This research builds on the quantification of learning processes and the integration of diverse learning data to analyze and extract learner characteristics, identify multidimensional learning states, and construct personalized learner profiles. The aim is to develop stable-state models of learners using multi-source data.

Learner modeling is, by nature, a dynamic and probabilistic process. To build models that reflect key learning features and their temporal and spatial evolution, the study follows a systematic approach: analyzing and extracting learner characteristics → identifying multidimensional learning states → dynamically constructing personalized learner profiles → conducting comprehensive learner profiling and evolutionary analysis.