PlanetCatfish.com

Sort of a spinoff from this...

Is there any way to link the type localities (where they exist in the Cat-eLog) to the Fuzzy Gazetter?

You mean like this:

<a href="http://tomcat-dmaweb1.jrc.it/fuzzyg/que ... ">Xingu</a> ?



Jools

The problem appears to be the search isn't that good. A sample type locality (Pim pictus):

Rio Hyavary, Peru-Brazil border.

would link to

http://tomcat-dmaweb1.jrc.it/fuzzyg/que ... zil+border.

Which isn't a very useful result. I could put this in place with only about 1.5 hours work, but results would be variable at best?

If I wrote something much more clever like getting it to remove common words like country names, rio, border etc and it just searched on the unusual words, it still is inaccurate:

Consider http://tomcat-dmaweb1.jrc.it/fuzzyg/query/?q=Hyavary which comes up with somewhere in Norway.

This would be great if it was less fuzzy - in being less fuzzy it would find less - something very like this would be a great addition, but I'm not certain this is quite the right thing.

Jools

I was thinking more going to this (which is what happens when you click a search result).

And as for the results with P. pictus, this entails a little bit of additional work. Searching for this yields the correct result (the first one).

We can't go straight to the digital map; that doesn't have an interface that allows searching on english names.

What fuzzyg does is provide a list of search results each with a link containing latitude and longitude co-ordinates i.e.

http://tomcat-dmaweb1.jrc.it/fuzzyg/map ... =Rio+Xingu

Unless you stored these co-ords in the cat-elog, you can't go direct from it to the map. What I could do is build a link around the type locality and distribution data items that passes the contents of those fields to the fuzzy search. Then it's up to the user to select the search result.

Jools

I'm probably signing myself up for some enormous amount of work here, but would it make sense to have a different format of "Type Locality" that (where possible and relevant) combines the Occurrence database and a Long/Lat location from an original description, and then a link to Google Maps or some such for those Long/Lat coordinates?

Of course, I suspect, just as with the distribution data, that the type locality is most likely contains a lot of meaningful information that can not be represented in any form other than free text, so not ALL the information can be used.

Of course, the original suggestion is quite old by now...

--
Mats

So, just to pull this and other threads together, this is all combined into this thread:

http://www.planetcatfish.com/forum/view ... =3&t=32861

Jools