Decomplexifying networks : a tool for RDF/Wikidata to network analysis

Linked Data is a common way to publish structured data in the Humanities (De Boer et al., 2014; Meroño-Peñuela et al., 2017). The non-discriminate model allows users to model knowledge as multimodal graphsmeaning theoretically any related objects can be represented as a graph. Inherently, this affords modeling data as networks and thus the implementation of network analysis. The analysis of networks from RDF is largely done with a pipeline of tools (i.e. Groth & Gil 2011; Gil & Groth 2011). In these workflow approaches, researchers specify SPARQL queries, extract networks and export data as matrices, and implement network analysis tools to investigate graphs. The development of such a pipeline can be a technical adversary, for domain experts (e.g. historians, literary scholars) with (traditionally) limited technical knowledge, but also for researchers with specific expertise in RDF or networks. In addition, in building such pipeline, we lose sight of the hermeneutics of the research objects (Gibbs & Owens 2013). Thus, we argue, there is a need within the DH community, to reduce this RDF-to-network analysis pipeline without creating another domain or research question specific tool and while maintaining oversight over the process from RDF-to graph-to network analysis.