Research in Collaborative Tagging Applications: Choosing the Right Dataset

Abstract

Collaborative tagging is an interesting approach that provides the flexibility to add description(s) to a resource according to the user’s own perception about that resource. These applications are the hottest in the areas of Social Bookmarking, Media Content Sharing and E-Commerce. Being favorite among users, these applications accumulate users’ interactions in the form of embedded datasets very quickly. These datasets are very important for further improving these applications and subsequently facilitating the user in better performing his/her activities. We feel there is a need to study these datasets to help researchers test their proposed algorithms on the right dataset and make valuable assessment and informed decisions. In this paper, we have identified measures for evaluating collaborative tagging applications’ datasets suitability for research experiments. The appropriateness of the identified measures is tested through experiments. Based on the results, recommendations are made on the suitability of the available datasets and how future dataset should look like. Researchers working not only in tagging but also in other disciplines can utilize these datasets to test their proposed algorithms without developing their own. This article provides measures which we dig out by reviewing existing available datasets. These measures are significant in selection of suitable and appropriate dataset(s), as selection of inappropriate dataset leads to errors in the results researchers are expecting. This work will prove extremely relevant and beneficial to all researchers who wish to use datasets of collaborative tagging applications for their research experiments.