Data Citation Practices in the CRAWDAD Wireless Network Data Archive

Abstract

Abstract. CRAWDAD (Community Resource for Archiving Wireless Data At Dartmouth) is a popular research data archive for wireless network data, archiving over 100 datasets used by over 6,500 users. In this paper we examine citation behaviour amongst 1,281 papers that use CRAWDAD datasets. We find that (in general) paper authors tend to cite datasets in a manner that is sufficient for providing credit to dataset authors, but also providing access to the datasets that were used. Only 11.5% of papers did not do so; common problems included (1) citing the canonical papers rather than the dataset, (2) describing the dataset using unclear identifiers, or (3) not providing URLs or pointers to datasets.

How research data is cited in scholarly literature: A case study of HINTS

“…The percentage of data citation (11.6%) is much lower than other studies. Mooney's () study, which analysed data citations using bibliographies from ICPSR in social sciences, reported 29%, and the study by Henderson and Kotz (), which analysed scholarly articles used Community Resource for Archiving Wireless Data At Dartmouth (CRAWDAD), reported that 88.5% cited a dataset. However, it was observed that the percentages of HINTS data citation improved from 5.6% (2005–2008) to 15.2% (2013–2016).…”

“…Several studies demonstrated that researchers use a wide variety of ways to refer to datasets (Callaghan, ; Henderson & Kotz, ; Major, ; Mooney, ; Pepe, Goodman, Muench, Crosas, & Erdmann, ; Zhao et al, ). Belter () supported this by noting that 377 variant formats of citing the same data were found.…”

“…Another issue suggested by previous studies involved whether or how frequently researchers cite datasets in their publications. It was suggested that most of the scholarly literature provided citations to data‐related publications rather than actual datasets (Mooney, ; Piwowar, Carlson, & Vision, ).While citing related publications helped understand the origin of data and acknowledge the originator of datasets, it did not enable readers to access the data (Henderson & Kotz, ). The portion of articles that provided citations to datasets differed in existing studies, ranging from 39% to 88.5%, depending on the characteristics of data and the methods of sampling scholarly literature.…”

Theory and practice of data citation

“…To this end, a fundamental requirement for citation methods is to automatically create citation snippets in order to ensure consistency of references (Thorisson, 2009). Indeed, recent studies about data citation practices showed that data citation snippets are often underspecified, with vital information missing (Mathiak & Boland, 2015), and may present several errors compromising the identification and retrieval of the data being cited (Henderson & Kotz, 2015). Automatic methods for building citation snippets are also required because humans cannot remember or know what the necessary information is, when a snippet is complete, what the citation format requires in a given context, and where to gather the relevant information, especially in a big data context where manual exploration of data is unfeasible (Buneman et al, 2014;Crosas et al, 2015;Jagadish, 2015;Minister, 2012;Peters, Kraker, Lex, Gumpenberger, & Gorraiz, 2016).…”

“…This system needs to be extended/revised to handle the citations of big dynamic data in order to be able to cite a subset of data on 1) selected variables and observations for large quantitative data; 2) time-stamp intervals; and, 3) spatial dimensions and also to be able to handle queries to the data. The use of DOI raised some concerns about the cost of minting identifiers for large data archives (Henderson & Kotz, 2015). Buneman and Silvello (2010) and Silvello (2017) discussed the case of XML data, where every single node of an XML file can be a citable unit.…”