Collocation resources

The currently available collocation dataset is a list of about 10,000 collocations in English collected and tagged in terms of Lexical Functions (LFs) by I. Mel’čuk. In order to facilitate the use of this dataset in downstream NLP applications, we disambiguated the collocation bases (or “keywords” in the terminology of LFs) with respect to BabelNet synsets.

Please consult  here  the Readme for the precise description of the dataset

A subset of this dataset has been used in the experiments described in L. Espinosa-Anke, L. Wanner, and S. Schockaert. “Collocation Classification with Unsupervised Relation Vectors”, ACL 2019, Short paper track, Florence, Italy

Files:

•  Collocations Dataset

Files:

•  training data

Links: 

• script that emulates the experiment in T. Milokov's paper "Exploiting Similarities among Languages for Machine Translation" that we used for collocation discovery

A dataset of comparable size on French is in preparation and will be published soon.