In-Depth Analysis of Anaphora Resolution Requirements

This paper aims to lay the foundations of an anaphora resolution framework able to process all types of hypertexts and treat all types of anaphors for the English language. To this end, we provide a linguistically unambiguous and extensive definition and categorization of the concept of anaphora. We introduce a new corpus, and use our proposed categorization to statistically analyze it. Finally, we describe a preliminary version of our framework and outline promising results of the first experimental evaluations.