Text Mining - A Toolbox for Text Classification

Abstract

In this thesis it will be explored in depth the process of text mining and further document classification. The main focus will be the development of a platform capable of achieving op- erations of data extraction, natural language processing, classification of data, and evaluation of constructed models, from a corpus of labeled documents. This will be integrated with a sentiment analysis dataset where the documents are polarity based, classified as positive or negative. It will be made an evaluation of the accuracy in the processing algorithms and an in depth comparison between the different ones used although this process. It was aimed to produce a user-friendly application, capable of providing the user with tools of text mining and predictive analysis with the integration of a polarity dataset.