According to R-Trees: Theory and Applications by Yannis Manolopoulos,
Alexandros Nanopoulos, Apostolos N. Papadopoulos and Yannis Theodoridis, there
are a number of algorithms for efficiently determining the nearest neighbor(s)
using an R-Tree (an internet search on the two terms will pull up several).
There are two things to keep in mind about this:
1. You would need to access SQLite's R-Tree "shadow" tables (xx_node,
xx_parent, xx_rowid) directly in to perform the traversals required by all of
the algorithms -- I don't know if this is officially supported by SQLite's
developers, or if these tables are guaranteed not to change;
2. If your dimension is > 1, the R-Tree alone can't give you a reliable
answer about who is closer, you would also need a method to calculate the
minimum distance between two indexed objects.
Peter
On Thursday, August 21, 2014 2:32 PM, skywind mailing lists
<[email protected]> wrote:
>
>
>Hello,
>
>does anybody have any experience with implementing a nearest neighbor search
>using SQLite's RTree functionality? Is a nearest neighbor search possible?
>
>Regards,
>Hartwig
>
>
>_______________________________________________
>sqlite-users mailing list
>[email protected]
>http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
>
>
>
_______________________________________________
sqlite-users mailing list
[email protected]
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
