The Open PHACTS VoID Editor helps non-Semantic Web experts to create machine interpretable descriptions for their datasets. The web app guides the user, an expert in the domain of the data, through a series of questions to capture details of their dataset and then generates a VoID dataset description. The generated dataset description conforms to the Open PHACTS dataset description guidelines that ensure suitable provenance information is available about the dataset to enable its discovery and reuse. The VoID Editor is available at http://voideditor.cs.man.ac.uk. The source code can be found at https://github.com/openphacts/Void-Editor2.
Users of systems such as the Open PHACTS Discovery Platform3 [
Open PHACTS does not publish its own datasets; it integrates existing
publicly available domain data. Typically the publishers of these scienti c data sets
are experts in their scienti c domain, viz. chemistry or biology, but not in the
semantic web. They need to be supported in the creation of VoID descriptions
of their datasets which may have been published in a database and converted
into RDF. A tool which hides the underlying details of the semantic web but
enables the creation of descriptions understandable to a domain expert is thus
needed.
3 https://dev.openphacts.org/ accessed July 2014
The aim of the VoID Editor (see screenshot in Figure 1) is to allow a data
publisher to create validated dataset descriptions within 30 minutes. In particular,
the data publisher does not need to read and understand the Open PHACTS
dataset descriptions guidelines [
The VoID Editor is a web application that guides the data provider through a series of questions to acquire the required metadata properties. The user is rst asked for details about themselves and other individuals involved in the authoring of the data. Core publishing metadata such as the publishing organisation and the license are then gathered. The user is then asked for versioning information and the expected update frequency of the data. The Sources tab helps the user to provide details of source datasets from which their data is derived. They can either select from the datasets already known to the Open PHACTS Discovery Platform or enter the details manually. The list of known datasets is populated by a call to the Open PHACTS API. The Distribution Formats tab allows the user to describe the distributions in which the data is provided, e.g. RDF, database dump, or CSV. The nal screen allows the user to export the RDF of their dataset description as well as providing a summary of any validation errors, e.g. not supplying a license which is a required eld, such errors will already have been indicated by a red bar at the top of the screen containing an error message. Note that the `Export RDF' button is only activated when a valid dataset description can be created, i.e. all required elds have been lled in.
At any stage, the generated RDF dataset description may be inspected by clicking the `Under the Hood' button. This button can also be used to save a partially generated description that can later be imported into the editor through the `Import VoID' button. The `Under the Hood' feature is also useful for semantic web experts to see what is being generated at any stage. 3
In companion with the VoID Editor, a Linkset Editor (see screenshot in Figure 2) has been developed. The Linkset Editor allows for the creation of descriptions of the links between two datasets. The same interface design and framework is used.
The Linkset Editor reuses the rst three tabs of the VoID Editor to capture
details of the authors, core publishing information, and details about versioning.
The Source/Target tab allows the user to select the pair of datasets that are
connected by the linkset. Again, the list of possible datasets is generated by
a call to the Open PHACTS API. The Link Info tab asks the user to declare
the link predicate used in the linkset and provide some justi cation to capture
the nature of the equality relationship encoded in the links. (For details about
linkset justi cations, please see Section 5 of [
The VoID and Linkset Editors have been implemented using AngularJS as a
Javascript framework for the web client with a server implementation using Jena
libraries. A user-centric approach was followed for the design and development
of the VoID Editor. A small number of data providers were consulted about the
type of tool they required with regular interviews and feedback on prototype
versions. A larger number of potential users were involved in an evaluation of
the VoID Editor. Full details can be found in [
In the future we plan to investigate how the VoID Editor can genearate template descriptions that can be populated as part of the data publishing pipeline. We also plan to look at how the editor could be adapted to other dataset description guidelines, e.g. DCAT4 or the W3C HCLS community pro le5. However, this is not a straightforward process since considerable care and attention is paid to the phrasing and grouping of questions to ensure a pleasant user experience. Acknowledgements The research has received support from the Innovative Medicines Initiative Joint Undertaking under grant agreement number 115191, resources of which are composed of nancial contribution from the European Union's Seventh Framework Programme (FP7/2007- 2013) and EFPIA companies in kind contribution.