Mining the link structure of social networks is a relatively old topic dating back to 1950, but the field has recently gained tremendous attention, due to the compilation of huge amounts of data and the development of sophisticated computerized methods to process these data. Among the many facets of these mining tasks in various types of social networks, the mining task of ranking researchers or conferences/journals in citation networks is of particular importance and challenge due to its uses in many practical situations, e.g. faculty promotion, funding. The development of a fair metric to perform the ranking is a very hard task and a lot of methods have appeared in the literature. Recently, the method that has stimulated the interest of the research community is the h-index method. The h-index metric corrects a lot of the older methods inefficiencies. But, the h-index presents a major disadvantage: it keeps increasing even if the researcher has stopped publishing good papers or if his present work is not that significant. In this paper, we present a generalization of the basic h-index, termed age decaying h-index to cope with this situation. The new index is able to identify the researchers and publication forums that are perceived as very significant for the latest years. We present an extensive experimental evaluation of the proposed novel citation index over the DBLP bibliographic database. The evaluation proves the robustness and virtues of the new index and reveals some very interesting results concerning the performance of scientists, the emergence of research topics and the preference of scientists to publish in new forums.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below