280 likes | 353 Views
AGE-DECAYING H-INDEX FOR SOCIAL NETWORKS OF CITATIONS. D. Katsaros 1,2 , A. Sidiropoulos 2 , Y. Manolopoulos 2. 1 Dept. of Computer & Communication Engineering University of Thessaly, Volos, Greece 2 Dept. of Informatics Aristotle University, Thessaloniki, Greece.
E N D
AGE-DECAYING H-INDEX FOR SOCIAL NETWORKS OF CITATIONS D. Katsaros1,2, A. Sidiropoulos2, Y. Manolopoulos2 1Dept. of Computer & Communication Engineering University of Thessaly, Volos, Greece 2 Dept. of Informatics Aristotle University, Thessaloniki, Greece Presentation by:Apostolos Papadopoulos2 SAW 2007 Poznan, Poland
Methods for Ranking Scientists • Evaluation of scientists by “experts” • e.g., surveys • Citation Analysis • Task: Compute a score for the “objects” • Hybrid method of previous two SAW2007
Drawbacks of various scientists ranking methods • They do not measure the importance of papers • They are affected by “big hits” • They do not measure productivity • Need to set administrative parameters SAW2007
Terminology • Np: total number of papers • Nc,tot: total number of citations • Y(now): present year • Y(i): year of publication of paper i • C(i): set of citations to paper i SAW2007
H-index • Published by J.E. Hirsch in Oct. 2005 • Was known at least two years earlier • Definition:A researcher has h-index h if • h of his Np articles have received at least h citations each • the rest Np-h articles have received no more than h citations each SAW2007
H-index example Author X has 5 published articles: Article1, citations 5 Article2, citations 10 Article3, citations 100 Article4, citations 6 Article5, citations 4 The H-index of X is 4: there are 4 papers with at least 4 citations each. SAW2007
H-index • Calculates the broadness of a researcher • Productivity • Impact • Not affected by “big hits” • Not affected by “noise” SAW2007
Index a • Nc,tot≥h2 • Definition:A researcher has index a if Nc,tot=ah2 • Second metric-index SAW2007
H-index drawbacks • It is a growing function over time • Does not show scientist’s inactivity or retirement • Scientists with short scientific life are out of competition SAW2007
Contemporary H-index • Definition (in Scientometrics, 72(2), 2007):A researcher has contemporary h-index hc if • hc of his Np articles have Sc(i)≥hc • the rest Np-hc articles have Sc(i)≤hc • Sc(i)= * (Y(now) - Y(i) + 1)- |C(i)| • In our experiments: =4 and =1 • An old article gradually loses its “value” • Show how “active” a researcher is SAW2007
Trend H-index • Definition (in Scientometrics, 72(2), 2007):A researcher has trend h-index ht if • htof his Np articles have St(i)≥h • the rest Np-ht articles have St(i)≤h • In our experiments: =4 and =1 • An old citation gradually loses its “value” • Shows how “trendy” the work of a researcher is SAW2007
Age Decaying H-index • Definition:A researcher has Age Decaying h-index had if • hadof his Np articles have Sad(i)≥had • the rest Np-had articles have Sad(i)≤had • In our experiments: =4 and 1=2=1 • An old article gradually loses its “value” • An old citation gradually loses its “value” SAW2007
The second metric “a” of h-index • A factor a is defined as: • Nc,tot=ah2 • For h-index generalizations, instead of Nc,totwe use the sum of each paper score Sad(i): SAW2007
H-index generalizations • Contemporary • Trend • Age Decaying • Scientists, journals, conferences or any other kind of semantic grouping of articles SAW2007
Experiments • DBLP collection (http://dblp.uni-trier.de/) • Data timestamp: March 2006 • DBLP includes data for authors, journals and conferences • Focuses in the DB area • “Names Problem” is solved SAW2007
Experiments – h-index SAW2007
Age-decaying h-index: Findings • May decrease to depict outdated “entities”, contrary to the h-index which always increases or remains constant • Very sensitive to less modern and less popular “entities” (up/down per year) • able to provide more accurate image of the performance of the “entity” SAW2007
Conclusion • Evaluation of scientists based on citation analysis • Evaluation of publication forums based on citation analysis • H-index shortcomings: • Active – inactive scientists • Significant works in the past – not any more significant • H-index generalizations on the time dimension SAW2007
AGE DECAYING H-INDEX FOR SOCIAL NETWORKS OF CITATIONS D. Katsaros,A. Sidiropoulos, Y. Manolopoulos Thank you for your attention! Presenter: Apostolos Papadopoulos SAW2007