Title: The GWP R package: Generalized word power approach of calibrating lexicons
Authors: Keven Bluteau - Institute of Financial Analysis, University of Neuchatel (Switzerland) [presenting]
Abstract: Text analysis tools are becoming increasingly important in financial and economic research as recent research projects focus on textual analysis to answer new hypotheses or increase forecasting power. The standard approach to extract sentiment from textual data is the bag-of-word method, where a list of words and corresponding scores, referred to as a lexicon, are used to quantify textual data to uncover, for example, textual sentiment. This list of words and scores are often manually set up by the researcher in an ad-hoc way. We present the R package GWP which implements the Generalized Word Power approach, a data-driven lexicon calibration framework. The package provides several helper functions to organize textual data as well as a Shiny interface to help validating the calibrated lexicons.