UTU at SemEval-2016 Task 10: Binary Classification for Expression Detection (BCED)

The SemEval 2016 DiMSUM Shared Task concerns the detection of minimal semantic units from text and prediction of their coarse lexical categories known as supersenses. Our approach is to define this task as a binary classification problem approachable by straightforward machine learning methods. We start by detecting semantic units by matching text spans against several large dictionaries, including the English WordNet, expressions derived from the Yelp Academic Dataset and concepts from the English Wikipedia, generating a set of potential supersenses for each matched span. For each potential supersense and text span pair a binary machine learning example is defined. We classify these examples using an ensemble method, taking as the final predicted supersense the one with the highest confidence score. Our system achieves good performance on the supersense classification task but has limited performance for detection of multi-word semantic units. We show that the task of supersense prediction can be effectively defined as a binary classification task.