Introduction to the Special Issue on Man aging Information Extraction

The field of information extraction (IE) focuses on extracting structured data, such as person names and organizations, from unstructured text. This field has had a long history. It attracted steady attention in the 80s and 90s, largely in the AI community. In the past decade, however, spurred on by the explosion of unstructured data on the World-Wide Web, this attention has turned into a torrent, gathering the efforts of researchers in the AI, DB, WWW, KDD, Semantic Web and IR communities. New IE problems have been identified, new IE techniques developed, many workshops organized, tutorials presented, companies founded, academic and industrial products deployed, and opensource prototypes developed (e.g., [5, 4, 3, 1, 2]; see [5] for the latest survey). The next few years are poised to witness even more accelerated activities in these areas. It is against this vibrant backdrop that we assemble this special issue. Our objective is threefold. First, we want to provide a glimpse into the current state of the field, highlighting in particular the wide range of IE problems. Second, we want to show that many IE problems can significantly benefit from the wealth of work on managing structured data in the database community. We believe therefore that our community can make a substantial contribution to the IE field. Finally, we hope that examining IE problems can in turn help us gain valuable insights into managing data in this Internet-centric world, a long-term goal of our community. Keeping in mind the above goals, we end this introduction by briefly describing the nine papers assembled for the issue. These papers fall into four broad categories.