This purpose of this research is to explore soft information contained within the text of the earnings conference call of Facebook.
Purpose:The rationale of this exploratory study is to contribute to the literature on textual analysis by expanding the quantitative and qualitative possibilities of sensemaking. By using the Natural Language Processing (NLP) IBM Watson™ platform, this paper expands the popular semantic classification of text (positive, negative, and neutral) to emotional (and linguistic) outputs typically associated with psychology (e.g. Ekman, 1992). Design/methodology/approach: By calculating (emotional and linguistic) tone potency, at sentence level, the main argument of this research is that the IBM Watson™ tool presents an opportunity to provide an additional layer of textual understanding to publicly traded companies’ earnings conference calls and the information signals from management.
Finding: This paper is exploratory and part of a wider research project which aims to contribute to the literature on machine learning (NLP, semantic techniques, etc.), investment management (financial forecasting), linguistics and behavioural economics (psychology taxonomy of emotion). Research limitations/implications: Many contributors have shown that textual analysis can be corpus dependent and imprecise. Therefore, any performance signals extracted, through the exploration of corporate narratives, should be measured with caution. Practical implications: Notwithstanding these challenges, this research attempts to contribute a technique, in an attempt, to unbundle the contextual polarity of soft information contained with the earnings conference call of public traded companies’ in general, and for this research project, that of Facebook Inc.
Keywords: Machine learning, Investment Management, Linguistics, Emotion.
