Agents for Intelligent Information Extraction By Using Domain Knowledge and Token-based Morphological Patterns

Abstract

Knowledge-based information extraction is known to have flexibility in recognizing various kinds of target information by exploiting the domain knowledge to automatically generate information-extraction rules. However, most of previous knowledge-based information-extraction systems are only applicable to labeled documents, and as a result, ontology terms must appear in the document in order to guide the system to determine the existence of the target information. To make a knowledge-based information-extraction system to be more general enough to handle both labeled and unlabeled documents, this paper proposes an enhanced scheme of knowledge-based wrapper generation by using token-based morphological patterns. Each document is represented as a sequence of tokens, rather than a sequence of logical lines, in order to capture the meaning of data fragments more correctly and recognize the target information contextually. The newly implemented system XTROS+ is presented and its performance is demonstrated.