Efficient Interval Queries

A common use case for certain types of data is the ability to search for
documents that have an interval field that overlaps or that is contained in
a target interval. Typically, this is implemented using an Interval Treehttp://en.wikipedia.org/wiki/Interval_tree or
a Range Tree http://en.wikipedia.org/wiki/Range_tree in other contexts.

I was wondering if ElasticSearch (or Lucene) provided such a capability,
perhaps by leveraging n-grams or geo-spatial indexes.

Thanks!

--

Lucene uses tries in range queries since
Schindler/Diepenbroek, http://epic.awi.de/17813/1/Sch2007br.pdf

Best regards,

Jörg

--

Hi Jörg,

Thanks for this very informative reference. I was very curious how Lucene
was performing RangeQueries, and this certainly gives a better explanation
than the Javadocs. So, if I understand the approach correctly, one should
be able to use a document structure with two fields, say start and end.
Both of these fields should be NumericField. Then, two RangeQueries can be
specified to efficiently (by way of the two prefix tries) find all
documents that are contained in a target interval. Is my understanding
correct?

Cheers

On Monday, 14 January 2013 18:37:49 UTC-5, Jörg Prante wrote:

Lucene uses tries in range queries since Schindler/Diepenbroek,
http://epic.awi.de/17813/1/Sch2007br.pdf

Best regards,

Jörg

--