Glue: improving tag-based contents retrieval exploiting implicit user feedback

Tagging systems are increasingly becoming an interesting feature of web 2.0. An important metric to evaluate the quality of a tagging system is the efficacy of its internal search engine. Traditional tagging systems search engines are based only on the direct resource tag association performed by the users when a resource is initially uploaded to the system. This way, the user is not able to avoid the influence of misleading tags or to exploit tags similarity. In this paper we propose Glue: a technique exploiting tag-resource and tag-tag associations built-up on implicit user feedback coming from its search activity, in order to increasingly improve search efficacy. Experimental results, obtained by analyzing a custom tagging service used by a restricted community of users, show that our approach is able to discover and emphasize proper associations and that it outperforms a traditional search in terms of research efficacy.

@InProceedings{ACA08,
  author       = {Alimonda, A. and Clemente, M. and Agelli, M.},
  title        = {Glue: improving tag-based contents retrieval exploiting implicit user feedback},
  booktitle    = {Proceedings Of The 8th Dutch-Belgian Information Retrieval Workshop (DIR 2008)},
  pages        = {29--35},
  year         = {2008},
  editor       = {Eduard Hoenkamp and Martine de Cock and Veronique Hoste},
  note         = {isbn: 978-90-5681-282-9idxproject: ?},
  keywords     = {tagging system, tags, search engine},
  url          = {https://publications.crs4.it/pubdocs/2008/ACA08},
}