instructions to reduce the difficulty of such questions can be generated.

[0058] After generating the reconstruction constraint instructions, the current complexity constraints are reconstructed in a step-by-step manner based on these instructions. Step-by-step reconstruction refers to making hierarchical adjustments to the complexity constraints of the homework questions based on feedback information to ensure that the adjustments can be implemented gradually and to avoid adaptability problems caused by large-scale changes at one time. Specifically, based on the feedback, the difficulty distribution of the questions is appropriately adjusted. For example, if the difficulty of the questions is too high, the complexity level of the questions will be gradually reduced to ensure that students can gradually transition from basic to more difficult questions; the proportion of questions of different difficulty levels in the homework is adjusted. For example, if students indicate in the follow-up visit that there are too few basic questions, the proportion of basic questions can be increased, and vice versa.

[0059] After completing the step-by-step reconstruction of the complexity constraint, the subsequent first homework question bank will be constructed based on the new complexity constraint results. Based on the revised complexity constraint, reselect questions that meet the new complexity constraint in the question database to construct a new first homework question bank. The constructed new question bank will continue to accept student feedback to form a closed-loop feedback adjustment mechanism to ensure that the system can continuously and dynamically optimize homework allocation to meet students' personalized learning needs.

[0060] Furthermore, the method further comprises:

[0061] A long-term window record of the homework assignment feedback for students is established to create a long-term window record result; rolling window feature extraction is performed on the long-term window record result to create a learning rule feature, wherein the learning rule feature includes a volatility feature and a stability feature; a compensation optimization strategy is generated based on the learning rule feature, and construction optimization of the first homework question bank is performed based on the compensation optimization strategy.

[0062] Preferably, a long-term window record is performed on the feedback of students' homework assignments, that is, the information such as students' homework completion status, homework feedback, homework supervision feedback, etc. is continuously collected and recorded over a long period of time (such as several weeks or months). The long-term window record includes each homework performance of the students and related feedback, forming long-term historical data, namely, the long-term window record result; rolling window feature extraction is performed on the long-term window record result, that is, the long-term data is divided into multiple small time windows (for example, weekly or monthly), and the students' learning performance, knowledge point mastery, homework completion status, etc. are analyzed in these time windows to extract valuable learning regularity features, which include volatility features and stability features; volatility features refer to whether there are significant fluctuations in students' learning performance, for example, in certain time periods, the quality or accuracy of students' homework completion may fluctuate greatly, suggesting that they may have encountered some difficulties or challenges in learning; stability features reflect whether students' learning performance shows a relatively stable trend over a long period of time, for example, students' mastery of certain knowledge points may remain at a relatively stable level, indicating that the knowledge points have been mastered more solidly.

[0063] Based on the learning regularity characteristics extracted from the rolling window, the corresponding compensation optimization strategy is generated. The compensation optimization strategy aims to adjust and optimize the students' learning fluctuations to ensure that students maintain steady improvement in the learning process. Specifically, for students with large fluctuations in learning performance, basic questions can be added, homework difficulty can be reduced, or more review questions can be assigned to help students return to stability in certain weak stages to avoid excessive challenges that lead to a decrease in learning efficiency; for students with relatively stable learning performance, the proportion of high-difficulty questions can be appropriately increased, or more challenging questions can be introduced to help students further improve their abilities on the basis of mastering stability. Finally, based on the compensation optimization strategy, the construction of the first homework question bank is further optimized. Specifically, according to the compensation strategy, the distribution of questions in the question bank is adjusted, such as adding basic questions to cope with learning fluctuations, or adding challenging questions to promote students to further improve; based on the students' long-term learning rules, the difficulty and type of questions are dynamically adjusted in the subsequent homework assignments, so that the homework content is more in line with the students' learning rhythm, helping students gradually stabilize and improve their learning ability; by comprehensively analyzing the learning regularity characteristics of each student, the homework question bank content can be more personalized to ensure that each student gets the homework task that best matches their learning status.

