Automated text classification for fast feedback -investigating the effects of document representation

Abstract

New trends such as increased product complexity, changing customer requirements and shortening development time, have given rise to an increase in the number of unexpected events within the Product Development Process (PDP). Traditional tools are only partially adequate (either insufficient coverage or simply too late) to cover these unexpected events. As such, new tools are being sought to complement traditional ones. This paper investigates the use of one such tool, textual data mining, for the purpose of facilitating fast feedback. The motivation for this paper stems from the need to handle widely ignored and "loosely structured textual data" within the PDP. In particular this study would focus on the automated classification of call center records from a Multi National Company (MNC). Different document representation schemes are studied in view of determining the most appropriate scheme that maximizes classification accuracy.