Text Classification Using Ensemble Features Selection and Data Mining Techniques

Text categorization is a task of text mining/analytics which involves extracting useful information from unstructured resources followed by categorizing these documents. In this paper, we classify the TechTC dataset collected from various Web directories. We employed feature selection methods such as Gini index, chi-square, t-statistic, correlation which drastically reduced the model building time. Various neural network models such as probabilistic neural network, group method of data handling, multi layer perceptron yielded higher accuracies compared to other techniques applied in literature.