A Subject-Oriented Approach to Citation Prediction Model Through Web Metrics in Mendeley, Figshare, PLOS, Scopus systems

A comparative study of citation prediction model through web altmetrics (visibility, save and download, readers) in the fields of health science, life science, physical science, humanities and social science is the aim of this study.

The present study is a scientometric study that has been done with the method of citation analysis and web data analysis.   Sampling was done by random and stratified method. The Sample size was 2000 articles from 4 subject areas, the indicators of which were  extracted from Mendelian, Figshare, PlOS and Scopus systems and analyzed by Multiple Regression Analysis method.

The results showed that in four subject areas, web measures act as a predictor of citation indicator and there is a significant correlation between them. The extent of this correlation and predictive power depends on the subject area and covers a range of negative to positive correlations.

Conclustion:

 The difference between regression model of citation     prediction through web altmetrics in the variety of fields indicates the distinction among subject areas and their patterns in web metrics which should take in to account for assessments to avoid interdisciplinary    comparisons.  In the areas with powerful prediction model, web metrics can use separately and as an early predictor of citation. In other areas with weak prediction model, it is suggested that both metrics are applied for the best result.