Brundlefly at SemEval-2016 Task 12: Recurrent Neural Networks vs. Joint Inference for Clinical Temporal Information Extraction

摘要

We submitted two systems to the SemEval-2016 Task 12: Clinical TempEvalchallenge, participating in Phase 1, where we identified text spans of time andevent expressions in clinical notes and Phase 2, where we predicted a relationbetween an event and its parent document creation time. For temporal entity extraction, we find that a joint inference-based approachusing structured prediction outperforms a vanilla recurrent neural network thatincorporates word embeddings trained on a variety of large clinical documentsets. For document creation time relations, we find that a combination of datecanonicalization and distant supervision rules for predicting relations on bothevents and time expressions improves classification, though gains are limited,likely due to the small scale of training data.