| Popis: |
Research standard and quality should be continuously monitored to direct progress of science in right direction. With exponential growth and continuous expansion in citation network, manual and static analysis is becoming insignificant. To fill in the gap, application of machine learning models might prove to be useful. In this paper, we propose some of the problems that we intend to solve using machine learning. Among various applications outlier analysis for early detection of anomalies in citation network, long term prediction of high impact and seminal authors, papers and field of study, deriving inherent features on diverse temporal and demographic scale governing citation structure etc. Starting with empirical analysis of open academic graph dataset, we try to understand the complex relational citation structure of entities. As a preliminary step, we do time series clustering of citation data and study characteristics of diverse profiles of citation curves. When compared to static classification in past literature, we overcome drawbacks of past study and get better insights. |
