Suggesting tags is not always a good idea
When we are tagging a resource on a social tagging system, it could save us work by suggesting us tags for that resource. Therefore, we can choose the tags we consider relevant or representative for us. This tag suggestion can be performed in several ways, e.g., by suggesting tags that other users have annotated earlier for the same resource, or by using machine learning techniques to look for suitable tags.
Amongst others, Amazon uses the latter approach to suggest tags when nobody has annotated a book yet. In this case, we can see a subsection “Suggested Tags from Similar Products” for that book. Nonetheless, this doesn’t seem to be a good approach, and it sometimes suggests us weird tags, as happens for O’Reilly’s book “Cassandra: The Definitive Guide” [retrieved on November 3, 2010]:
As stated by Delicious founder Joshua Schachter:
Automatic tags lose a lot – doesn’t help the user really achieve their goals. That’s why the ‘add to del.icio.us’ badges don’t let you suggest tags.
Well, I think we should pay attention to Schachter’s quote, and let the users tag by themselves, without suggesting them automatic tags.
More posts will come soon on tag suggestions.
Category: social-bookmarking, social-tagging, tagging