An Intelligent Recommendation of Chinese Traditional Culture Learning Resources Based on DistilBERT with Collaborative Filtering

Chinese traditional culture is a precious cultural heritage, so, it is essential to revive the traditional culture resources in today's era. This heritage holds significant guidance for the direction, approach, and specific content of socialist cultural development with Chinese characteristics. However, the existing Deep Learning (DL) based Long Short Term Memory based Recurrent Neural Network (LSTM-RNN) model neglected student's individuality to recommend culture resources. Therefore, this research presents an intelligent recommendation of Chinese traditional culture learning resources based on Distilled version of Bidirectional Encoder Representations from Transformers (DistilBERT) with Collaborative Filtering (CF) algorithm. Initially, the student data is gathered from various educational institutions. The collected data is further preprocessed with the removal of stop words, and punctuations, and grouped together by Lemmatization. Then, DistilBERT is employed to extract the features and classify the culture learning resources. Finally, CF recommender is demonstrated to recommend the personalized learning resources to students according to the past tastes. As per the results, the proposed DistilBERT-CF model provided outstanding results in accuracy of 95.64% and recall of 90.32% respectively when compared to existing Traditional Cultural Network integrated with and Knowledge Tracking (TCN-KT) model.