Information Extraction (IE) is a natural language processing (NLP) task that processes unrestricted text and attempts to extract specific types of items from the text. This form of shallow text processing has attracted considerable attention recently with the growing need to intelligently process the huge amounts of information available in the form of text documents. While learning methods have been used earlier to aid in parts of an IE system [Riloff, 1993; Soderland and Lehnert, 1994], it has been argued quite convincingly [Califf and Mooney, 1999; Craven and Slattery, 2001] that relational methods are necessary in order to learn how to directly extract the desired items from documents. The reason is that the target concepts require the representation of relations over the source document and learning those might require induction over structured examples and FOL representations. Indeed, previous works [Califf and Mooney, 1999; Freitag, 2000] have demonstrated the success of ILP methods in this domain. This is therefore an ideal domain to study our proposed relational learning paradigm.