2*n is better than n²: Decomposing Event Coreference Resolution into Two Tractable Problems

Event Coreference Resolution (ECR) is the task of linking mentions of the same event either within or across documents. Most mention pairs are not coreferent, yet many that are coreferent can be identified through simple techniques such as lemma matching of the event triggers or the sentences in which they appear. Existing methods for training coreference systems sample from a largely skewed distribution, making it difficult for the algorithm to learn coreference beyond surface matching. Additionally, these methods are intractable because of the quadratic operations needed. To address these challenges, we break the problem of ECR into two parts: a) a heuristic to efficiently filter out a large number of non-coreferent pairs, and b) a training approach on a balanced set of coreferent and non-coreferent mention pairs. By following this approach, we show that we get comparable results to the state of the art on two popular ECR datasets while significantly reducing compute requirements. We also analyze the mention pairs that are "hard" to accurately classify as coreferent or non-coreferent 1 . Mention Pairs Heuristic Predictions Coreferent Not Coreferent Lemma Heuristic Triggers, Sentences, Synonyms Colts clinch playoff berth with 20 -13 win in K . C . Colts beat Chiefs 20 -13 to clinch playoff berth P + easy Colts clinch playoff berth with 20 -13 win in K . C . Colts clinch playoff spot by beating Jags P - hard Colts clinch playoff spot by beating Jags. Indianapolis made a comeback this season to lock up the five seed in the AFC with this win. P + FN Colts clinch playoff berth with 20 -13 win in K . C . Game slips away late as Jaguars fall to Indy.