Experience with transformation of bibliographic data into linked data

Abstract

Academic Bibliography Database of the University of Economics in Prague consists of bibliographic records of publications involving journal articles, conference papers, lecture notes, monographs and monograph chapters created by the academic staff of the university. We would like to discuss the experiences gained in the process of this dataset’s transformation from its current data format to RDF-based data. During the course of conversion we will specify the entity types in our data and choose a way to model them. For data description we are going to use the most popular and widely implemented vocabularies and domain ontologies. Next, we will discuss the issues associated with interlinking the data with relevant and well-established datasets constituting a part of the web of data (e.g., DBLP Computer Science Bibliography). We would like to examine further possibilities of data re-use by making the data suitable for other applications such as citation managers (e.g., Zotero) to generate citations automatically. We are also interested in providing users with the information about availability of a resource as full text or as library holding, so we discuss cooperation with link resolvers (SFX, in our case) as well as other possible ways of linking. In this way, aligned with the concept of the semantic web, we maintain that the potential of the data would be maximized, as the information in bibliographic records would become easy to share, more visible due to incoming links, and more ready to be processed by web applications. We argue that academic bibliography data should be openly available, so it increases transparency of publishing activity of the university’s academic staff. Beyond being re-usable, the data can be easily linked to so that they must not be duplicated at multiple locations. Likewise, these features together can provide a few benefits for bibliometric evaluation of science. The adoption of linked data publishing model for academic bibliography datasets involves moving to a more flexible data format for bibliographic data. While linked data is seen as a pragmatic implementation of the semantic web, we argue that it is not possible to implement the library data vision with the MARC, current standard for bibliographic data. Obviously, there is a need for more web-compatible and web-friendly data model. With this in mind, we suggest linked data as a part of the pragmatic implementation of the vision for bibliographic data on the Web.