TB-BGAT With TinyBERT and BiGRU in Personalized Course Recommendations

TB-BGAT With TinyBERT and BiGRU in Personalized Course Recommendations

Jing Chen, Weiyu Ye
DOI: 10.4018/IJICTE.345358
Article PDF Download
Open access articles are freely available for download

Abstract

Aiming at the problems of inadequate user feature extraction, cold start and sparse data, a personalized course recommendation algorithm that utilizes TB-BGAT is suggested. First, the Tiny Bidirectional Encoder Representation from Transformers (TinyBERT) model is utilized to output character-level word vectors; then, Bidirectional Recurrent Neural Network (BiGRU) model is utilized to obtain the embedded contextual semantic features. Finally, the attention mechanism is utilized to allocate weights to various course features by assigning their importance and to obtain the output results. The results of experiment on the publicly available dataset MOOCs-Course prove that the proposed method improves at least 3.62%, 3.04%, and 3.33% in precision, recall, and F1-score, correspondingly, in contrast to several other state-of-the-art course resource recommendation algorithms. The proposed method can enhance the effectiveness of the course recommendation model, enhance the quality of learners' online learning, and provide good technical support for online education learning platforms.
Article Preview
Top

Collaborative Filtering Recommendation Algorithm

Collaborative filtering algorithm is the most frequently employed algorithm in recommender systems, it is primarily subdivided into user-based collaborative filtering and item-based collaborative filtering (Goldberg et al., 2001). The principle of user-based collaborative filtering is to recommend to the target items that the target user likes but has not bought, whereas the principle of item-based collaborative filtering is to suggest to the target items that are similar to the items the target user likes (Jain et al., 2020; Ajaegbu, 2021). On e-learning websites, the cost of computing the user similarity matrix is typically significantly greater than the cost of calculating the course similarity matrix, making the item-based collaborative filtering recommendation algorithm more efficient.

Li & Ye (2020) modified the conventional recommendation algorithm according to collaborative filtering so that the results of course recommendation are more aligned with the user's interests, thereby significantly enhancing the recommendation’s accuracy and efficiency. Chen et al. (2020) suggested a new course-recommendation algorithm that employs collaborative filtering to facilitate student decision-making. This algorithm employs an enhanced cosine similarity based on a student's previous course selections to enhance the accuracy of the recommendation task and meet the requirements of the users. Wu et al. (2021) proposed a collaborative filtration recommendation model built on deep learning for art and MOOC resources. The model employs embedding vectors that utilizes meta-path context for learning. Second, an attention mechanism is added to promote the interpretability of the recommendation results, and then the relational network information is effectively integrated by suggesting the Laplacian matrix into the a priori distribution of the implicit factor feature matrices. Compared to the conventional approach utilizing scoring matrices, the model utilizing the text-word vector is superior. Li et al. (2020) proposed the Bayesian personalized ranking network (BPRN), which trains to match a user's course preferences with their enrollment history. Experiments indicate that the BPRN framework outperforms contemporary item-based course recommendation methods. Collaborative filtering-based recommendation algorithms are commonly used in recommendation systems, but they encounter several issues. These include cold start, data sparsity, insufficient feature extraction, poor scalability, new item and user problems, user preference drift, and lack of flexibility. Despite improvements, further research is needed to address these challenges and enhance the accuracy and personalization of recommendation algorithms.

Complete Article List

Search this Journal:
Reset
Volume 20: 1 Issue (2024)
Volume 19: 1 Issue (2023)
Volume 18: 3 Issues (2022)
Volume 17: 4 Issues (2021)
Volume 16: 4 Issues (2020)
Volume 15: 4 Issues (2019)
Volume 14: 4 Issues (2018)
Volume 13: 4 Issues (2017)
Volume 12: 4 Issues (2016)
Volume 11: 4 Issues (2015)
Volume 10: 4 Issues (2014)
Volume 9: 4 Issues (2013)
Volume 8: 4 Issues (2012)
Volume 7: 4 Issues (2011)
Volume 6: 4 Issues (2010)
Volume 5: 4 Issues (2009)
Volume 4: 4 Issues (2008)
Volume 3: 4 Issues (2007)
Volume 2: 4 Issues (2006)
Volume 1: 4 Issues (2005)
View Complete Journal Contents Listing