Experiment 0 (baseline): Linguistic
We began with the constraint and preference set used by Ferrández
et al. 2000. This algorithm is based on linguistic information
only, and its results have been successfully tested over a non-dialogue
corpus with a resulting precision of 82% for pronominal anaphora resolution
(we have no information about the precision for adjectival anaphora).
The initial configuration included the following constraint and preference
set and definition of anaphoric accessibility space.