Text Data Pre-Processing for Time-series Modelling

The paper proposes an algorithm of text data representation for time-series econometric modeling. We link two areas in empirical econometrics, text mining and time-series modeling, and provide a multi-step methodology for processing qualitative text data for quantitative time-series models. We also present an empirical example of text data processing of research article meta-data in economics and show its applicability to sentiment analysis using annual data from 1933-2015