Knowledge Representation Issues in Information Extraction

The advent of computing has exacerbated the problem of overwhelming information. Advanced information management strategies such as Information Extraction, Information Filtering, Information Retrieval, and Text Categorization are becoming important to manage the deluge of information. Information Extraction (IE) systems can be used to automatically extract relevant information from free-form text for update to databases or for report generation. This paper describes the major challenge of knowledge representation issues in an information extraction task – representing the meaning of the input text, the knowledge of the field of application (or domain application) and the knowledge about the target information to be extracted. In this research, we have chosen a directed graph structure to represent the input text meaning, a domain ontology to represent the domain application and a frame representation to capture the target information to be extracted. We discuss in this paper how these knowledge structures interplay to perform the task of information extraction.