Dynamic Patterns of Academic Forum Activities

Abstract: A mass of traces of human activities show rich dynamic patterns. In this article, we comprehensively investigate the dynamic patterns of 50 thousands of researchers' activities in Sciencenet, the largest multi-disciplinary academic community in China. Through statistical analyses, we found that (i) there exists a power-law scaling between the frequency of visits to an academic forum and the number of corresponding visitors, with the exponent being about 1.33; (ii) the expansion process of academic forums obeys the Heaps' law, namely the number of distinct visited forums to the number of visits grows in a power-law form with exponent being about 0.54; (iii) the probability distributions of time intervals and the number of visits taken to revisit the same academic forum both follow power-laws, indicating the existence of memory effect in academic forum activities. On the basis of these empirical results, we propose a dynamic model that incorporates the exploration, preferential return and memory effect, which can well reproduce the observed scaling laws.