Calculating Concept Similarity Heuristics For Ontology Learning from Text

We present experimental results of an approach to learning ontological concepts from text. The ontologicalsemantic analyzer OntoSem and its knowledge resources – in particular, its NLP-oriented ontology and semantic lexicon – are used to dynamically create the feature values on which our learning approach is based. We expand upon our previously reported work, with emphasis given to development of a new metric for calculating similarity between two ontological concepts. The specific use for this metric in our approach is to compare an automatically generated candidate ontological concept with concepts already in the ontology, to find the best position for a new concept in the inheritance network of the ontology. Our longterm goal of bridging the knowledge acquisition bottleneck through “learning by reading” is assisted in this manner by facilitating the placement of acquired ontological concepts into an existing ontology.