Year of Graduation
Extraction of Structured Data from News Reports for Predicting Financial Indicators
Applied Mathematics and Information Science
Stock prediction has always drawn attention among different fields of research for several decades. It has been shown that news reports could influence stock prices. However, most of previous works use news’ text features, encoded with bag-of-words techniques, which could not qualitatively capture underlying information, and hence do not show impressive results. We suggest adapting works from the field of information extraction in order to retrieve structured events from the reports. Then, event representations can be used for stock price movements predictions. Such pipeline, combined with a complex model to investigate hidden relations between events and stock prices, could significantly outperform bag-of-words solutions. Also, we investigate application of aforementioned models to Russian news articles.