Cross-Domain Recommendation Algorithm Based on Sentiment Analysis of Reviews and Early Warning Field Side Information
Chen Jingwen, Liu Chunjia, Li Juan, Lan Zhang · 2024
Collaborative filtering has been successfully used to provide users with personalize products and services. However, data sparseness and cold start issues remain unsolved. Considering the problem of insufficient historical data and strong domainality of Jiangsu's early warning information, a cross-domain recommendation algorithm with deep fusion of side information is proposed. Introducing CICDR framework, the bidirectional GRU model is used to optimize the calculation of user comment sentiment. The bidirectional GRU model is used to optimize the calculation of user comment sentiment. Matrix factorization (MF) is modeled in both source and target domains for more accurate recommendations. Simulation experiments show that the algorithm can effectively improve the recommendation speed and obtain better RMSE and MAE values, which can meet the needs of rapid and accurate recommendation of early warning information.