Title: FiGAS: Fine-Grained Aspect-based Sentiment analysis on economic and financial lexicon
Authors: Luca Barbaglia - European Commission Joint Research Centre (Italy)
Sergio Consoli - National Research Council of Italy (CNR) (Italy) [presenting]
Sebastiano Manzan - European Commission (Italy)
Abstract: The extraction of sentiment from news, social media and blogs for the prediction of economic and financial variables has attracted great attention in recent years. Despite many successful applications of sentiment analysis (SA) in these domains, the range of semantic techniques employed is still limited and mostly focus on the detection of sentiment at a coarse-grained level, i.e. whether the sentiment expressed by the entire sentence text is either positive or negative. However, coarse-grained methods might not be precise enough in evaluating the sentiment polarity of a specific topic of interest contained in a sentence. For this reason we propose FiGAS, a Fine-Grained Aspect-based Sentiment analysis approach that is able to identify the sentiment associated to specific topics of interest within a text, by assigning real-valued sentiment polarity scores to those topics. The approach is completely unsupervised and customized to the economic and financial domains, being built upon a large specialised lexicon in these areas. We provide a statistical comparison of the performance of FiGAS against other popular lexicon-based SA approaches on a humanly annotated data set in the economic and financial domain. FiGAS outperforms the other methodologies and shows to be a promising alternative to extract sentiment from news.