Haven’t tried it but just read about it after seeing your post and, indeed, it seems a very interesting DB (graph DB) that could be a good for XWiki as its store engine. However, that’s a huge work and would need either some research project or some sponsoring to have someone spend time working on this.
Now, in your message, you don’t suggest replacing the store engine but to add a new store on the side. In practice this is exactly what we do with SOLR, which is synced with the RDBMS (when a doc is created/modified/deleted, it’s put in an indexing queue for SOLR).
What would be the main differences with SOLR in your opinion? I mean, the SOLR index format is extensible and it’s possible to add attributes (like related pages) so that queries can then be made and the search results displayed with more information (related pages).
Thanks for starting the idea.