[0064] In summary, the embodiments of the present application have at least the following technical effects:

[0065] First, a question database is established. After the data in the question database is standardized, the question database is subjected to dual association recognition of knowledge points, and the knowledge point identification of the questions in the question database is established. The dual association recognition includes surface semantic association recognition and implicit semantic analysis recognition. Then, the student's historical homework database is configured, and the student group is clustered based on the homework completion status in the historical homework database, the first group clustering result is established, and the knowledge point weak identification of the first clustering result is configured. Further, the student behavior data in the historical homework database is read, the weak link of the knowledge point is located based on the student behavior data, and the first group clustering result is adjusted based on the weak link location result and the weak identification of the knowledge point, and the second group clustering result is established, and the second group clustering result has a corrected weak identification. Then, the knowledge point identification of the question database is matched using the corrected weak identification and complexity constraints to establish a matching result. Finally, the first homework question bank is constructed based on the matching results, and the second homework question bank is configured based on the complexity constraints. The second homework question bank is a random test homework question bank, and the homework of the second group clustering result is assigned according to the first homework question bank and the second homework question bank. The technical problem that the existing technology cannot accurately assign homework according to the personalized needs of students is solved, and the technical effect of personalized homework assignment is achieved through data analysis.

[0066] Embodiment 2 is based on the same inventive concept as the personalized job allocation method based on data analysis in the above embodiment. Figure 2 As shown, the present application provides a personalized job allocation platform based on data analysis. The platform includes:

[0067] An association identification module 11 is used to establish a question database, standardize the data in the question database, perform dual association identification of knowledge points on the question database, and establish knowledge point identification of the questions in the question database, wherein the dual association identification includes surface semantic association identification and implicit semantic analysis identification; a clustering module 12 is used to configure a student's historical homework database, cluster student groups based on the homework completion status in the historical homework database, establish a first group clustering result, and configure a knowledge point weakness identification of the first clustering result; a clustering adjustment module 13 is used to read student behavior data in the historical homework database, Based on the student behavior data, weak links of knowledge points are located, and based on the weak link location results of knowledge points and weak identifiers, the first group clustering results are adjusted to establish a second group clustering result, and the second group clustering result has a corrected weak identifier; a matching module 14, the matching module 14 is used to match the knowledge point identifiers of the question database with the corrected weak identifier and complexity constraints to establish a matching result; a homework assignment module 15, the homework assignment module 15 is used to construct a first homework question bank based on the matching result, configure a second homework question bank based on complexity constraints, the second homework question bank is a random test homework question bank, and the homework assignment of the second group clustering result is performed according to the first homework question bank and the second homework question bank.

[0068] Furthermore, the association identification module 11 is used to execute the following method:

[0069] A knowledge point database is established, and the knowledge points in the knowledge point database are classified based on subjects, and the knowledge points are sorted out through the education syllabus to update the knowledge point database; keywords are extracted from the question database after data standardization to establish high-frequency key phrases; high-frequency key phrases are used to match the updated knowledge point database to establish surface semantic association recognition; the question database after data standardization is used as input data and input into the deep learning model, and the question text is converted into a sentence vector using the deep learning model, and the semantic similarity between the sentence vector and the knowledge point database is calculated to establish a first analysis result; after calling the historical data of question answers, collaborative filtering analysis is performed on the historical data of question answers to establish a second analysis result; implicit semantic analysis and recognition are established based on the first and second analysis results, and surface semantic association recognition and implicit semantic analysis and recognition are integrated to complete the dual association recognition of knowledge points.

[0070] Furthermore, the clustering module 12 is used to perform the following method:

[0071] Based on the completion status of the homework, feature extraction is performed to establish feature extraction results, which include homework performance features and knowledge point performance features; after the feature extraction results are normalized, hierarchical clustering is used to perform hierarchical clustering on the normalized results to establish a first group clustering result; a knowledge point weakness analysis is performed on the first group clustering result, and the analysis features of the knowledge point weakness analysis include knowledge point accuracy analysis features, knowledge point mastery curves, and knowledge point difficulty and accuracy relationship features; a knowledge point weakness identification is established based on the knowledge point weakness analysis results.

