Author Extraction: A Test Experience for Flexible Information Extraction

摘要

This paper presents an experience in the area of information extraction within the framework of the on-going research ESPRIT IV project FLEX. The work presented is part of a larger effort aimed at building the first prototype of a flexible information system and it has consisted in the detection and extraction of named entities from a collection of newspaper articles. Although name extraction has received the attention of the Information Extraction research community from the beginning, this attention has been recently increased by the inclusion of a Named Entity task both in the Seventh Message Understanding Conference and the Information Retrieval and Extraction Exercise. The work presented is a practical application of techniques discussed and developed in recent years by the Information Extraction community. Besides that practical application, the results obtained serve for validating our overall approach to cost-effective, re-usable information extraction in the context of a viable, marketable system.