CMStatistics 2021: Start Registration
View Submission - CFE
Title: Sentiment analysis of economic text: A lexicon-based approach Authors:  Elisa Tosetti - University of Venice (Italy) [presenting]
Luca Tiozzo Pezzoli - JRC European Commission (Italy)
Luca Barbaglia - European Commission Joint Research Centre (Italy)
Sergio Consoli - Joint Research Centre (JRC) (Italy)
Sebastiano Manzan - City University of New York (United States)
Abstract: With the increasing availability of opinion-rich web resources, such as news, discussion forum and personal blogs, a growing body of research in economics and finance focuses on constructing sentiment indicators from these sources with the aim to predict in a timely manner economic and financial developments. Several studies automatically determine the sentiment of a piece of text by looking at how many positive and negative words can be found in the text according to a predefined lexicon. While this approach is very popular, little work has been done to develop a lexicon for calculating sentiment scores specifically for text with economic content. We fill this gap by proposing a domain-specific lexicon suitable for applications in the area of economics and finance. We do this by carrying a semantic analysis of the language used in economic news and extracting the set of most frequent terms used to describe economic concepts. We use the proposed lexicon in a small empirical study to investigate to what extent the forecasting power of a regression model for predicting stock returns can be improved by incorporating sentiment extracted from economic text.