[0072] Furthermore, the cluster adjustment module 13 is used to perform the following method:

[0073] Analyze student behavior data to obtain answering time, number of redos, answering order, and prompt dependency features; use question level and homework level to perform feature analysis on answering time and establish a first analysis result; evaluate knowledge point mastery based on the number of redos and establish a second analysis result; use answering order to perform confidence analysis on knowledge points and establish a third analysis result; calculate the dependency degree of prompt dependency features and establish a fourth analysis result; use the first analysis result, second analysis result, third analysis result, and fourth analysis result to locate weak links in knowledge points.

[0074] Furthermore, the matching module 14 is used to perform the following method:

[0075] The student's current level score is established based on the historical homework database; the current level score is used to establish the complexity ratio of the question, and the complexity ratio of the question is used as the complexity constraint. The complexity constraint and the correction of the weak mark are used to match the knowledge point mark of the question database to establish the matching result.

[0076] Furthermore, the job allocation module 15 is used to execute the following method:

[0077] Reading the homework assignment feedback of the students, wherein the homework assignment feedback includes the students' follow-up feedback and homework supervision feedback; generating reconstruction constraint instructions based on the homework assignment feedback, and reconstructing the complexity constraint ladder based on the reconstruction constraint instructions; and constructing the subsequent first homework question bank based on the ladder reconstruction result.

[0078] Furthermore, the job allocation module 15 is used to execute the following method:

[0079] A long-term window record of the homework assignment feedback for students is established to create a long-term window record result; rolling window feature extraction is performed on the long-term window record result to create a learning rule feature, wherein the learning rule feature includes a volatility feature and a stability feature; a compensation optimization strategy is generated based on the learning rule feature, and construction optimization of the first homework question bank is performed based on the compensation optimization strategy.

[0080] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0081] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0082] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A personalized job allocation method based on data analysis, characterized in that: The method comprises: Establishing a question database, standardizing the data in the question database, performing dual correlation recognition of knowledge points on the question database, and establishing knowledge point identifiers of the questions in the question database, wherein the dual correlation recognition includes surface semantic correlation recognition and implicit semantic analysis recognition; Configuring a historical homework database of students, clustering student groups based on the homework completion status in the historical homework database, establishing a first group clustering result, and configuring a knowledge point weakness indicator for the first clustering result; Reading student behavior data in a historical homework database, locating weak links of knowledge points based on the student behavior data, adjusting the first group clustering result based on the weak link locating result of the knowledge points and the weak identification, and establishing a second group clustering result, wherein the second group clustering result carries a corrected weak identification; Use the correction of weak identification and complexity constraints to match knowledge point identifications in the question database and establish matching results; A first homework question bank is constructed based on the matching results, and a second homework question bank is configured based on complexity constraints. The second homework question bank is a random test homework question bank. The homework allocation of the second group clustering result is performed according to the first homework question bank and the second homework question bank.

2. The personalized job allocation method based on data analysis according to claim 1, characterized in that: The dual association identification of knowledge points in the question database also includes: Establish a knowledge point database, classify the knowledge points in the knowledge point database based on subjects, sort out the knowledge points through the education syllabus, and update the knowledge point database; Extract keywords from the topic database after data standardization and establish high-frequency key phrases; Use high-frequency key phrases to match the updated knowledge point library and establish surface semantic association recognition; The question database after data standardization is used as input data and input into the deep learning model. After the question text is converted into a sentence vector using the deep learning model, the semantic similarity between the sentence vector and the knowledge point library is calculated to establish the first analysis result. After calling the historical data of question answers, perform collaborative filtering analysis on the historical data of question answers to establish a second analysis result; The first analysis result and the second analysis result are used to establish implicit semantic analysis and recognition, and the surface semantic association recognition and implicit semantic analysis and recognition are integrated to complete the dual association recognition of knowledge points.

