Indian Journal of Science and Technology

Abstract

The contemporary research in text or document mining is discerning towards syntactic components and semantic environment. In order to this and with the motivation gained from our earlier research contributions, here we explored a mining model to classify documents by Order of Context, Concept and Semantic Relations (OCCSR). This proposed model is classifying the documents in three levels and they are by context, by concept and semantic. The document context is defined through the document meta-data, the concept is defined by the order of features and semantic relations are assessed through the correlation off the activities observed in documents. The experimental results explored are indicating that the OCCSR is with high classification accuracy, scalable and robust. The research findings lead us to conclude that the context similarity along with concept and semantic similarity score is more significant to achieve classification accuracy in supervised learning. Assessment of the OCCSR is done by using confusion matrix and discriminator metrics. The model devised here is most useful, in particular to assess relation of the documents published in social communities like electronic journals, publishers and blogs.
Keywords: Concept Relations, Context Relations, Document Classification, Feature Selection, Semantic Relations, Supervised Learning, Text Mining, OCCSR