SESR: SEMANTIC ENTITY EXTRACTION FOR COMPUTING SEMANTIC RELATEDNESS

There are many systems for entity extraction which have been designed for different purposes. Here, an entity extraction system will be introduced to compute semantic relatedness. The previous works in this field convert a text to a set of words by different methods, and then compute semantic relatedness between these words of texts. In this paper a different approach will be proposed in which any text converts to a set of semantic entities to compute semantic relatedness. For this aim, it will be shown that using YAGO ontology, as one of the best background knowledge resources, can extract these entities from texts. To do this, a new algorithm for entity disambiguation will be suggested. Comparing the results with the literatures shows that the results of this new approach can be sufficiently reliable.