3. The personalized job allocation method based on data analysis according to claim 1, characterized in that: The clustering of student groups based on the homework completion status in the historical homework database to establish a first group clustering result also includes: Perform feature extraction based on the completion status of the homework and establish feature extraction results, wherein the feature extraction results include homework performance features and knowledge point performance features; After normalizing the feature extraction results, hierarchical clustering is used to perform hierarchical clustering on the normalized results to establish the first group clustering results; A knowledge point weakness analysis is performed on the first group clustering result. The analysis features of the knowledge point weakness analysis include knowledge point accuracy analysis features, knowledge point mastery curves, and knowledge point difficulty and accuracy relationship features. A knowledge point weakness identification is established based on the knowledge point weakness analysis results.

4. The personalized job allocation method based on data analysis according to claim 1, characterized in that: The positioning of weak links in knowledge points based on the student behavior data also includes: Analyze student behavior data to obtain answering time, number of redos, answering order, and prompt dependency characteristics; Perform feature analysis on the answering time using the question level and the assignment level to establish a first analysis result; Evaluate the knowledge points based on the number of redoes and establish a second analysis result; Use the answering order to conduct confidence analysis of knowledge points and establish the third analysis result; Calculating the dependency of the prompt dependency feature to establish a fourth parsing result; The first analysis result, the second analysis result, the third analysis result, and the fourth analysis result are used to locate the weak links of the knowledge points.

5. The personalized job allocation method based on data analysis according to claim 1, characterized in that: The method further comprises: Establish students' current level scores based on historical homework database; The current level score is used to establish the complexity ratio of the question, and the complexity ratio of the question is used as the complexity constraint. The complexity constraint and the correction of the weak mark are used to match the knowledge point marks of the question database to establish the matching result.

6. The personalized job allocation method based on data analysis according to claim 5, characterized in that: The method further comprises: Reading the homework assignment feedback of the students, wherein the homework assignment feedback includes the student return visit feedback and homework supervision feedback; Generate a reconstruction constraint instruction based on the job allocation feedback, and reconstruct the complexity constraint ladder based on the reconstruction constraint instruction; The subsequent first homework question bank is constructed based on the step-by-step reconstruction results.

7. The personalized job allocation method based on data analysis according to claim 1, characterized in that: The method further comprises: Long-term window recording of homework assignment feedback for students, and establishment of long-term window recording results; Perform rolling window feature extraction on the long-term window recording results to establish learning regularity features, wherein the learning regularity features include volatility features and stability features; A compensation optimization strategy is generated based on the learning regularity characteristics, and the construction optimization of the first homework question bank is performed based on the compensation optimization strategy.

8. A personalized job allocation platform based on data analysis, characterized by: For implementing the personalized job allocation method based on data analysis according to any one of claims 1 to 7, the platform comprises: A correlation identification module, which is used to establish a question database, standardize the data in the question database, perform dual correlation identification of knowledge points on the question database, and establish knowledge point identifiers of the questions in the question database, wherein the dual correlation identification includes surface semantic correlation identification and implicit semantic analysis identification; A clustering module, the clustering module is used to configure a historical homework database of students, cluster student groups based on the homework completion status in the historical homework database, establish a first group clustering result, and configure a knowledge point weakness mark of the first clustering result; A clustering adjustment module, the clustering adjustment module is used to read the student behavior data in the historical homework database, locate the weak links of the knowledge points based on the student behavior data, adjust the first group clustering results based on the weak link location results of the knowledge points and the weak identification, and establish a second group clustering result, and the second group clustering result has a modified weak identification; A matching module, the matching module is used to match the knowledge point identifications of the question database by using the correction of the weak identification and the complexity constraint to establish a matching result; A homework assignment module is used to construct a first homework question bank based on matching results, configure a second homework question bank based on complexity constraints, the second homework question bank is a random test homework question bank, and perform homework assignment based on the second group clustering results according to the first homework question bank and the second homework question bank.

Images