Mining the Factors Affecting the Quality of Wikipedia Articles

In order to observe the variation of factors affecting the quality of Wikipedia articles during the information quality improvement process, we proposed 28 metrics from four aspects, including lingual, structural, historical and reputational features, and then weighted each metrics indifferent stages by using neural network. We found lingual features weighted more in the lower quality stages, and structural features, along with historical features, became more important while article quality improved. However, reputational features did not act as important as expected. The findings indicate that the information quality is mainly affected by completeness, and well-written is a basic requirement in the initial stage. Reputation of authors or editors is not so important in Wikipedia because of its horizontal structure.