Quality Classification of Thai Wikipedia Articles using Ontology and Reference Feature Sets

Abstract

Wikipedia articles are widely used as references sources in documents and other medias. Since the framework of Wikipedia allows user to create and edit articles, therefore the quality is the main concern. We found that there is small number of good quality in Thai articles, therefore this research proposes a method to classify Thai Wikipedia articles into high and low quality. The hypothesis is that the high quality articles should contain content that covers most concepts in the domain and should be reliable. We investigated ontology with various machine learning methods and found that using ontology and decision tree with reference features provides promising result measured in term of F-Measure which is 0.73 in biography domain, 0.89 in animal domain, and 0.72 in place domain.