Sarigiannidis, Antonios; Karypidis, Paris-Alexandros; Sarigiannidis, Panagiotis; Pragidis, Ioannis-Chrysostomos
In: Internet Science, pp. 37–48, Springer International Publishing, Cham, 2018, ISBN: 978-3-319-77547-0.
Sentiment analysis refers to the use of natural language processing (NLP) and textual analysis approaches to identify and extract subjective information from textual sources. Extracting sensible financial knowledge from relevant textual material is significant in order to help leverage the predictive power of the financial and econometric forecasting models. However, the determination of the sentiment analysis from textual data such as headlines, news and user comments is not an easy task. One of the most arduous challenges when dealing with sentiment analysis is the accuracy. In this work, a new lexicon-based approach is presented which is based on supervised learning. The introduced model is able to create a new lexicon based on annotated textual data and then it applies that lexicon to determine the sentiment in new, not-annotated data. The proposed method seems able to work effectively with financial data while supporting accurate decisions.