Eran's blog

Author Tags vs. Reader Tags?

A very simple distinction between tagging systems is based on who creates the tags. Author tags, created by the same person who created the content, can be seen in places like Technorati where blog posts are aggregated along with their tags. Reader tags, on the other hand, are created by anyone else and so might be closer to annotation systems. You can see reader tags on sites like delicious that allow anyone to tag any document . Interestingly, systems like flickr combine both into one mix, allowing both author and (some) readers to add tags.

It’s been saidthat tags created by readers are better as information consumers, through numbers and a different point of view, see things in the content that the author may have missed or does not see at all. This means that data becomes easier to find and more connections are created between different pieces of information, improving associative browsing. But at the same time, reader tags generate more noise (for many reasons) thereby making it harder to find good information.

A comparison between two such systems, say Technorati and delicious, would be very interesting as it would allow us to quantify those differences, see how readers and authors differ in their tagging behavior and how much more information do reader tags actually add to the system. Any takers?


Filed under: General

7 Responses - Comments are closed.

  1. ryan king says:

    Damn it! I knew you were going to get to blogging it before I did. I guess I’ll take the task.

  2. limbo says:

    I expect to see results by the time I’m back from the desert…

  3. Assaf says:

    Interesting question.

    I wonder what the difference would be betwee author and reader (singular), author vs collection of readers, tags to promote vs tags to search

    settling the question whether reader tags add anything is easy: look for tags like ‘favorite’, ‘to-read’, ‘top-pick’. would you trust author or readers?

  4. limbo says:

    Assaf: Operative tags like “toread” or “wishlist” are very personal (even more so than most tags) and carry little information for anyone but their owner. The one exception I can think of is if you look at them as positive votes for the resource in question. Would I trust them? I don’t know.

    I’m more interested in the categorization value of tags – when I tag a post with “folksonomy” and “tagging” I assume that people understand it’s also about the Internet, the Web, knowledge management, etc. but when a resource gets tagged on delicious by hundreds of users some of those connection will be made explicit as some users will tag it with “web2.0” or “knowledge management.” Could we glean that information from the author tags alone? I dont know. Are readers affected by the tags used by the author? Are they affected by tags from other readers? There are plenty of questions that are yet unanswered about people’s tagging behavior.

  5. vanderwal says:

    The difference between author and reader tags has to do with vocabulary, definition, terms, personal/community usage, small group ontologies, etc.

    The grey area in author tagging is what are they tagging? The tags can be one link, the whole post/article, a subject not specifically named (tagged with geek, but never mentioned geek in the text), marketing (tag with a hot word to draw attention), etc. The variation of options makes the use of the tags harder to discern. When I use Technorati Tags I and search on folksonomy I get things that do not relate at all, or do not fit my definition of folksonomy. I find the keyword search results in Technorati and Feedster far more valuable as the words are actually being used in the text and I can see their intent and definition based on their words use context.

    Reader generated tags can still have some of the problems with tags listed above. But if the person tagging is a distinct data point we can filter them out if their tagging is not helpful to our uses. This was a large part of the realization of what separated del.icio.us from all tagging efforts before it. The distinction was named folksonomy to separate it from general tagging that had gone before it. A folksonomy has three distinct data points: the person tagging, the tag, and the object being tagged. In folksonomies the information is done for the person tagging’s own use above others and the tags are shared with the community. The difference between a del.icio.us and Flickr tagging is the person tagging is not their own distinct datapoint.

    Vocabulary and definitions are very important in tagging and in a reader generated tag environment we can use the user as a data point and find community, cross-cultural connections, and cross discipline connections. Lets say Jeff tags something with mash-up and web2.0 and his friend Tara tags the same object with mash-up and AJAX. Does mash-up have the same connotation and definition? Most likely, but we can easily see other uses of the same mash-up tag as they have applied it to other things and assess if there definitions are similar or exact or not. Reader generated tags are much richer than author generated tags as they can be used to pivot and follow the tag, the person tagging, the tag only by one person, or follow all tags for that object being tagged.

    Extracting connections from tags and assuming hierarchies can be dangerous. Just as people have variations on their definitions the related words or subjects is less stable in larger groups. In small groups this can work as is shown in academic research in small group ontologies. Extracting explicit relationships in larger groups breaks the value in the distinct data points of the person tagging. Tags applied are associative and not hierarchical and inference of association outside the explicit tags listed has less of a correlation in pairings that also have weaker values. A person tagging with “tag” may not have any derived meaning in that term being related to knowledge management, while an other person may.

  6. Assaf says:

    I disagree. I think this distinction was invented when people were trying to justify tags as better classification scheme then categories, so tags where compared to categories. and so the question became, what do you do with the noise?

    I don’t see the noise. sometimes ‘wishlist’ is just more interesting than ‘web2.0’, sometimes 43Things is a better read than TechCrunch. my Amazon wishlist tells more about me than my del.icio.us tags, it certainly has more cool toys. now let’s say I’m learning about Ruby, and there’s a million posts out there tagged with Ruby. I’m going to narrow down based on ‘toread’, ‘top-pick’, ‘linkto’, etc. why? because all these “personal” tags reveal something really important — they tell me what’s on other people’s priority queue.

    it’s a gesture. and gestures are great ways to find interesting stuff.

    back to the learning about Ruby example. so there’s a tag cloud and it’s related to ‘web2.0’ and I really couldn’t care. but if you can find me all the posts tagged with ‘talk to suzy about this’, I’d pay money. why? because that gestures means the post is worth reading. the best posts are the ones on your in/out priority queues.

    and to answer the last question. if when I tag something with del.icio.us it would pull out and suggest tags from the original post (maybe it does?), most of the time I’m going to stick with the author tags. I’ll probably add some of my own, remove some redundancy, rename tags because I use different terms. but right now, if I blog about a tagged post, I do tend to carry some of these tags over.

  7. […] Author tags vs reader tags […]

%d bloggers like this: