Structure in the flow

I was recently talking with someone about bookmarking and he mentionned it’d be great if we could simply click the star button in Firefox 3 and have it automagically add machine-selected tags to the new bookmark. I thought I had heard about similar services, but had never tried any yet. Now, I did say I now “bookmark” in a personal wiki, but I still use a few social bookmarking services for links I want to share, so this would be helpful for me in that perspective.

Searching quickly, the software that approaches this idea the most closely seems to be HandyTag, a plugin for Firefox. I haven’t tried it yet but it looks for tags in the page title and metadata, tries to see what tags users on delicious are using and, lastly, uses another Firefox extension called KGen to use word frequencies on the page as yet another source for the suggestions.

The web applications I had heard about were Zigtag and Faviki, which were mentionned in previous months on ReadWriteWeb. I haven’t yet found a way to add a 1-click bookmark with any of these, though. Both Faviki and Zigtag come as bookmarklets. Zigtag also has a Firefox extension. They both present a series of automatically generated suggestions in a box similar to Google search suggestions. For niche pages, based on 4-5 ad hoc tests, Faviki seems to have more suggestions. For popular pages, they’re competing with delicious‘ suggestions drawn from the pool of user-added tags for a given URL, which in my experience are usually adequate.

Another web app which seems promising is Favo, which its homepage says will offer automatic cataloging, but it’s still in private beta so I’m not quite sure how that’ll end up in practice.

Ambiguity in tagging systems

Both services attack the problem of lack of semantics in tags, which is about ambiguity in tagging systems: if I tag a page “Apple”, is it the fruit or the company I’m referring to? Similarly, if I want to tag a page “computers”, do I use “computers”, “computer”, “personal_computer”.

So, basically, what these services do is provide suggestions for your tags, those suggestions having proper semantics. Faviki, for example, will have tags refer to Wikipedia entries, which have well-defined meaning (given Wikipedia has its own disambiguation process). In both cases you end up having suggestions like “Apple Inc.”, “Apple (fruit)”, etc. in a completion list.

Back to productivity

On the practical level, though, the suggestions can be quite numerous, and I’m not quite sure if, from a strictly personal perspective on the bookmark-retrieve process, they’d actually improve my productivity in the end: you have to go through the whole list of disambiguated suggestions until you find what you want, and I don’t have have that many bookmarks relating to apples as a fruit. Therefore it might actually make the process more tedious. But if I want my bookmarks to merge well in public lists, then it’d make much sense, and that’s why we’re using social bookmarking in the first place, right?

Yet, for the original purpose of 1-click bookmarking, I guess it’s not very appropriate. At least, if you feel lazy, they add some machine-extracted suggestions which you didn’t have in delicious, so it makes for more pages for which you don’t have to reach for the keyboard to bookmark.

As for Favo, we’ll have to wait and see.