The impact of word sense disambiguation on stock price prediction
作者：
Highlights：
• Event-based stock price prediction systems typically ignore word meaning in text.
• We assess word sense disambiguation merits in event-based stock price prediction.
• Word sense disambiguation reduces the number of detected events with over 30%.
• There are less noisy price movement predictions for buy/sell signal generation.
• Modest improvements in buy/sell signal precision yield 70% increased excess returns.
摘要
•Event-based stock price prediction systems typically ignore word meaning in text.•We assess word sense disambiguation merits in event-based stock price prediction.•Word sense disambiguation reduces the number of detected events with over 30%.•There are less noisy price movement predictions for buy/sell signal generation.•Modest improvements in buy/sell signal precision yield 70% increased excess returns.
论文关键词：Stock price prediction,Event detection,Word sense disambiguation,Natural language processing
论文评审过程：Received 9 February 2020, Revised 23 August 2020, Accepted 4 July 2021, Available online 9 July 2021, Version of Record 13 July 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.115568
