A Knowledge-aware and Time-sensitive Financial News Recommendation System Based on Firm Relation Derivation

Investors are increasingly accessing media information and constantly updating their expectations of the market, which has led to the media-aware stock movements. Most previous studies have focused on modeling the relationship between media and stock volatility through advanced algorithms to improve prediction. Only a few studies have attempted to identify and recommend influential media news. Here, we argue that the previous recommendation system is not suitable for this financial scenario in two facts. For one thing, media are distinguished by the meaning they convey, and their value needs to be measured by the quality and comprehensiveness of their content, not by user interest. Second, the recommended media needs to be selected based on the time-sensitive and firm-related characteristics of the stock market. To address these issues, a financial news recommendation system is proposed to accommodate media-aware stock movements. Specifically, a knowledge-aware method is utilized to capture media news containing key content, and a finance-oriented mechanism is further proposed to recommend news articles that incorporate firm relationships and are within their impact time. Experiments performed on real datasets demonstrate that the proposed method is beneficial for financial news recommendation applications.