Abstract:

Taking papers from 2006 to 2008 in the field of library and information science journalsasresearch objects, this paper conducts a twostep cluster analysis on the downloads and citations of every paper. Then, the correlation between downloads and citations in various patterns has been studied in a dynamic perspective. It has been found that there is a linear relationship between cumulative downloads and cumulative citations, which increases with time. Stronger correlation between downloads and citations has been observed in the pattern with the most and fewest average downloads, and this phenomenon is more prominent at the early stage of publication. Similar findings were found in different citation patterns. The result suggests that downloads and citations tend to be unified in the evaluation of the use and usefulness of articles and readers' download behavior is likely to convert into citation behavior; the highest and lowest downloads of papers can be used as a predictor for the latter citations, while isn't applicable in papers of medium downloads.

Key words: Citation, Downloads, Correlation, Library and information science, Download pattern, Citation pattern