CLARIN’s Key Resource Families

CLARIN’s Key Resource Families Darja Fišer1,2, Jakob Lenardič1, Tomaž Erjavec2 Faculty of Arts, University of Ljubljana, Aškerčeva 2, 1000 Ljubljana, Slovenia Jožef Stefan Institute, Jamova cesta 39, 1000 Ljubljana, Slovenia {darja.fiser, jakob.lenardic}@ff.uni-lj.si, tomaz.erjavec@ijs.si Abstract CLARIN is a European Research Infrastructure that has been established to support the accessibility of language resources and technologies to researchers from the Digital Humanities and Social Sciences. This paper presents CLARIN’s Key Resource Families, a new initiative within the infrastructure, the goal of which is to collect and present in a uniform way the most prominent data types in the network of CLARIN consortia that display a high degree of maturity, are available for most EU languages, are a rich source of social and cultural data, and as such are highly relevant for research from a wide range of disciplines and methodological approaches in the Digital Humanities and Social Sciences as well as for cross-disciplinary and trans-national comparative research. The four resource families that we present each in turn are newspaper, parliamentary, CMC (computer-mediated communication), and parallel corpora. We focus on their presentation within the infrastructure, their metadata in terms of size, temporal coverage, annotation, accessibility and license, and discuss current problems. Keywords: language resources, research infrastructure, open science, digital humanities and social sciences, CLARIN 1. Introduction CLARIN is a European Research Infrastructure that has been established to support the accessibility of language resources and technologies to researchers from the Humanities and Social Sciences (Krauwer and Hinrichs, 2014). CLARIN’s vision, mission and design are aimed at findability, accessibility, interoperability and re-usability of its resources, tools and services to support researchers in the Humanities and Social Sciences (SSH) (de Jong et al., 2018; De Smedt et al., 2018). At the time of writing, CLARIN has 20 member and 2 observer countries which provide numerous language resources and tools through certified data centres. Access to these resources is enhanced by the Virtual Language Observatory (VLO) a portal which enables searching for resources and provides uniform display of highly-granular Component MetaData Infrastructure (CMDI) metadata (Van Uytvanck et al., 2012). Similar to other service-oriented e-research infrastructures (Chunpir et al., 2015), the developmental phase of the CLARIN infrastructure and services was followed by analyses of user experience and needs. Odijk (2014) looked into the problems with resource descriptions, granularity of metadata and resources, and string as well as faceted search in the VLO through the eyes of a linguist. Lušicky and Wissik (2016) conducted a similar survey of resource discovery that was tailored to the needs in translation studies. Sanders (2017) carried out two focus groups in order to obtain user experience and desiderata from two distinct types of users: humanities and social sciences researchers, and language technology and information technology experts. All these surveys reached similar conclusions: both SSH researchers as well as IT experts need easier access to the desired resources and more data type-specific guidance, including comprehensive metadata on the provenance and annotation of the resources, standard formatting, uniform concordancing and text analytics options that enable comparisons …

Content truncated to 500 words