CRIM at SemEval-2018 Task 9: A Hybrid Approach to Hypernym Discovery

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 725–731 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics CRIM at SemEval-2018 Task 9: A Hybrid Approach to Hypernym Discovery Gabriel Bernier-Colborne g.b.colborne@gmail.com Caroline Barri`ere caroline barriere@yahoo.ca Abstract This report describes the system developed by the CRIM team for the hypernym discovery task at SemEval 2018. This system exploits a combination of supervised projection learning and unsupervised pattern-based hypernym dis- covery. It was ranked first on the 3 sub-tasks for which we submitted results. 1 Introduction The goal of the hypernym discovery task at Sem- Eval 2018 is to predict the hypernyms of a query given a large vocabulary of candidate hypernyms. A query can be either a concept (e.g. cocktail or epistemology) or a named entity (e.g. Craig An- derson or City of Whitehorse). Two types of data were provided to train the systems: a large unla- beled text corpus and a small training set of exam- ples comprising a query and its hypernyms. More details on this task may be found in the task de- scription paper (Camacho-Collados et al., 2018). The system developed by the CRIM team for the task of hypernym discovery exploits a com- bination of two approaches: an unsupervised, pattern-based approach and a supervised, projec- tion learning approach. These two approaches are described in Sections 2 and 3, then Section 4 de- scribes our hybrid system and Section 5 presents our results. 2 Pattern-Based Hypernym Discovery Pattern-based approaches to relation extraction have been discussed in the literature for quite some time (see surveys by Auger and Barri`ere (2008) and Nastase et al. (2013)). They can be used to dis- cover various relations, including domain-specific ones (Halskov and Barri`ere, 2008) and more gen- eral ones, such as hypernymy. The pattern- based approach to hypernym discovery was pi- oneered by Hearst (1992), who defined specific textual patterns (e.g. Y such as X) to mine hy- ponym/hypernym pairs from corpora. This approach is known to suffer from low re- call because it assumes that hyponym/hypernym pairs will occur together in one of these patterns, which is often not the case. For instance, using the training data of sub-task 1A, we found that the ma- jority of training pairs never co-occur within the same paragraph in corpus 1A, let alone within a pattern that suggests hypernymy. To increase recall, we extend the basic pattern- based approach to hypernym discovery in two ways. First, we identify co-hyponyms for each query and add the hypernyms discovered for these terms to those found for the query. These co- hyponyms are identified using patterns, and fil- tered based on distributional similarity using the embeddings described in Section 3.3. Further- more, we discover additional hypernyms using a method based on the following assumptions: most multi-word expressions are compositional, and the prevailing head-modifier relation is hypernymy. The co-hyponym patterns we use are limited to enumeration patterns (e.g. X1, X2 and X3). For hypernyms, we use an extended set of Hearst-like …

Content truncated to 500 words