Examining text categorization methods for incidents analysis

Abstract

Text mining saves the necessity to sift through vast amount of documents manually to find relevant information. This paper focuses on text categorization, one of the tasks under text mining. This paper introduces fuzzy grammar as a technique for building text classifier and investigates the performance of fuzzy grammar against other machine learning methods such as decision table, support vector machine, statistic, nearest neighbor and boosting. Incidents dataset was used where the focus was given on classifying the incidents events. Results have shown that fuzzy grammar has gotten promising results among the other benchmark machine learning methods. © 2012 Springer-Verlag.