While the page relations application  introduced the ability to draw basic relations between pages outside the page content itself, a new branch I've been working on  implements "typed relations", for instance: "The Jungle Book" "has author" "Rudyard Kipling", where each of these items is represented by a page, and page "has author" itself has relation "is a" with page "Relation". It is similar to what MediaWiki extensions such as Semantic MediaWiki  or Wikibase  propose. A documentation draft covers the approach in more details in the XWiki development area . Since the application data has an underlying graph structure, I was wondering whether we could use the name "application-graph" for it, considering also the current traction that graphs seem to have .
The branch code is now installable, it comes with a simple demo around Nobel Prize laureates. The typed relations are managed by dedicated Solr fields and the application implements a generic graph API that I came up with  because I couldn't find one with the needed features among the existing libraries such as Apache Giraph  or JGraphT .
What do you think about this approach? Do you have anything against adopting the name "application-graph"? Would it make sense in your opinion to host the graph API it contains under the xwiki-commons namespace?