Ontology enrichment with causation relations

Ontology learning is considered a potential approach that can help to reduce the bottleneck of knowledge acquisition. However it suffers from a lack of standards to define concepts, besides the lack of fully automatic knowledge acquisition methods. In performing this learning process, the discovery of non-taxonomic relationships has been identified as being the most difficult. This study is then an attempt to create an enhanced framework for discovering and classifying ontological relationships by using a machine learning strategy. We take into consideration the context of the input text in performing the classification of the semantic relations, in particular, causation relations. The proposed framework extracts initial semantic patterns for causation relation from the input samples, then filters these patterns using two novel algorithms, namely, the “Purpose Based Word Sense Disambiguation” which helps in determining the causation senses for input pair of words and the “Graph Based Semantics” which determines the existence of the causation relations in the sentence and to extract their cause-effect parts. The results show a good performance and the implemented framework cut off many steps of the usual process to produce the final results.