Towards FAIR Data Access

Background In the past decade many different national, EU and global projects have been successful in raising awareness about Open Science importance of making data findable accessible such as stated FAIR principles (Wilkinson et al. 2016). this respect, there advances with respect to options for discovering data. A multitude either thematic or general catalogues are providing faceted browsing interfaces humans Application Programming Interfaces (APIs) use by machines similarly, data-citations publications offer references resources hosted repositories. However, using data-citations, researchers not guaranteed obtain access itself. Mostly resource link catalogue (and also metadata) citation is a “landing-page”, description meant human consumption. The landing-page may contain instructions how download itself but usually it difficult parse machines. Thus approach sketched above does meet requirements scenarios where applications need assured quick Also interpretation from GO states*1 that these “emphasise machine-actionability (i.e., capacity computational systems find, access, interoperate, reuse none minimal intervention) because increasingly rely on support deal result increase volume, complexity, creation speed data.” requirement Persistent Identifier (PID) resource*2, mostly interpreted meaning PID resource’s metadata only. Note we ignore user authentication authorization prior accessing data, here will only consider ‘freely’ accessible. To improve situation machine accessibility number technologies approaches discussed CLARIN Social Sciences Humanities (SSH) infrastructure domain can be useful. We present some comment their suitability. Signposting Signposting*3 technology proposed van de Sompel (Sompel Nelson 2015) release relevant technical bibliographical attributes URI. It's well described, uses HTTP protocol provide additional information via Link Headers*4. Alternatively, HTML type resources, provided elements. community signposting concept was accepted, its implementation deviated made less dependent (Arnold 2021). However downside, embedded specific Component Metadata (CMDI) (Broeder 2012), so makes specific, at least requires clients knowledge CMDI. Digital Object Gateway (DOG) One currently worked research DOG library*5 (later) service provides proxy gateway actual implicit repository solutions used B-type centres*6 repositories outside infrastructure. works two steps: first obtaining secondly extracting links metadata. Each registered within has configuration specifying fields interest resource's For centres content negotiation primary way CMDI format. infrastructure, primarily relies API order resources. solution scalability problems, limited centres, until better one becomes available. Limited kernel (limited) assumes every (DO) (Berg-Cross Handle (CNRI 2020) issued record store associate important (meta) handle value types example checksum This simplification architecture work done RDA context: Kernel Info recommendations (Weigel 2018). Consistent records could solve problem, just strategy, strong discipline maintain source. Examples smaller exist do manage eg. DARIAH-DE repository*7. Objects (FDO) FDOs*8 attempt overcome management challenges posed heterogeneity complexity combination abstraction, virtualization encapsulation (Schwardmann 2020). practice, context our FDO seen both generalization upgrade approach. key characteristics (conceptual) objects structure services allow aware recognize bitstream format, process intended programmer. Eligible processing services, ones communities, found through typing mechanism, directly linked FDO. rich set permit signaling FDOs including instance supported protocols APIs. What do? collaboration others, now would prefer invest get closed unscalable technologies. propose combine signposting. First testing URIs (obtained resolving PID) presence Headers. If missing, (extended) idiosyncratic workflow. Long term see advantages general, scalable independent offer. Hybrid conceivable proxies sit between machinery compliant