I am creating an application for the iPhone that will use a weighted schedule (maybe about 1000 nodes and 500-700 relationships, the assessment of the ball park may be completely wrong).
The core data would be great as it worked easily and is optimized for the iPhone. At the same time, the nodes of the graph must be weighed. I could add a level of indirection (editing / refining: I would present weighted edges as relationships.) In the master data, but I believe that this can lead to a decrease in performance.
SQLite would allow me to do this without any additional indirectness, the performance problem.
Another consideration is that I use the main data to store the material, and then only save the weights in sqlite in the tables and create special kernel data objects which, when I extract or create, creates the necessary links in sqlite and then just use sqlite for the calculation (since the calculation will only go through and look at the weights, so I do not need anything but the final result / id of the object). Then I looked at the actual object in the master data.
In addition, there will be a server that takes into account all the graphs of users on a much larger graph. Thus, it would be easier to use sqlite directly, since replicating information on a brutal server would be easier :-)
Thanks for the help.