Text mining is a young and innovative subfield of the data mining techniques, and it is specifically designed for unstructured and free-form text data. As a result, it gives a great opportunity to analyse speech recognition outputs. Additionaly, speech recognition and automatic text processing has the same difficulties, because natural languages have formed for the communication between people, not for computer processing. The aim of my thesis is the use of this two technologies support each other. I will analyse a speech recognition output with different text mining techniques, using the SPSS Modeler software and the Text Analytics module. The purpose of the study is to make useful conclusions from a speech recognition output of a call center, about the succesful insurance contracting.
In the fisrt part of my thesis I introduce the theoretical backgrounds, algorithms, and the main concepts of the speech recognition and text mining. In the third, and the fourth part, I describe the steps of the design and implementation of my text mining modell. In the fifth part I make classification and predictive modells using decision trees from the database made by the text mining modell of the fourth part. In the end of my thesis I describe the further possibilites and my conclusions.