It will be good to have suggestions for tags when you type them in text field. It will help to avoid duplicates with different spelling (for example now there are some duplicates differs only by dashes instead of spaces).
Autosuggest is on the to-do list and we have it working on the backend. We have some vacation time for key developers that will slow us down a bit, but we will be back on track in August.
And thanks for all your tagging and commenting today Denis, it’s much appreciated. You are the first ever “Robot Herder” on the Qt Developer Network, grats! :)
Not so sure. I find that in my work (bibliometric analysis) we can do very nice things with linking based on content. You need to do it smartly, of course, but it can work.
Another tag-related thing that I encountered, is that I think rights do not work properly. I get the option to remove tags, while I don’t have Hobby Entomologist status yet (I just reached Ant Farmer).
P.S. Congrats on reaching Robot Herder, Denis!
Yeah, MariusG said that tag search is not 100% ready in this blog [developer.qt.nokia.com] comments.
What might be nice too in this context, is to show a list of related items to the one currently being viewed.
Andre is at the core right there, we want to do exactly that. We have some way to go before everything is ready, but it has been the goal all a long. We have more content coming from other parts of Qt that will make that feature scale in usefulness.
We will obviously use tags and ratings to find that related content, but we’re open to other solutions as well. Text semantic analysis, navigation summary, author relations, custom lists / play lists, bookmarking feature… So much to try, so little time :)
If you guys know of sites that have excellent “related content” solutions please let us know.
I don’t know how other sites implement this kind of thing (I am not a big fan of web development myself), but I have plenty experience with building up this kind of networks on other datasets. That’s what the software I build is for…
Perhaps you could look into using a Jaccard or Salton index on the keywords and/or title words and/or content words to calculate the similarity between pieces of content. They are not too difficult to calculate.
:)We will obviously use tags and ratings to find that related content, but we're open to other solutions as well. Text semantic analysis, navigation summary, author relations, custom lists / play lists, bookmarking feature… So much to try, so little time
Hm, so user relations are in roadmap too? ;)
Nice (not needed, but fun!) feature would be to be able to generate a network of authors based on which authors post in the same topics. That would be an indication of the interests/knowledge areas of those authors. Same goes for contributions to the same wiki pages, or perhaps even for the tags users use, or… The possibilities to play with this kind of thing are endless :P
These kind of things are fun to do and play around with. :-)
I think I have seen it done for commits in the KDE repos too, where a relation between two contributors is established when committing to the same modules or files or something like that. Nice way to see how some people specialize in one area, while others operate more all-round.
But hey, I’m a sucker for this kind of thing, so don’t pay too much attention to my ramblings here :-)