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 discovery. It was ranked first on the 3 sub-tasks for which we submitted results. 1. Create the empty set Q, which will contain an extended set of queries.

We describe the systems developed by the National Research Council Canada for the Cuneiform Language Identification (CLI) shared task at the 2019 VarDial evaluation campaign. We compare a state-of-the-art baseline relying on character n-grams and a traditional statistical classifier, a voting ensemble of classifiers, and a deep learning approach using a Transformer network. We describe how these systems were trained, and analyze the impact of some preprocessing and model estimation decisions. The deep neural network achieved 77% accuracy on the test data, which turned out to be the best performance at the CLI evaluation, establishing a new state-ofthe-art for cuneiform language identification.

We describe the National Research Council Canada team's submissions to the parallel corpus filtering task at the Fourth Conference on Machine Translation.

In this paper, we describe a methodology used to create a test corpus for the evaluation of term extractors. This methodology relies on term annotation: terms in a corpus on automotive engineering are selected based on specific criteria pertaining to the terminological setting as well as linguistic and formal properties of terms and term variations. The test corpus accounts for the variety of ways in which terms are realized in running text, and provides a means of automatically evaluating the relevance of term candidate lists produced by term extractors. Due to the XML annotation scheme used, the corpus can be customized, e.g. by filtering out some of the annotated terms based on the type of term or term variation, or frequency. In this paper, we focus on the methodological aspects of this work.

No abstract