Semantic Oriented Clustering of Documents

Semantic web-based approaches and computational intelligence can be merged in order to get useful tools for several data mining issues. In this work a web-based tagging process followed by a validation step is carried to tag WordNet adjectives with positive, neutral or negative moods. This tagged WordNet is used to define a semantic metric for text documents clustering. Experimental results on movie reviews prove that the introduced semantically oriented metric is extremely fast and gives improved results with respect to the classical frequency based text mining metric from the accuracy point of view.