A Data-Driven Unified Framework for Predicting Citation Dynamics

Abstract

With the rising interest in predicting the scientific output, various efforts have been made to predict a scientist's h-index or the citation trajectory of a publication. In this work, we employ a dynamic categorization for scientists to ensure at each stage of their careers a comparison amongst their peers and combine this grouping with predictive models to estimate a scientist's future impact, as expressed by citation counts. Moreover, we investigate a wide range of factors identifying their importance in determining the future of science for different performance and academic levels with particular emphasis on features describing a scholar's position in multi-layered collaboration and citation networks. The robustness of the approach is examined on a longitudinal dataset centered around 700,302 data points representing Computer Scientists in various time periods with their complete networks of over 18 million collaboration links and 36 million citations. Our results indicate up to 30% improvement in prediction performance compared to baseline methods along with an average R 2 =0.96 for short term and R 2 =0.91 for long term predictions.

Modeling a successful citation trajectory structure for scholar's impact evaluation in Israeli academia

“…For the prediction task, they automatically generated 44 features for each scholar, such as publication and citation scores, h-index , and changes in citations, h-index , and mean number of citations per publication over the last two years, and used them in several regression models. Gogoglou and Manolopoulos [ 12 ] created clusters of academic peers according to their citation scores and predicted the future impact of the scholars that belong to each group based on several features such as: the publication and citation rates, co-authorship, citers and cited authors, differences in citations and h-index for every ten-year sub-period. They found that the position of the scholar within the social network of citers and cited authors is crucial for future impact.…”

“…Studies that compared scholarly productivity across disciplines found that the publication rates of natural scientists exceed those of social scientists and humanists [ 8 , [52] , [53] , [54] , [55] , [56] ]. According to Gogoglou and Manolopoulos [ 12 ] and Sarigöl et al [ 57 ], social links and networking of scholars may have effect on their future impact. In this study, our goal was to develop measures of scholarly impact that are not susceptible to biases toward certain fields, seniority levels or source citation index.…”

“…As has been shown in previous research, there is a wide diversity in publication, citation and h-index scores even among prominent scientists, such as tenured professors [ 8 ]. These scores strongly depend on the scholar's field of study, source citation index, age, seniority, networking, and other factors [ 1 , [8] , [9] , [10] , [11] , [12] , [13] ]. Hence, several universal impact measures have been proposed that produce normalized scores comparable for scholars from various research fields and career stages [ 2 , [14] , [15] , [16] , [17] ].…”

“…Several recent studies applied multi-feature models to predict the future citation number (or h-index ) of a scholar based on his/her past performance [ 12 , 25 , 26 ]. Slope-based trajectory analysis was formerly used in citation prediction and knowledge diffusion of academic papers [ 27 ].…”