Exploiting and improving institutional repositories for bibliometrics

Institutional repositories have a dual role, being both open archives, disseminating without barriers research results financed by public funds, and a directory of institutional scientific publications. Today, a new kind of functionality may be added to these repositories : they allow to evaluate University's scientific production through bibliometric analysis. Indeed, two special types of metadata are included into institutional repositories: bibliographic and administrative information (attribution of publications to each author or research unit from the institution for instance), making them particularly interesting for bibliometric use. It is common to make use of international and subject databases to feed and improve the content of the institutional repository. These mines of bibliographic information are currently used to import references into archives, individually or in batch. Using Web services and AJAX techniques, they are also exploited during data entry: with a simple identifier, bibliographic metadata is injected into a web form or proposed while typing. At the same time, bibliographic metadata is enriched with unique identifiers that connect the archive with external databases. These identifiers allow the addition of deep links to external databases from the archive or university web pages display. In bibliometrics, these identifiers also play a major role because they allow information import and update. Indeed, for a journal article, the ISSN makes the inclusion of journal impact indicators easier. Other unique identifiers can be used to import the citations count received for a document, calculated by different databases and open archives: UT for the Web of Science, PMID for PubMed Central and PubMed and DOI for CrossRef and SCOPUS The use of data from an institutional repository as a source of bibliometrics analysis has two advantages. On the one hand, this requires different units of the university to agree and work together to improve the archive's metadata and keep it up to date. On the other hand, the effort invested in sorting, selecting and assigning publications to authors and research units may be exploited by the university evaluation services in order to solve ambiguities generated by homonyms that are common in general databases and makes the bibliometric analysis complicated. Since the creation of Lausanne University's open archive "SERVAL" in late 2008, we are working in this direction at the Faculty of Biology and Medicine. Currently, the data from SERVAL is used to calculate bibliometric indicators for faculty research units' evaluation. Up to now, these indicators only took into account the impacts of journals in which researchers publish. We are currently working to establish a new system that allows the archive to be enriched with the number of citations for each item. This information will allow us to calculate the h-index and use it as new indicator in our bibliometrics. Managing this new kind of information requires to improve the current system and to make modifications at two levels: upstream, completing and correcting metadata from the archive; and downstream, processing and enriching data with citations information. Indeed, bibliometric indicators are changing rapidly, particularly the number of citations that needs to be updated frequently. The aim of this project is to create a new flexible system that can be integrated with other institutional repositories. We believe that the use of these data for bibliometrics is an opportunity to combine forces around the archive and fulfill even more important functions within the institution