Knowledge Discovery in Unstructured Data: An analysis in achat of online game

Abstract

This paper presents a study on knowledge discovery applied in a chat of a massively multiplayer online game
(MMO). In this study, the tool was used Text Mining Suite, where the process of knowledge discovery, besides
the mining of texts, is based on the creation of concepts defined by a person who has some understanding about
the texts to be analyzed. Also, as a complementary way of research, Sobek text mining tool was used as a support
to the Text Minig Suite for the creation of concepts. This paper proposes an experimental methodology for the
application of the knowledge discovery in textual data generated from the conversations of the players. With this
methodology, we evaluated the results of three tests. The frst test involved Text Mining Suite tool to extract
knowledge from the textual database; the second one was used Sobek tool to extract the concepts from the text;
finally, the last test added the two tools. A quantitative analysis made possible the presentation of the quality of
the results of the mining of texts. We observed an improvement in the results, a gain of knowledge when the
tools Text Mining Suite and Sobek were used together for the process of the knowledge discovery.