39 Hints to Facilitate the Use of Semantics for Data on Agriculture and Nutrition
Recommendation Title: 39 Hints to Facilitate the Use of Semantics for Data on Agriculture and Nutrition
Impact: The document contains a number of highlights on issues to be addressed in order to make the use of semantics more straightforward and widespread. We expect the document to help share awareness about semantics among experts in different areas (eg, SW development, data curation, policy makers) and support strategic allocation of resources in future development.
|Recommendation package DOI: http://dx.doi.org/10.15497/RDA00036|
Summary of the Recommendations from the RDA Agrisemantics Working Group
This document presents the recommendations of the RDA Agrisemantics Working Group (WG) to promote the use of semantics for agricultural data for the purpose of enhancing data interoperability in agriculture. These recommendations are high-level, to encourage researchers and practitioners to extend them according to their area of expertise.
The activities of the Agrisemantics WG started off with an analysis of the current landscape of use of semantic resources with agricultural data, based on first-hand experience of our group members, as well as on bibliometric investigation, and an analysis of existing repositories. From that first activity, we found that despite the many possible applications of semantics and the interest shown by research and industry alike, their actual implementation and use is lagging behind. A large number of resources are not in machine-readable formats or do not have public APIs, while those available are often used beyond their intended area of use, possibly leading to problems. Our conclusion is that there is a strong need for “lifting” semantic resources to the web, and create new ones appropriate to their intended use.
Based on the insights gained from that analysis, we moved on to look in more detail at the actual reality of working with semantic resources, we opened a call for use cases, to collect examples of actual work and limitations, together with evidence of work already ongoing to address them. From that, we found that a large variety of profiles (and so skills and roles) are involved in working with semantic resources, and that there seem to be as many tools and toolkits as projects. Following the explicit suggestions provided by our respondents, we identified three types of implications of this fact, namely:
Implications on software usability. Given the variety of profiles involved, tools designed for use with SRs should be accessible to non-ontologists. In particular, more attention should be payed to graphical interfaces, support for validation, and for methodological support in each task.
Implications on tools availability. Reportedly, not all institutions or organizations can afford the whole set of activities associated to them, online platforms are needed to lift the burden of local (or ad-hoc) installations and maintenance from users or individuals.
Implications on tools interoperability. From the use cases presented, it emerges that the different tasks commonly required to work with SR are commonly achieved by pipelines of different tools. Given this fragmented landscape, we recommend that tools to perform common tasks involving SRs (e.g. editing, format conversion, etc.) should be integrated, or integratable, to form flexible and interoperable workflows. This approach would contribute to minimizing the breadth of skills required to work with SRs.