Analysis of Genetic Programming Fusion for Text Classification

Nyamagoudar, Raghavendra (2006) Analysis of Genetic Programming Fusion for Text Classification. Technical Report TR-06-11, Department of Computer Science, Virginia Tech.

Abstract

The underlying motivation behind this project is the analysis of preliminary experimental results that good text classification is possible when documents are noisy or have limited amounts of text (e.g., short metadata records) by fusing or combining multiple sources of evidence through genetic programming (GP) methods. Pilot studies have shown that improvement of classification can be achieved relative to other machine learning approaches if genetic programming (GP) methods are combined with classifiers such as kNN. In this project, we try to understand why the solutions, unveiled by GP, work so well, so we can understand the key factors that contribute to accurate text classification.