One of GBIF’s goals is to make biodiversity data, from whatever source, interoperable. In the current GBIF information architecture, one way this is accomplished is for data providers to install a web services software package called DiGIR (Distributed Generic Information Retrieval), which is a protocol for single point access to distributed data sources. DiGIR is an open source collaborative initiative, originally developed by The University of Kansas Natural History Museum and Biodiversity Research Center, The California Academy of Sciences, and The Museum of Vertebrate Zoology in Berkeley, all of which are GBIF data providers. GBIF is dedicated to the open source philosophy, and the free and open sharing of software developments and data.
DiGIR as initially given by GBIF to its nodes and data providers was designed to make databases containing data derived from a single source accessible via the GBIF network. It was not originally capable, without human intervention, of handling large databases holding multiple datasets stored in a single table. When the UK node, the National Biodiversity Network (NBN), began to work toward making its large database containing millions of records from over 80 data sources available via GBIF, this limitation of DiGIR became a problem. The Canadian Biodiversity Information Facility was one node that had already begun to tackle the problem, because they serve data from 50 sources through one large data management system, as well as from distributed databases.
Because developments by GBIF nodes are openly shared with others who need them, Canada made their workaround available to the NBN. However, the UK node did not have the programming skills needed to implement the system. Through the GBIF NODES committee's reporting system, NBN knew that the Expert Center for Taxonomic Identification (ETI) had the skills and contracting capability needed to help with an implementation. A quick fix could have been cobbled together, but ETI noted that other nodes were experiencing the same problem, and a more robust, longer term solution would be desirable. Canada suggested that Brazil's Centro de Referência em Informação Ambiental (CRIA) could contribute significant expertise.
Two of GBIF’s defining characteristics are its roles in providing technical information and in bringing together different partners to solve problems. In this case, the GBIF Secretariat helped NBN to understand how to fix the problem and then acted as an intermediary to establish an agreement between NBN and CRIA to implement the fix. The work carried out by this partnership has resulted in a new formulation of DiGIR that solves the problem of serving data from multi-source databases by enabling the use of generic local filters on each DiGIR resource configuration. This may also be used to filter sensitive data, non-validated data, etc, which may help promote further collaborative data-sharing, because it gives each data provider greater control over his/her data. The UK node and the Secretariat are now testing the modified DiGIR, making sure it is robust, so that this software package can be shared, as GBIF’s data are, openly and freely with others.
One solution to interoperability is to bring many data sources together in one large data management system. Another is to use a web services protocol like DiGIR to make individual, distributed databases interoperable. Thanks to the teamwork of NBN, CRIA, ETI, the Canadian GBIF node and the GBIF Secretariat, now it is possible to make multiple-source, large data management systems interoperable both with individual databases and with other multiple-source systems in a flexible, highly useful manner.
Acknowledgements: Lawrence Way, Dora Ann Lange Canhos, and Renato De Giovanni contributed substantially to the substance of this article.
|