Nominal Coreference Resolution Using Semantic Knowledge

Coreference Resolution is a challenging task, considering the required linguistic knowledge and the sophistication of language processing techniques involved. Several other Natural Language Processing tasks may benefit from it, such as named entities recognition, relation extraction between named entities, summarization, sentiment analysis, among others. We propose a process for nominal coreference resolution in Portuguese, based on syntactic-semantic linguistic rules. Such rule models have been efficiently applied in other languages, such as: English, Spanish and Galician. They are useful when we deal with less resourceful languages, since the lack of sample-rich corpora may prevent accurate learning. We combine different levels of linguistic processing, using semantic relations as support, in order to infer referential relations between mentions. The proposed approach is the first model for Portuguese coreference resolution which uses semantic knowledge.