A model for new media data mining and analysis in online English teaching using long short-term memory (LSTM) network

Area of Science:

• Educational Technology
• Machine Learning
• Data Science

Background:

• Effective teacher-student relationships require understanding student progress.
• Current methods for monitoring learning status can be time-consuming and lack depth.
• Technology offers potential for objective and insightful learning analytics.

Purpose of the Study:

• To develop an intelligent classification approach for analyzing student learning time series data.
• To leverage deep learning, specifically Long Short-Term Memory (LSTM) networks, for recognizing student learning states.
• To provide educators with a tool for gaining an intelligent grasp of students' learning status for targeted teaching improvements.

Main Methods:

• Data collection from learners using software via a network platform.
• Feature engineering incorporating user learning characteristics, screen lighting time, inertial sensor data, and network strength.
• Implementation of a deep network model based on Long Short-Term Memory (LSTM) for time series classification.

Main Results:

• The proposed LSTM model achieved highly precise time series recognition.
• The model demonstrated accuracy exceeding 95% in identifying learning states.
• The approach proved effective using relatively straightforward features.

Conclusions:

• The intelligent classification model based on LSTM shows significant promise for learning state recognition.
• High accuracy in recognizing learning states can aid teachers in understanding student progress.
• This technology facilitates targeted teaching improvements and enhances the teacher-student relationship.