Tag Specificity

After the Carson Summit and listening to numerous speakers, specifically Joshua Schachter the inventor delicious, Cal Henderson who works on Flickr, and Tom Coates' inspirational speech about the convergence of data in future of the web, I got thinking about tagging.

Both Flickr and Delicious use tagging quite extensively to organise their users data. And a common way of combining data is using tags, for example finding related posts using Technorati.

Before the conference started Ryan Carson, the organiser of the summit - The Future of Web Apps - asked the audience to use a specific tag when uploading photos or blogging the conference. He said to use the tag "futureofwebapps".

The message seemed to work. Currently there are over 500 photos tagged on Flickr and plenty of articles found via Technorati.

Now, the first problem I had this tag was... did he mean "thefutureofwebapps" because that was the name of the conference. He repeated himself and it was definitely "futureofwebapps". Not sure why he decided to drop the "the" but this could have lost a few pieces of data relating to the summit, if people misheard him.

Indeed, there are just over 50 photos on Flickr (from two different people) using the "thefutureofwebapps" tag, but they also feature the "correct" tag "futureofwebapps". However, there was one article on Technorati incorrectly tagged, which also doesn't feature the correct tag. This article maybe missed if this data is ever used.

The second problem I had with this tag was that it isn't a unique reference to this specific conference. If the conference is repeated in following years, that information may become tangled together. Although, this may not be a major problem, it may not have been the intended use of the data. A simple solution for this yearly repetition could be to append the year to the end of the tag. Although, this would solve the problem above, it doesn't stop other conferences with the same name in the same year placing their data in the same pile.

A potential solution for tagging events could be to use the title of the event, date and place name. So for the Future of Web Apps Carson Summit on February 8, 2006 in London the tag could look something like - thefutureofwebapps20060208london - no exactly short but pretty specific.

The other reason in deliberating the specificity of tags is the promotion and potential convergence of data for a new project; The Multipack. The Multipack is a social-based group who are passionate about web standards. The name was chosen because of it's social connotations, such as a multipack of beer.

However, this connotation could lead to problems in aggregating the specific content we want. A perfect example of this has already appeared on Flickr, by searching for the tag "multipack". Although three of the photos are from the group, two are most definitely not!

Multipack as a tag for this group makes perfect sense, although themultipack could avoid the situation described above. Using the tag "multipack" would be OK for the group, but we also have meetups. I could simply use the method described previously, resulting in multipack20060204wolverhampton for the previous Multipack meeting.

If "multipack20060204wolverhampton" is used for events instead of "multipack" then aggregation for the general group would miss this data. This brings me to a more appropriate solution for tagging specific groups and events.

Tags should describe the data you're seeing in small snippets so the data can be compared easily between users, computers and different types of data.

Every piece of data for the group should have a specific set of basic tags. For the case of the Multipack the following tags could be used to always specifically target the group.

  • multipack
  • web
  • midlands

These tags describe the data quite well for both machine and user. At a quick glance you're able to see the data is related to the multipack group, which talks about the web and is based in the Midlands.

Feeds are usually used for aggregation of information on a website and most websites with tags (such as Flickr & Delicious) have feeds for every tag. However, all the basic tags should be used for this type of aggregation.

The idea is that you use the feed which is most applicable to your aggregation, in the Multipacks case, it is the tag "multipack". Then, this data can be filtered based upon the other tags which allows for more specific data aggregation.

How about calling it: multi-tag filtered aggregation?

If we want to aggregate all the information for a specific Multipack event, such as the meetup in February, the following tags could be added, in addition to the basic tags:

  • meetup
  • wolverhampton
  • februay
  • 4th
  • 2006

Again, it is easy for the user to see that there was a Multipack meetup (related to the web and in the midlands) which was held in Wolverhampton on Februay the 4th 2006. This perfectly describes a specific event but also allows for data comparing of any of the tags. Such as other meetups or events in February.

The date in the format YYYYMMDD could also be added, to aid processing of the data, as well as the tag first discussed. This would give the final set of tags for the Multipack meeting in February...

  • multipack
  • web
  • midlands
  • meetup
  • wolverhampton
  • februay
  • 4th
  • 2006
  • 20060204
  • multipack20060204wolverhampton

Applying this to the Carson Summit example, the following tags should be added to any information pertaining to the event.

  • carsonworkshop
  • thefutureofwebapps
  • web
  • london
  • Februay
  • 8th
  • 2006
  • 20060208
  • thefutureofwebapps20060208london

It's important to realise that tags are simply a tool to mark data in a easy way. However, some guidelines for adding tags could vastly improve the comprehensive convergence of data.