Retrieving data in constant time sounds very promising especially for routing purposes. Did you allready have thoughts on algorithms that use hierarchies or contraction of the network? In our students project we implemented some approaches to reduce search space and have thoughts on the heuristics used for A*. When travel speed comes into concern we got very different results with SLD heuristics. Maybe this could be of interest for Neo4j. When doing bounding box queries the only advantage of Neo4j could be better spatial locality of data ('nodes' and 'ways' both stored in the same container). Everything else is up to the r-tree performance as in PostGIS, which uses an improved r-tree in GIST. Maybe you can do an r-tree search for the first node near the center of the bounding box and with some magic retrieve the surrounding nodes with graph traversing... At least I will think about this, if you can improve bounding box queries over PostGIS. When I switch to a low zoomlevel I want to use generalization and simplification of mapdata. In Neo4j I can model that within subgraphs I guess? For example have different relationships for different zoomlevels (I skip some nodes of a way for example). I could also build a routing network within subgraphs where I skip non-junction nodes for example? Btw what do you think from building a graph inside a relational database? As I understand now it would be very inefficient because you would have at least two tables (vertices and edges) and have to do time demanding joins with the two foreign key colums in the edge table.
Regards, Daniel -- View this message in context: http://neo4j-community-discussions.438527.n3.nabble.com/Neo4j-in-GIS-Applications-tp3393925p3409062.html Sent from the Neo4j Community Discussions mailing list archive at Nabble.com. _______________________________________________ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user