Combining Enterprise Knowledge Graph and News Sentiment Analysis for Stock Price Prediction

Abstract

Many state of the art methods analyze sentiments in news to predict stock price. When predicting stock price movement, the correlation between stocks is a factor that can't be ignored because correlated stocks could cause co-movement. Traditional methods of measuring the correlation between stocks are mostly based on the similarity between corresponding stock price data, while ignoring the business relationships between companies, such as shareholding, cooperation and supply-customer relationships. To solve this problem, this paper proposes a new method to calculate the correlation by using the enterprise knowledge graph embedding that systematically considers various types of relationships between listed stocks. Further, we employ Gated Recurrent Unit (GRU) model to combine the correlated stocks' news sentiment, the focal stock's news sentiment and the focal stock's quantitative features to predict the focal stock's price movement. Results show that our method has an improvement of 8.1% compared with the traditional method.

Domain-specific knowledge graphs: A survey

“…In particular, the constructed KG (named FinKG) and the implemented embeddings performs learn informative representations based on both the relations of event argument and the lead-lag relations amongst the entire KG. Liu et al [143] demonstrated the use of a KG embedding framework to predict stock prices using news sentiment analysis. Although the authors did not provide much discussion on the validity of the mechanism followed to construct the KG, the utility was demonstrated in the prediction task.…”

Stock Market Analysis and Prediction

“…However, few consider inter-market or inter-company relations in the current financial market environment. Liu et al [10] comprehensively use the news sentiment as the correlation between stock companies, accompanied by stock price data, to predict the future trend. Utilizing the knowledge graph and its embedding model, they find that their method can improve significantly over baseline.…”

Deep Attentive Learning for Stock Movement Prediction From Social Media Text and Company Correlations

“…Despite the popularity of NLP and graph-based stock prediction, multimodal methods that capture inter stock relations and market sentiment through linguistic cues are seldom explored. Jue Liu (2019) combines feature extraction from news sentiment scores, financial information (price-earnings ratio, etc.) along with knowledge graph embeddings through TransR.…”