Research networking tools use data-mining, social networking and semantic web approaches to enable expertise discovery, matchmaking and more. Several research networking platforms have been built--the commercial product SciVal Experts (formerly Collexis Expert Profiling), U Pittsburgh's Digital Vita, Harvard Catalyst Profiles, U Florida's VIVO, U Iowa's Loki, Stanford CAP, and others--and many additional institutions have also deployed at least one of these products. On August 6th 2010 the CTSA Research Networking Group facilitated a day-long meeting at the University of California, San Francisco (UCSF) with CTSA and non-CTSA participation including leads of major research networking tools, to discuss interoperability between products and institutions and to design a pilot for a federated national network. As a result of the meeting, initial guidelines for a pilot national network were established and a 'pilot project' to create the first test instance of the pilot network was kicked off.
The design of the pilot network addresses a simple use case--searching for a potential biomedical research collaborator across multiple institutions in a way that provides value over existing methods, such as Google or Facebook. The pilot was further defined by our belief that individual institutions can provide "cleaner" and more complete data about their own researchers by combining external sources of data with their own local databases. Therefore, we decided that the focus of our pilot network would be to generate buy-in from institutions so that they will be both willing and eager to share their information and encourage their researchers to adopt the tool.
Against a backdrop of allowing individual institutions as much control as possible over the data shared and the user experience, the network, as currently envisioned, has two components that define it: a) a technical architecture and common interface and b) an agreement between participating institutions. The agreement between institutions emphasizes that upon the execution of a federated search, any participating institutional website will only show the number of 'hits' a search term produces at each of the pilot participating institutions, and a URL back to each institutions local research networking website. In this way, participating institutions leave the presentation of research expertise to each individual institution, while still allowing for the benefits of a search across institutions. The technical design of the pilot network achieves the requirements of the pilot agreement by using a federated architecture with users initiating searches from within the framework of their local institution's website. Key aspects of the architecture are that there is no central database, search index, or website; a global ranking algorithm is not needed; institutions can define which populations to load into their databases, what a search "match" is, and how to sort/rank people within their institution; and institutions may remove themselves from the network at any time.
The DIRECT2Experts network is open to any institution that is interested, while the initial pilot project (proof of concept) requires that institutions have a mature, deployed research networking tool and can commit resources to assessments. We recognize that once we reach a critical number, followup meetings will be necessary to discuss governance, scalability, user experience, data quality, and other issues. The CTSA Research Networking group is the facilitator of this process, but participants also include non-CTSA institutions.