Detecting Bridge Anaphora

The paper presents one of most important issues in natural language processing (NLP), namely the automated recognition of semantic relations (in this case, bridge anaphora). In this sense, we propose to recognize automatically, as accurately as possible, this type of relations in a literary corpus (the novel Quo Vadis), knowing that the diversity and complexity of relations between entities is impressive. Furthermore, we defined and classified the bridge anaphora type relations based on annotation conventions. In order to achieve the main goal, we developed a computational instrument, BAT (Bridge Anaphora Tool), currently still in a test (and implicitly an improvable) version. This study is intended to help especially specialists and researchers in the field of natural language processing, linguists, but not only.