Say It Right: IS Prototype to Enable Evidence-Based Communication Using Big Data
- 2.4k Downloads
Financial investors face an increasing information abundance when making their valuation decisions of financial assets. As Information Systems research demonstrated, valuation not only builds on quantitative facts, but also on qualitative information such as the language used in financial disclosures and the readability of the texts. As an originator of financial disclosures, e.g. a company, it is thus essential to thoughtfully steer the creation of new textual information. While regulators provide guidelines on what content to publish, corporate communication departments can flexibly steer how they communicate. We have developed an IS prototype that accounts for the importance of textual information and provides corporate communications with a decision-support tool to assure a high readability and a positive sentiment. Our IS prototype builds on a two-step process. First, we extract a dictionary with the most relevant words for investors from a large inventory of regulatory filings with Bayesian learning algorithms. Second, we use this dictionary as input for a Microsoft Word add-in that highlights positively or negatively connoted words and suggests alternative words with a more positive investor perception to corporate communications professionals.
KeywordsBayesian learning Capital markets communication Decision-support system Evidence-based communication IS prototype Readability Sentiment
We want to acknowledge the support of the start-up TonalityTech, our research collaboration partner on this IS prototype development project, for the provision of the above screenshot (Fig. 15.2).
- Cornelissen J (2014) Corporate communication: a guide to theory & practice. Sage, LondonGoogle Scholar
- Hastie TJ, Tibshirani RJ, Friedman JH (2013) The elements of statistical learning: data mining, inference, and prediction. Springer, New YorkGoogle Scholar
- Manning CD, Schütze H (1999) Foundations of statistical natural language processing. MIT Press, CambridgeGoogle Scholar
- Pröllochs N, Feuerriegel S, Neumann D (2015) Generating domain-specific dictionaries using bayesian learning. In: 23rd European conference on information systems (ECIS 2015), Münster, GermanyGoogle Scholar