One of the main problems in anaphora resolution is how to find the correct antecedent for each anaphoric expression that appears in the discourse. As shown in a previous study (Rico, 1993), the type of knowledge required to select an antecedent comprises different sources of linguistic information (morphology, syntax, semantics and pragmatics) and it is only through the coordination of all these sources that we will be able to approach anaphora from an effective point of view. The strategy for anaphora resolution that I present here allows the integration of the mentioned sources by means of a vector-like model. According to it, I consider that each anaphoric expression and discourse entity is a vector and therefore, can be subject to vector comparison techniques in such a way that closer vectors will indicate closer anaphoric links.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados