A WordNet-based semantic similarity measure enhanced by Internet-based knowledge

Approaches for measuring semantic similarity between words have been widely employed in various areas such as Artificial Intelligence, Linguistics, Cognitive Science and Knowledge Engineering. A new semantic similarity measure is proposed in this paper, which exploits the knowledge retrieved from a semantic network (i.e., WordNet) and the Internet. In particular, the structure information from WordNet and the statistic information obtained from the Internet are combined to quantify the semantic similarity between words. The new similarity measure is evaluated by comparing the rating results with two sets of human benchmark data. Experimental results indicate that, the proposed similarity measure outperforms previous WordNet-based semantic similarity measures.