Text data analysis has found its way in many applications, and our study focuses on the financial fields. Previous studies in financial indicator prediction are mostly based on econometric models. In recent years, with the advance of text mining techniques, more and more studies employ financial news as the data source for analysis. Most studies, however, aim to predict stock prices, identify the trend of stock market, and detect company bankruptcy or company fraud. We observe that company’ revenue, which can imply the company's cash flow and market share, is indeed an important financial indicator. In our study, we identify a few features that potentially impact company’s revenue and further propose an approach to deriving feature values from financial news data. Specifically, we develop a lexicon-based method that involves the automatic expansion of existing financial sentiment dictionary and the aggregation of sentiment values. Preliminary experimental results show that we are able to predict the revenue trend through the news articles in the last quarter with the accuracy up to 80%.
Hsieh, Wei-Lin; Hwang, San-Yih; Huang, Hsin-Ching; and Chang, Shanlin, "PREDICTING COMPANY REVENUE TREND USING FINANCIAL NEWS" (2016). PACIS 2016 Proceedings. 316.