AMORE-UPF at SemEval-2018 Task 4: BiLSTM with Entity Library
This work addresses the problem of learning from sparse or unbalanced data in NLP tasks, with potential broad relevance, though it is incremental as it builds on a standard model.
The authors tackled character identification in multiparty dialogues by introducing an entity library to a standard BiLSTM model, which significantly improved identification of infrequent characters, achieving winning results in SemEval 2018 Task 4.
This paper describes our winning contribution to SemEval 2018 Task 4: Character Identification on Multiparty Dialogues. It is a simple, standard model with one key innovation, an entity library. Our results show that this innovation greatly facilitates the identification of infrequent characters. Because of the generic nature of our model, this finding is potentially relevant to any task that requires effective learning from sparse or unbalanced data.