Sigurbjörnsson, B., and van Zwol, R. Flickr tag recommendation based on collective knowledge. In WWW ’08: Proceeding of the 17th international conference on World Wide Web (New York, NY, USA, 2008), ACM, pp. 327–336. [PDF]
This paper presents a technique for supporting the generation of metadata in Flickr through tag recommendation. The authors used tag co-occurrence to generate a list of tags that could complement some user-entered tags. The initial motivation of the author is that photo annotations provided by the author of the pictures relfect a personal perspective and cotext that is important to the owner but that might be confusing for other users trying to retrieve the same picture. Additionally, they note how suggesting non-obvious tags might be confusing for the users. They validated the tag recommendation using tag assessors that had to verify the “descriptiveness” of the generated tags.
Given a photo with user-de?ned tags, an ordered list of m candidate tags is derived for each of the user-de?ned tags, based on tag co-occurrence. The lists of candidate tags are then used as input for tag aggregation and ranking, which ultimately produces the ranked list of n recommended tags.