Semantic Indexing For Question Answering System

Abstract

With the vast growth of various forms of digital data, automated indexing has become very important so that it enables the needs of the current users to be fulfilled. Keywords based indexing has failed to accommodate to the needs of the present demands. The representation of the document content as well as the indexing process is a crucial factor that ensures the success of retrieval process. Therefore, this research introduces a new approach in creating semantic indexing that uses Skolem representation which automatically indexes multiple documents into a single knowledge representation. This knowledge representation will then be used by the proposed question answering system in retrieving the answers as well as pointing to the documents the answer contains based on the user’s query. The system managed to achieve 93.84% of recall and 82.92% of precision.