Conversational ontology construction based on human emotions

In this paper we describe a solution for incorporating human emotions into the conversational system for automatic ontology construction. We find semantic similarity using LSA, Latent Semantic Analysis, between the new sentence and the corpus corresponding to each human emotion stored in ontology, structured database of sentences corresponding to human emotions. If the semantic similarity of the new sentence and the corpus has a highest value then new sentence will be inserted as a child of that corpus on the ontology and the emotion of the new sentence is the emotion of that corpus but the degree of the emotion depends on the percentage of the semantic value. We analyzed the 100 sentences and we got the estimation accuracy of about 75%.