The first step of any terminological work is to setup a reliable, specialized corpus composed of documents written by specialists and then to apply automatic term extraction (ATE) methods to this corpus in order to retrieve a first list of potential terms. In this paper, the experiment we describe differs quite drastically from this usual process since we are applying ATE to unspecialized corpora. The corpus used for this study was built from newspaper articles retrieved from the Web using a short list of keywords. The general intuition on which this research is based is that ATE based corpus comparison techniques can be used to capture both similarities and dissimilarities between corpora. The former are exploited through a termhood measure and the latter through word embeddings. Our initial results were validated manually and show that combining a traditional ATE method that focuses on dissimilarities between corpora to newer methods that exploit similarities (more specifically distributional features of candidates) leads to promising results.

The TermEval 2020 shared task provided a platform for researchers to work on automatic term extraction (ATE) with the same dataset: the Annotated Corpora for Term Extraction Research (ACTER). The dataset covers three languages (English, French, and Dutch) and four domains, of which the domain of heart failure was kept as a held-out test set on which final f1-scores were calculated. The aim was to provide a large, transparent, qualitatively annotated, and diverse dataset to the ATE research community, with the goal of promoting comparative research and thus identifying strengths and weaknesses of various state-of-the-art methodologies. The results show a lot of variation between different systems and illustrate how some methodologies reach higher precision or recall, how different systems extract different types of terms, how some are exceptionally good at finding rare terms, or are less impacted by term length. The current contribution offers an overview of the shared task with a comparative evaluation, which complements the individual papers by all participants.