Hello,
After much delay I finely have something to post from my efforts to develop
a Spatial Metadata Extraction Tool. All proprietary tools have be replaced
with open source so it can be freely shared. Thanks goes out to Tyler
Mitchell for code that got me going with this aspect. At this point, SMET
doesn't yet do much crawling or exporting to xml, but this is largely due to
implementation questions that need to be resolved.
Regarding xml output, the extracted metadata is currently placed in a python
dictionary whose structure mimics that of a xml, so saving this output to
xml will be simple. The problem is "what then?". What is the work flow
here? Do I create a new ISO19139 record and then edit it in GeoNetwork, or
create a record in GeoNetwork and then merge it with the SMET output or
what? I would love to hear some thoughts on this.
These output question get larger when implementing the crawler
functionality, simply because the number of records become more difficult to
manage. In addition, to make these data more useful, I see the need to be
able to determine two things: if metadata already exist and if so is
metadata is current for the record being crawled. Otherwise, I run the risk
of creating multiple metadata records for data and more of management
nightmare than I already have. This could possibly be solved by placing a
.mef file (or even just a simple .txt file with the metadata guid) with the
data and comparing this with the metadata in the catalogue. I am not sure
the best approach here.
Please share with me any thoughts you may have on these issues. Also, I
would like to post this somewhere like sourceforge for collaboration, if
people think it is worthwhile. How should I proceed?
Attached is SMET.zip that includes SMET.py, companion files GeoObject.py and
MetadataRecord.py, and a readme.txt.
http://www.nabble.com/file/p20688903/SMET-package.zip SMET-package.zip
--
View this message in context: http://www.nabble.com/SMET-metadata-extraction-tool-version-0.3-tp20688903p20688903.html
Sent from the geonetwork-devel mailing list archive at Nabble.com.