Download patterns of journal papers and their influencing factors

A two-step cluster analysis was performed on the absolute downloads and the relative downloads of Chinese journal papers published between 2006 and 2008. Four patterns were identified from the perspective of absolute downloads; the first three patterns can be expressed as power functions, signifying their evident aging trends, although this does not apply to pattern 4. Two patterns were identified from the perspective of relative downloads, and both present power distributions with minor differences in decline speed. Furthermore, we delved into the relationships between total downloads and article features in varying patterns and found that there are only weak correlations between total downloads and title length, number of authors, and number of keywords. However, there are moderate to high correlations between initial downloads—defined as downloads made during the first year after publication—and total downloads, suggesting that it is possible to forecast total downloads according to initial downloads. Additionally, it was found that total instances of highly downloaded papers have no correlations with article features.