Making Serendipitous Connections via Research Publications

Universities throughout the UK are gathering information for the UK Research Excellence Framework in 2014 and as a result they are gathering large databases of information about their research activities, and about the research activities, interests and especially the publications of individual researchers within each institution. Many UK universities are using Pure (Publications and Research) software to manage this information. Heriot-Watt University alone has a database of about 15 000 published research papers within Pure from all of its research desciplines, and this database is still growing.

As part of the SerenA project, we are interested in making serendipitous connections between researchers. We have focused our attention on the kinds of information about publications which are typically available in Pure. Document references within Pure may be linked to external publications databases, including links to Digital Object Identifiers (DOI). The CrossRef consortium of publishers have recently made metadata available publicly as Linked Open Data for a large number (about 46 million) DOI records . Given the DOI reference for a document from Pure, it is possible to retrieve information about the document from this metadata– what kind of document it is, all of its authors, its publisher, journals and page references. Depending on the subscription status of the publications, the title, keywords, abstract and even full text are available for many documents. Individual researchers also have DOI author references.

As well as storing information about researchers’ publications, Pure stores a profile for each researcher. The profile includes basic information about names, titles, department, professional roles, and the individual’s areas of research interest. Researchers can describe their own interests in narrative text or by choosing from a menu controlled keywords. Researchers are linked to grant awards, projects and organisations. The profile can also hold links to the researcher’s identifier from major publications databases such as Thomson-Reuters’ Web of Science.

This gives us a route to identify connections between researchers. We have been experimenting with using information about publications, extracting annotations from the titles and abstracts of published papers using natural language processing analysis OpenCalais and Dbpedia Spotlight. These annotations can be combined with existing keywords to represent potential research interests of the authors, In future, as researchers put more information into their personal profiles and with the agreement of the researchers involved, similar techniques might be applied directly to their user profiles. Given these keywords and annotations, we can identify and link researchers who have similar interests, including those who are from different departments within the university and those who have not previously published together or worked on the the same research projects, and who therefore may not know each other already.

Pure offers an applications programming interface which makes it potentially possible to use the data within Pure to connect researchers within a university and across universities. Data from a university’s Pure database can potentially be made available as a web service, and it can be converted to web standard forms such as RDF. If universities choose to make the Pure API available internally then connections can be identified between researchers within a university, and this information can be linked to public sources of information to make connections with researchers in external institutions. If appropriate information from Pure were shared across institutions, then it becomes possible to identify connections between researchers from different institutions who have similar or complementary interests.

–
Diana Bental