PARSEME Survey on MWE Resources

PARSEME Survey on MWE Resources Gyri Smørdal Losnegaard1, Federico Sangati2, Carla Parra Escart´ın3, Agata Savary4, Sascha Bargmann5, Johanna Monti6 1University of Bergen, Norway; 2FBK Trento, Italy; 3Hermes Traducciones, Spain; 4Universit´e Franc¸ois Rabelais Tours, France; 5Goethe University Frankfurt, Germany; 6University of Sassari, Italy Gyri.Losnegaard@uib.no, federico.sangati@gmail.com, carla.parra@hermestrans.com, agata.savary@univ-tours.fr, bargmann@em.uni-frankfurt.de, jmonti@uniss.it Abstract This paper summarizes the preliminary results of an ongoing survey on multiword resources carried out within the IC1207 Cost Action PARSEME (PARSing and Multi-word Expressions). Despite the availability of language resource catalogs and the inventory of multi- word datasets on the SIGLEX-MWE website, multiword resources are scattered and difficult to find. In many cases, language resources such as corpora, treebanks, or lexical databases include multiwords as part of their data or take them into account in their annotations. However, these resources need to be centralized to make them accessible. The aim of this survey is to create a portal where researchers can easily find multiword(-aware) language resources for their research. We report on the design of the survey and analyze the data gath- ered so far. We also discuss the problems we have detected upon examination of the data as well as possible ways of enhancing the survey. Keywords: multiword expressions, language resources, language resource infrastructures 1. Introduction Despite the ample progress in Natural Language Processing (NLP) in the past decades, the high degree of ambiguity and fuzziness in human languages remains a challenge. Multi- word expressions (MWEs), which often show idiosyncratic morphological, syntactic, or semantic properties, strongly contribute to that challenge (Sag et al., 2001). Current research on MWEs shows that most proposals still concentrate on the creation of MWE lexicons or the au- tomatic recognition of MWEs in text, whereas only very few address the links between MWEs and a comprehen- sive linguistic analysis of text. The IC1207 COST Action PARSEME (PARSing and Multi-word Expressions)1, an interdisciplinary scientific network devoted to the role of MWEs in parsing, aims to contribute to the further im- provement of NLP in this direction (Savary et al., 2015). In an effort towards consolidating previous and ongoing research, PARSEME is currently conducting a meta survey of language resources (LRs) containing MWEs. Exam- ples of such LRs are monolingual and multilingual lists of MWEs, MWE dictionaries and lexicons, corpora and tree- banks with MWE annotations, and any other type of lexical or linguistic resource with MWEs as part of its inventory. The aim is to provide a portal that gives researchers access to previously unavailable or newly created MWE(-aware) resources. In this paper, we present the preliminary results from this ongoing survey. Section 2 reports on how the survey was designed and conducted. Sections 3 and 4 analyze the col- lected data. Whereas Section 3 offers a statistical analysis, Section 4 is devoted to a qualitative one, focusing on each LR type separately. Section 5 discusses our preliminary findings and planned future work, and Section 6 concludes the paper. 1http://www.parseme.eu/index.php/ the-action 2. Methodology 2.1. Survey Design The survey (PARSEME, 2014c) was designed using Google Forms …

Content truncated to 500 words