Predicting Economic Indicators from Web Text Using Sentiment Composition
Abby Levenberg, Stephen Pulman, Karo Moilanen, Edwin Simpson, Stephen John Roberts · International Journal of Computer and Communication Engineering · 2014
Of late there has been a significant amount of work on us-ing sources of text data from the Web (such as Twitter or Google Trends) to predict financial and economic variables of interest. Much of this work has relied on some form or other of superficial sentiment analysis to represent the text. In this work we present a novel approach to predict-ing economic variables using sentiment composition over text streams of Web data. We treat each text stream as a separate sentiment source with its own predictive distribu-tion. We then use a Bayesian classifier combination model to combine the separate predictions into a single optimal prediction for the Nonfarm Payroll index, a primary eco-nomic indicator. Our results show that we can achieve high predictive accuracy using sentiment over big text streams.