Metadata Collection Records for Language Resources

Metadata Collection Records for Language Resources Henk van den Heuvel*°, Erwin Komen°, Nelleke Oostdijk* *CLST / CLS, Radboud University °Humanities Lab, Radboud University Erasmusplein 1, Nijmegen, the Netherlands {h.vandenheuvel, e.komen, n.oostdijk}@let.ru.nl Abstract In this paper we motivate the need for introducing more elaborate and consistent metadata records for collections of (linguistic) data resources in the CLARIN context. For this purpose we designed and implemented a CMDI profile. We validated the profile in a first pilot in which we populated the profile for 45 Dutch language resources. Given the complexity of the profile and special purpose requirements we developed our own interface for creating, editing, listing, copying and exporting descriptions of metadata collection records. The requirements for this interface and its implementation are described. Keywords: LR infrastructure; metadata; collection records; Collection Bank 1. Introduction In the context of CLARIN (Common Language Resources Infrastructure)1 various stakeholders have been working towards the realisation of an integrated, interoperable research infrastructure. Apart from the technological achievements and the services offered, key factors to the success of this infrastructure are the availability and accessibility of the language resources. Ideally the infrastructure is populated with as many resources as possible. Therefore continued efforts are being put into gathering what resources there are and curating them, targeting also the many resources that have been created in the context of research projects but have not (yet) found their way to one of the data centres or repositories.2 In order to provide an entry-point to the language resources available in the CLARIN infrastructure, the Virtual Language Observatory (VLO) was developed (Van Uytvank 2014; Van Uytvank et al. 2010). The VLO offers a faceted browser which allows users to search for resources through their metadata. User experiences with the VLO have shown that discovering resources, and especially resources of which one is not aware they exist, is problematic. From the analysis of Odijk (2014) we know that while some of the problems arise from the current limitations of the VLO, in many cases the cause of the problem lies in the nature of the metadata. The set of facets used for search in the VLO is (too) small, not all facets are relevant for the discovery of resources, and some facets are lacking and should be added.3 In 2015 the CLARIAH Metadata Curation Taskforce4 was charged 1 https://www.clarin.eu/ 2 In the Netherlands the Data Curation Service was set up as a centre of expertise to assist researchers in preparing their data for delivery to one of the CLARIN centres (van den Heuvel et al. 2015). 3 Lušicky and Wissik (2017) also note the need for additional metadata. while using the VLO for discovering resources in the field of translation studies and urge other user groups to assess the VLO from their perspective and specify what should be added to satisfy their needs. 4 In the Dutch CLARIAH project (https://www.clariah.nl/en/), the Metadata Curation Taskforce is concerned with providing metadata for various resources. with the task to come up with a profile for …

Content truncated to 500 words