Oops, forgot to include solr-user@ in the original email. FYI below...
------ Forwarded Message
From: "Mattmann, Chris A (388J)" <chris.a.mattm...@jpl.nasa.gov>
Reply-To: <d...@lucene.apache.org>
Date: Tue, 24 Aug 2010 07:02:58 -0700
To: <d...@lucene.apache.org>
Subject: [Spatial] Geonames and extension to Spatial Solution for Solr
Hi Folks,
You may have noticed over the past few days a bunch of Spatial related
contributions, in particular:
SOLR-2073 Geonames.org UpdateProcessor for Spatial
SOLR-2074 GeoRSS ResponseWriter
SOLR-2075 SpatialQParserPlugin and HostIP adaptor
SOLR-2076 Spatial example schema updates
SOLR-2077 Spatial example solconfig updates
SOLR-2079 Expose HttpServletRequest object from SolrQueryRequest object
SOLR-2081 BaseResponseWriter isStreamingDocs causes SingleResponseWriter.end
to be called 2x
SOLR-2082 Geopost.jar for loading geonames data
These contributions were the results of the final project of one my students
in CSCI 572: Search Engines and Information Retrieval [1] at USC this past
Summer (2010). William and I built off of the existing Spatial code in Solr
to provide some more robustness, and out of the box ease-of-use for Spatial.
Basically what we did at a high level:
* allow easy loading of Geonames.org data using an UpdateRequestProcessor.
You can simply use the Geopost.jar tool we provided to load up exports from
Geonames.org, which contains city, state, and locale information as well as
coordinates (lat, lon)
* load up geo-tagged data (lat, lon). This was already provided, but we
included some default schema fields to automatically suck these up on
ingestion. For our examples, we grabbed a regular RSS news feed, ran it
through the geonames.org GeoRSS converter, and transformed it to a Solr
input XML file.
* Once Geonames.org data has been loaded, and you've loaded up a couple of
docs that have been geotagged (with lat, lon), you can use our extended
spatial Qparser, to issue queries like:
{!spatial ct=[city] s=[state] c=[country] d=[search radius]}search text
e.g. {!spatial ct=Orlando s=FL c=US d=400}NASA
(which in our example looked up articles about NASA and JPL stories within
400 km (or miles) of Orlando, FL in the US).
* allowed for Host IP->lat, lon detection and a pluggable framework to
incorporate services like hostip to provide this functionality. That way, if
a user doesn't include city, state, etc., in their spatial query, they still
get articles "close" to them when using the spatial filter based on host IP
detection. Though this isn't perfect, and largely dependent on the
architectural topology, it's a really good start.
Those are the high level features. We had to fix a bug along the way, and
additionally allow for access to some of the objects that Solr tries to
insulate you from (of course, with good reason), but it's still a fairly
robust spatial solution, and we are incorporating it into several of our
projects here at JPL.
We hope it helps out the community and that folks find it useful.
Thanks!
Cheers,
Chris
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Chris Mattmann, Ph.D.
Senior Computer Scientist
NASA Jet Propulsion Laboratory Pasadena, CA 91109 USA
Office: 171-266B, Mailstop: 171-246
Email: chris.mattm...@jpl.nasa.gov
WWW: http://sunset.usc.edu/~mattmann/
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Adjunct Assistant Professor, Computer Science Department
University of Southern California, Los Angeles, CA 90089 USA
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
------ End of Forwarded Message
