Tagging vs. Controlled Vocabulary: Which is More Helpful for Book Search?

The popularity of social tagging has sparked a great deal of debate on whether tags could replace or improve upon professional metadata as descriptors of books and other information objects. In this paper we present a large-scale empirical comparison of the contributions of individual information elements like core bibliographic data, controlled vocabulary terms, reviews, and tags to the retrieval performance. Our comparison is done using a test collection of over 2 million book records with information elements from Amazon, the British Library, the Library of Congress, and LibraryThing. We find that tags and controlled vocabulary terms do not actually outperform each other consistently, but seem to provide complementary contributions: some information needs are best addressed using controlled vocabulary terms whereas other are best addressed using tags.