Design and Construction of Semantic Document Networks Using Concept Extraction

Abstract

Processing of unstructured documents according to their content is required in many disciplines; e.g., machine translation, text analysis and mining, and information extraction and retrieval. Whilst research in fields like text analysis, conceptualisation, or design of semantic networks progressed crucially over the last years, we still observe gaps between state-of-the-art algorithms to extract concepts from documents and how these concepts are linked effective and efficiently. This paper proposes a framework to store processed documents in a specialised semantic network database to enhance retrieval and analysis of common concepts in documents. We apply natural language reduction to calculate semantic cores for the concept-based indexing of stored documents. The developed prototype demonstrates an advanced document storage as well as a fast (semantical) retrieval of documents based on given key concepts.