Posted, visited, exported: Altmetrics in the social tagging system BibSonomy. Issue 3 (August 2016)

• Record Type: Journal Article
• Authors: Zoller, Daniel; Doerfel, Stephan; Jäschke, Robert; Stumme, Gerd; Hotho, Andreas
• Is Part Of: Journal of informetrics. Volume 10:Issue 3(2016:Jul.)
• Page Start: 732
• Page End: 749
• Publication Date: 2016-08
• Subjects: Altmetrics -- Scholarly impact -- Social bookmarking -- Collaborative tagging; Library statistics -- Periodicals; Information science -- Statistical methods -- Periodicals; Bibliometrics -- Periodicals; Bibliothèques -- Statistiques -- Périodiques; Sciences de l'information -- Méthodes statistiques -- Périodiques; Bibliométrie -- Périodiques; 020.727
• DOI: 10.1016/j.joi.2016.03.005
• Languages: English
• ISSNs: 1751-1577

Abstract

Abstract : Highlights: We examine various usage features of a scholarly bookmarking system as indicators for impact of publications. Users of a scholarly bookmarking system are biased towards using publications with higher (citation) impact. Usage metrics do not only correlate with total citation counts, but also with counts of citations in the future. A bookmarking system's most inherent feature – tagging – is suitable for identifying topic subsets of publications where usage and future citations exhibit higher correlations. The traces of user behavior, found in a web systems logs, bear predictive power over future citations. Abstract: In social tagging systems, like Mendeley, CiteULike, and BibSonomy, users can post, tag, visit, or export scholarly publications. In this paper, we compare citations with metrics derived from users' activities (altmetrics) in the popular social bookmarking system BibSonomy. Our analysis, using a corpus of more than 250, 000 publications published before 2010, reveals that overall, citations and altmetrics in BibSonomy are mildly correlated. Furthermore, grouping publications by user-generated tags results in topic-homogeneous subsets that exhibit higher correlations with citations than the full corpus. We find that posts, exports, and visits of publications are correlated with citations and even bear predictive power over future impact. Machine learning classifiers predict whether the number of citations that a publication receives in a year exceeds the median number of citations in that year, based on the usage counts of the preceding year. In that setup, a Random Forest predictor outperforms the baseline on average by seven percentage points.