Enhancing Anaphora Resolution for Czech

Warning

This publication doesn't include Faculty of Arts. It includes Faculty of Informatics. Official publication website can be found on muni.cz.
Authors

NĚMČÍK Václav

Year of publication 2007
Type Article in Proceedings
Conference RASLAN 2007
MU Faculty or unit

Faculty of Informatics

Citation
Web https://nlp.fi.muni.cz/raslan/2007/papers/4.pdf
Field Use of computers, robotics and its application
Keywords anaphora resolution; linguistic resources; verb valency; semantic plausibility; WordNet; Czech
Description Resolution of anaphoric reference is one of the most important challenges in natural language processing (NLP). Functionality of most NLP systems crucially relies on an accurate mechanism for determining which expressions in the input refer to the same entity in the real world. The immense complexity of this issue has led the research community to adopt predominantly knowledge-poor methods, despite the fact that these are known to be incapable of solving this task reliably. This paper suggests several ways of extending such methods by further resources and mechanisms in order to arrive at a more adequate anaphora resolution procedure. First, the paper sketches how to exploit information about verb valencies or co-occurrence statistics to account for semantic plausibility of individual antecedent candidates. Further, several ways of adapting ML-based AR methods are suggested, so that they account for the structure of the AR task more closely.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.