Dynamic Industry-Specific Lexicon Generation for Stock Market Forecast.

摘要: Press releases represent a valuable resource for financial trading and have long been exploited by researchers the development of automatic stock price predictors. We hereby propose an NLP-based approach to generate industry-specific lexicons from news documents, with goal dynamically capturing, on daily basis, correlation between words used in these documents fluctuations. Furthermore, we design binary classification algorithm that leverages our predict magnitude future changes, individual companies. Then, validate through experimental study conducted three different industries Standard & Poor’s 500 index, processing press published globally renowned sources, collected within Dow Jones DNA dataset. Classification results let us quantify mutual dependence prices, help estimate predictive power lexicons.