relin.letras.ufmg.br. Paper Website abstract bibtex
Referring expressions and their textual design are fundamental factors in discourse interpretation. Computational approaches for interpreting information that is contained in textual bases find in anaphoric and correference processes a great challenge. Recent work shows the use of sophisticated techniques for the discovery of anaphoric relations in texts. This kind of research requires textual data bases where anaphoric relations are identified in an accessible way for computer systems. Building such data bases, also called annotated corpora, is very important for the development of research in this area, considering the Portuguese language. Studies in the area show that semantic and pragmatic knowledge are used not as frequently as other types of anaphoric relations. These are, however, the most difficult cases to solve computationally and recent research in the area show that currently available resources, although very sophisticated, are still not sufficient to deal with the problem in a satisfactory way. On the other hand the importance of anaphora resolution is recognized as essential for other natural language tasks, in this paper we see in particular the case of summarization.