G. Ramirez

Processing Overlaps


In semi-structured text retrieval, processing overlap techniques are used to reduce the amount of overlapping (thus redundant) information returned to the user. The existence of redundant information in result lists is caused by the nested structure of semi-structured documents, where the same text fragment may appear in several of the marked up elements. In consequence, when retrieval systems perform a focused search on this type of documents and use the marked up elements as retrieval objects, very often result lists contain overlapping elements. In retrieval applications where it is assumed that the user does not want to see the same information twice, it may be needed to reduce or completely remove this overlap and return a ranked list of no overlapping elements. Thus, depending on the underlying user model and retrieval application, different processing overlap techniques are used in order to decide, given a set of relevant but overlapping elements, what are the most appropriate elements to return to the user.