Investigation of data mining algorithms for hierarchical classifications

OData support
Supervisor:
Budai Péter István
Department of Control Engineering and Information Technology

People often use hierarchies, most of the time when they would like to find things efficiently in a large data set. It is the same with the patents, where the huge amount of data divided into different logical groups. This classification to groups requires a lot of time because the lot of consideration. People have been trying to speed up this process with computer programs.

The computer programs – which made to do this text classification processes, – analyzes the texts and process them with statistical and mathematical way. I will describe this process with a text classification algorithm.

The text classification schemes currently used are generally categorized only one logic level. In my dissertation I examine computer text classification method what is walking through on the patent classification hierarchy and compare the achieved results with the traditional single-level text classification.

Can be known or hidden logical relationships between certain elements of the hierarchy. These mapping and utilization can increase the accuracy of the categorization. In my thesis I will present a method for this, and also appreciate its effectiveness.

Downloads

Please sign in to download the files of this thesis.