Research On Learner Model And Pedagogical Model In Adaptative Learning System

Adaptive learning system can provide adaptive learning content and adaptive navigation to the learners according to their various characteristics and behavior tendencies,and recommend personalized learning paths and learning resources to the learners by adopting the personalized teaching strategies,in order to meet the needs of personalized learning.Research on adaptive learning system has achieved fruitful results in universities or research institutions.A theoretical and technical basis has been formed on adaptive learning system reference model,domain knowledge model,learner model,pedagogical model and system application et al.With the development of learning analysis technique,big data and other related fields,online education institutions have developed mature adaptive learning platform using their rich educational resources and advanced adaptive technologies,which provides development platforms for exploring practical and breakthrough technology theory.This study takes item response theory,social comparison theory and meta-cognitive theory as the theoretical guidance of system design,designs framework of adaptive learning system orientating learning experience,focuses on the two core components of learner model and pedagogical model after constructing domain knowledge model,and makes evaluation on four aspects of meta-cognitive learning experience,social-comparison learning experience,learning behavior analysis and subjective evaluation.The study mainly includes the following four parts:(1)Framework design of adaptive learning system and construction of domain knowledge model.This study takes adaptive educational hypermedia systems proposed by Brusilovsky professor of School of information science,University of Pittsburgh,as system framework reference model,and designs framework of adaptive learning system orientating learning experience adaptive.The system consists of domain knowledge model,learner model,pedagogical model,interface module and adaptive engine.Domain knowledge model is the foundation of adaptive learning system.This study lays out structure diagram of domain knowledge model taking junior middle school math as an example,sets attributes of domain knowledge elements based on learning object metadata specification(CELTS-3.1),designs domain knowledge model reference specification from three dimensions of subject,curriculum and learning objects according to model structure diagram and attributes of domain knowledge elements,and constructs domain knowledge model based on the domain knowledge model reference specification.(2)Construction and visualization of learner model.Based on knowledge element structure in domain knowledge model and learners' response to texts,videos and exercises,information of knowledge states,knowledge level,learning behaviors and other personality characteristics in learner model are built and updated by adopting overlay modeling and data-driven technique.During construction and real-time update of learner model,one focus is knowledge level estimation on knowledge units and knowledge points.Knowledge level of knowledge points is estimated using Rasch model with personalization of initial value based on learner's response to exercises,and knowledge level of knowledge units is estimated using the weight algorithm according to the relationship between knowledge unit and knowledge unit,knowledge unit and knowledge point.The other focus is visualization of the learner model.Open learner model based on mind map,called MindOLM,releases curriculum knowledge states in learner model to learner,and shows his/her learning progress in mind map.Open social learning model based on table,called TableOSLM,releases curriculum knowledge states in learner models of the learner and peer learners to the learner and shows their learning progresses of the learner himself/herself,group and peer learners in tabular form.(3)Design of pedagogical model.Pedagogical model emphasizes learners' adaptability to system and system adaptability to learners on the design idea,and consists of learner-driven strategies and system-driven strategies.Learner-driven strategies emphasizes autonomy of learners in the learning process,allows the learner plan the learning path,and presents their preferred learning content,such as knowledge points,learning objects(text,video,excercises),review of error excercises,peer learners and so on,according to their own choice.System-driven strategies focuses on system adaptability to learners and system intelligence,plans learning path for the learner,and recommends appropriate knowledge points,learning objects,review of learning objects(text,excercises),peer learners and other learning content to the learner,according to his/her current knowledge level of knowledge point and learning behaviors.(4)Research on implementation mechanism and evaluation.Implementation mechanism designs system architecture process and student learning process from the system and learners,and lists source codes of the core techniques.In the evaluation process,this study choose the students in sixth grade and seventh grade as participants,designs evaluation module including meta-cognitive learning experience,social-comparison learning experience,learning behavior analysis and subjective evaluation.The evaluation results show as the following four aspects:(1)MindOLM,TableOSLM and teaching strategies can well induce meta-cognitive learning experience.(2)TableOSLM has obvious effect on comparison method,self-improvement and social-comparison learning experience.(3)Learning behavior analysis results show that the sixth grade students in the absence of formal learning experience on knowledge points finally realize no significant difference learning effect on numbers of completed knowledge points and knowledge level comparing seventh grade students by learning resources of learning object and learning help,which indicates that the system is conducive to cultivating self-regulated learning ability.(4)Evaluation results from subjective questionnaire demonstrate that students have shown strong acceptance to usability,usefulness,and satisfaction with the system.The innovations of this study are mainly reflected in three aspects: knowledge level estimation of knowledge point,visualization views of learner model and teaching strategies.First,according to experimental conclusion of difficulty and knowledge level estimation based on Rasch model with personalization of initial value,we realize more accurate estimation on knowledge level of knowledge point using Rasch model with personalization of initial value which optimizes the knowledge level initial value of knowledge point by adopting pretest data.Second,there are two visualization views of learner model.One view is using combination of mind map and stars to release learner model to the learners,in order to stimulate learners to use metacognitive skills of self reflection and learning planning,and induce meta-cognitive learning experience.The other view is using combination of tabular form and stars to release learner models of the learner and peer learners to the learner,which induces meta-cognitive learning experience on self reflection and learning planning,and trigger social-comparison learning experience on comparison method and self-improvement.Third,system-driven and learner-driven teaching strategies can lead learners to meta-cognitive learning experience on self reflection and seeking help learning.