A Question Answer Approach to Building Semantic Memory

Semantic memory is an integral part of intelligent systems dealing with natural language processing (NLP). Building these memories is a challenging task. Different approaches have been proposed and tested, using a variety of corpuses. The corpora used to build the semantic memories vary from well structured to highly unstructured. The more structured a corpus, the easier it is to build a semantic memory using it. This is because a structured corpus delivers the NLP system more knowledge about the language and its grammar. In this paper we show how a question answering based approach can be used in learning of concepts and building the semantic memory.