Construction And Application Of Student Portrait Based On Online Homework Behavior

The 19 th National Congress of the Communist Party of China(CPC)proposed that the country should move towards a stage of high-quality development,and put forward the requirements for the construction of a high-quality education system in the field of education.As a key link to improve the quality of education,homework plays an important role.With the development of education informatization,homework has experienced the development of paper-based,semi paperless and paperless.Online homework and intelligent homework have become the important application methods of current homework scene.At the same time,the CPC Central Committee and the State Council(SC)issued the overall plan for deepening the reform of education evaluation in the new era,emphasizing the need to "innovate evaluation tools,make use of modern information technologies such as artificial intelligence and big data,and explore the whole process of longitudinal evaluation of students' learning situation in all grades." The CPC Central Committee and the SC issued the general plan for deepening the reform of education evaluation in the new era.Under the background of deepening the reform of education evaluation,it is a research topic worthy of attention to use modern information technology to promote the optimization and innovation of homework.As a user feature description technology based on data mining,portrait technology is becoming more and more accurate.In the field of education,it has been proved that learner portrait plays an important role in learning performance and effect evaluation.In this study,the learner portrait is applied to online homework scenarios.The research object is a large number of homework behavior data stored in the online operating system log,and the portrait construction and application based on the portrait technology are carried out.Based on the relevant research and literature systematically combing,this paper expounds the important concepts and theoretical basis of this study,summarizes the current research status and data mining technology,and lays a foundation for the further development of this study.In the stage of portrait construction,firstly,from the perspective of theoretical model and practical experience,this paper analyzes the existing homework model and mature measurement scale,interviews with front-line teachers,extracts the "initial dimension set" of the portrait model,and then uses two rounds of Delphi method to finally determine the main dimensions of the portrait.Then,according to the online homework data,the index features of each dimension are selected and screened.Finally,a portrait model based on basic features,behavior features(homework participation,homework effort,homework habit)and result features is generated.In the application stage of portrait,firstly,according to the application process of portrait,the portrait target is determined,and the data is cleaned and processed.Then,the students' performance is analyzed and visualized from the basic features,behavior features and result features.Then,based on the random forest regression algorithm,the important index features are selected as the main clustering features.The K-means clustering algorithm is used to identify three kinds of students with different behaviors,namely "high input-high efficiency-high efficiency","high participation-low efficiency-delay","low participation-low efficiency-delay".On this basis,linear stepwise regression is used to explore the important index features that affect the performance of different students.Finally,three kinds of student group portraits are output based on the analysis results of portrait,including group features and guidance.In order to verify the application effect of learner portrait based on online homework behavior,the study invited researchers in learning analysis and first-line teachers to interview.There are some innovations in the study.Firstly,based on the analysis of specific scenarios of homework behavior,construct the homework portrait model from the theoretical and practical point of view,enrich the application scenarios of learning analysis;Secondly,analyze and visualize students' homework performance from multiple angles,provide a diversified new perspective for students' homework evaluation;Thirdly,based on big data,different types of student groups are identified,which provides a reference for the realization of personalized homework and service.This study is based on the perspective of big data to solve practical problems in education,and provide scientific basis and effective methods for effective evaluation of homework and personalized support services.