In making the networks from this open and linked social data, only three similar SparQL queries were used to retrieve the edges:

SELECT ?i1 ?l1 ?i2 ?l2 WHERE { ?i1  prefix:relatesTo ?i2 .
                               ?i1  prefix:someLabel ?l1 .
                               ?i2  prefix:someLabel ?l2 . }

or:

SELECT ?i1 ?l1 ?i2 ?l2 WHERE { ?item prefix:relatesToFoo ?i1 .
                               ?item prefix:relatesToBar ?i2 .
                               ?i1   prefix:someLabel ?l1 .
                               ?i2   prefix:someLabel ?l2 . }

or:

SELECT ?i1 ?l1 ?i2 ?l2 WHERE { ?item1 prefix:relatesToBaz ?item2 .
                               ?item1 prefix:relatesToFoo ?i1 .
                               ?item2 prefix:relatesToBar ?i2 .
                               ?i1    prefix:someLabel    ?l1 .
                               ?i2    prefix:someLabel    ?l2 . }

where often “relatesToFoo” is “relatesToBar” and ?i1 ?i2 ?l1 ?l2 stand for both individual ids (?i1 and ?i2) and their names or nicks (?l1 ?l2, which are usually dispensable). These patterns are used alike for relationship and interaction networks as can be seen in this implementation.

The semantic queries were executed pretty fast (~20ms), a good surprise probably because there was no reasoner involved. The major drawback was the time needed for loading the rdf files: up to 4 minutes for a file of size ~100MB. This directed the efforts (again) to using the Jena/Fuseki endpoint to serve the data. This can take a whole day of work since individual files are to be input into Jena, incorporated to the configuration file if correctly loaded or corrected. In rdflib, problems were found in loading some ttl files and also some strange characters in the IRC logs (e.g. ^\ and ^A) prevented rdf/xml files to load. A workaround was to edit the files with vim.

Steps:

• upload RDF to Jena
• query to get network
• make Erdös partitioning
• write graphml file (or similar),
• write RDF file with ids of Erdos Partitioning and other info
• if necessary, write also the text produced by each agent
• upload files into repo when feasible.