Research on Deep Learning Recommendation System Based on Book Scoring
Keyu Xiang · 2024
Traditional recommendation systems, they primarily rely on collaborative filtering, which has rapidly developed and gained widespread adoption. However, these conventional recommendation systems suffer from sparsity, scalability, synonymity, and cold-start issues, etc. Therefore, we compare our model based on traditional recommendation systems, with deep learning and evaluate its fitting time on both training and testing sets. Furthermore, by extracting potential features from project descriptions using deep learning techniques, we calculate the similarities between books through a fully connected layer, forming new matrices, and compare them on the test set. Our deep learning algorithm outperforms all other algorithms in this study.