关键词: 大数据, 跨学科性, 中国, 美国, Rao-Stirling指标

Abstract: Due to the advancement of science and technology, and the complexity of social issues, research integrating knowledge from different disciplines becomes increasingly common. As a symbol of innovation and creativity, interdisciplinary research is considered to have great potential success in achieving breakthroughs, generating outcomes and addressing societal problems. All these help to rejuvenate science and promote its ongoing “health”.Big data research, for example, is such a typical interdisciplinary field involving various domains such as science, engineering, medicine, healthcare, finance, business, public administration, and so on. Because of its rapid and widespread application, big data research has also attracted growing attention from academic communities, as can be reflected by the enormous publications in recent years. In the context of the importance of big data development which has already been part of the national strategy in many countries, and the diversity of its research areas, this current paper, based on publications indexed in the Web of Science (WoS) during 2009-2016, examines and compares the interdisciplinarity of big data research in China and the USA by applying bibliometric methods and visualization tools. Perspectives include interdisciplinarity measured as Rao-Stirling Index, distribution of core disciplines, relationship between interdisciplinarity and citation impact, as well as discipline maps based on journals in publication references. The reason for applying the Rao-Stirling Index in this study is because it takes into account not only variety and balance but also distance between subject categories, and has been widely used in interdisciplinary studies. The current study contributes to a more comprehensive understanding of the interdisciplinarity of big data research in the two most-productive countries in the world, and may shed light on related efforts.The results confirm that big data research involves a wide range of fields as expected. Combining all publications of China and the US, on average one paper cites publications from six disciplines-respectively five and seven for publications of China and the US. Although the interdisciplinarity of research in both countries is increasing with time, that of the US is always higher than that of China in the studied period. In addition, the US publications not only cover more disciplines but have a more balanced disciplinary distribution. Main disciplines involved in big data research include Mathematics and Computer Science, Biomedical and Health Sciences, as well as Social Sciences and Humanities. To be specific, the vast of the references of the US publications belong to sub-fields such as multidisciplinary science, computer information systems, biochemistry and molecular biology, electrical and electronic engineering and artificial intelligence. By comparison, publications of China cite fewer disciplines with Mathematics and Computer Science as the main fields, inclusive of electrical and electronic engineering, artificial intelligence, computer information system, theoretical methods and software engineering. In recent years, the USA becomes growing active in applying big data with studies relevant to social sciences and humanities, where big data is considered as an efficient tool to solve practical problems in social economy development. Meanwhile, problems and challenges brought by big data are highlighted, for instance, data security, privacy, copyright, protection, media information integrity and so on. Nonetheless, the dominant share of Mathematics and Computer Science references in both the US and Chinese papers reveal that these disciplines are core knowledge foundation in big data field. Based on the clustering and visualization functions of VOS viewer, the cited journal maps of both China and the US can be displayed, providing more intuitive and comprehensive view of the distributions of involved disciplines. With regard to the relationship between interdisciplinarity and citation impact, positive correlations has been found in our study. In other words, the more disciplines referenced, the more citations a paper may receive. One of the possible explanations is that, more and wider sources of knowledge contributing to publications may attract larger group of audience, and thus may result in higher citation impact. Nevertheless, such citation benefit differs within the two countries: significant correlation exists between interdisciplinarity and citation impact of the US papers, but it was not the case in China. According to the findings above, this research provides the evidence that, in big data research, USA authors tend to refer more literature from more disciplines which fall within both basic and applied fields. In contrast, the Chinese researchers prefer to absorb knowledge form basic disciplines, typically as Computer Science, which contain core technologies and algorithms, and play a decisive role in promoting the progress of big data technology. To some extent, this situation can be interpreted by the gap in the development level of big data as well as the dissimilar strategic targets of the two countries. To sum up, the results evidently specify that China and USA have a lead on big data research in the world as reported by the proliferating publications and expansion of disciplines covered, noticeably indicating the clear signals of the growing interest and focus of both countries. Despite of the faster growth rate of Chinese publications, the USA performs better in interdisciplinary research measured by Rao-Stirling index and obtained more citations. It is for this reason that we suggest that, apart from focusing on the development of technologies, Chinese researchers should be encouraged to break down disciplinary barriers and explore more collaboration opportunities with scholars from other fields, especially those from non-representative disciplines, in order to combine more knowledge sources and improve the breadth and depth of big data research.With regard to limitations of the current research, we would like to mention the following: First, only document types of journal article and review indexed in WoS are used, the results of the current study do not represent the situation when conference proceedings and patent data are included. Coming to the interdisciplinarity indicator Rao-Stirling index, although it combines three dimensions, it is too general and is hard to distinguish and compare among different dimensions. Furthermore, the indicator is based on subject categories of journals, whereas such field classification itself is problematic because a journal may involve more than one fields, thus is difficult to classify multidisciplinary journals. Further study for a better measurement of interdisciplinarity is necessary.

Key words: big data, interdisciplinarity, China, USA, Rao-Stirling index