del.icio.us is stemming its tag searching
In an earlier post I was pointing out that del.icio.us tag search wasn’t very effective as it restricted to stemming only, what this means is if you do a search like, tag:rss, you may get some bookmarks in the results that don’t have this tag, but may have a tag like, rsssearch, rss_tools, etc
…this means it returns results in tags that have the characters, rss…although I’m told it shouldn’t return a tag like, searchrssblog (as the, rss characters are not at the start of the string)
It would be good if there was a way to turn stemming off, if I search for tag:rss, I don’t want to see hits that have the tag:rssreaders (unless the tag, “rssreaders”, has been applied to the same bookmark of course)
…if I search, tag:rss, it should just be like clicking on the tag “rss” in the tag cloud.
So in other words I’m saying that the search, tag:rss, does not bring up this URL http://del.icio.us/tag/rss…it doesn’t make a difference if I use quotations marks either like, tag:”rss”.
Some people might say, why is this a problem anyway if you can select that tag from the tag cloud or enter it to the URL in the address bar. Well what if I want to search for a free text search within a tag, eg. tag:rss enterprise, or even structured like, enterprise tag:rss, or search for a term within a tag.
I don’t mind the stemming of tags, but there sould be a way to unstem…as I said in my other post, it would be good to see an intermediate page of a search like, tag:rss, where it would list all the tags with the characters, rss, then you choose the exact tag you want to see
…so if you do a fielded tag search like, tag:rss, may it could return a tag cloud, then you choose the tag you want.
See Simpy for some great search features…although I’m not sure if Simpy stem their tag searches…if so there should be an option to unstem…I like the idea of having both options as they are both useful.
Also I did a tag field search, tag:rss -tag:wiki…this worked great as I didn’t get any bookmarks tagged with the string, wiki, if I did the search, tag:rss, I would get bookmarks with the tag, wiki, so the “-” operator works…but we still have the stemming problem, as mentioned before I’m still getting bookmarks with tags like, rss_tools.
More…
When you are in your own collection, instead of using boolean operators with tag fielded searching, what about offering symbols for OR, and NOT, just like they have for AND (using a “+” symbol called an intersection)
In your own collection you can use the “+” symbol to add tags, why not the “NOT”, and the “OR” symbol…you can even use the “+” symbol in the URL address bar if you like, what about the others
…and what about offering this for all of del.icio.us, so you can search across the whole database.
Improbulus has some insight as always.
Anyway del.icio.us is slowly getting there, here are the new search goodies.
![email.png](http://services.nexodyne.com/email/icon/GxbiQvgbwBgCnw%3D%3D/oIp%2BJMQ%3D/R01haWw%3D/0/image.png"){kind=icon}
Hola,
Just a quick comment from a search/Simpy guy - searches like “rss” vs. “rss_tools” vs. “rsstools”, etc. are not really about stemming. If you are curious about technical details, feel free to email. Got to run now.
Comment by Otis — November 13, 2005 @ 1:28 am