A use of Text Mining: give meaning to customer knowledge

Abstract

Data Mining technologies have enhanced management research’s predictive capability.
In recent years, many improvements have been made, among others by incorporating nonstructured
data to traditional models. This is an important challenge as non-structured data
accounts for more than 80% of an organization’s knowledge. Text Mining allows researchers
to use this type of data to optimize decision making processes. The goal of this paper is to
describe Text Mining implementation and its contribution to management, in other words,
the way non-structured data’s integration to traditional Data Mining models can optimize
the predictive outcome of such analysis. The added-value of Text Mining is demonstrated as
follows: first we show that Text Mining allows considerable enrichment of traditional data
mining models through identification and analysis of the most relevant textual data; second,
through showing that the model with textual data over performs other models with structured
data only. We analyze a case in the automotive industry that illustrates how a manufacturer
can anticipate vehicles recall by combining structured and non-structured data, and
avoid consequently the risk for its brand due to a bad crisis management.