“Power tags” in information retrieval

Many Web 2.0 services (including Library 2.0 catalogs) make use of folksonomies. The purpose of this paper is to cut off all tags in the long tail of a document‐specific tag distribution. The remaining tags at the beginning of a tag distribution are considered power tags and form a new, additional search option in information retrieval systems.

In a theoretical approach the paper discusses document‐specific tag distributions (power law and inverse‐logistic shape), the development of such distributions (Yule‐Simon process and shuffling theory) and introduces search tags (besides the well‐known index tags) as a possibility for generating tag distributions.

Search tags are compatible with broad and narrow folksonomies and with all knowledge organization systems (e.g. classification systems and thesauri), while index tags are only applicable in broad folksonomies. Based on these findings, the paper presents a sketch of an algorithm for mining and processing power tags in information retrieval systems.

This conceptual approach is in need of empirical evaluation in a concrete retrieval system.

Power tags are a new search option for retrieval systems to limit the amount of hits.

The paper introduces power tags as a means for enhancing the precision of search results in information retrieval systems that apply folksonomies, e.g. catalogs in Library 2.0 environments.