Building a customized federated search engine for health sciences

The Arizona Health Sciences Library (AHSL) developed their own federated search application. The Journal of the Medical Library Association (JMLA) published a paper about the development of the application.

The paper’s abstract briefly describes the project:

The AHSL provides services to the University of Arizona’s (UA’s) health sciences programs and to the University Medical Center. Librarians at AHSL collaborated with UA College of Medicine faculty to create an innovative search engine, Evidence-based Medicine (EBM) Search, that provides users with a simple search interface to EBM resources and presents results organized according to an evidence pyramid. EBM Search was developed with a web-based configuration component that allows the tool to be customized for different specialties.

What’s interesting about the AHSL federated search library project? A number of things:

1. It appears that AHSL built their own search framework (connector language, search submit mechanism, results parser, …)
2. AHSL built their own user interface as well. They used Cold Fusion with a Microsoft SQL server database backend. I don’t read much about organizations building their own federated search user interfaces (or applications) with any software, much less so with those components.
3. AHSL documented their thought process in building their application. This kind of information is pretty rare.
4. They built a federated search application around needs specific to their industry: search of evidence-based medicine, and the evidence pyramid.
5. They paid great attention to delivering relevant results as their users, physicians and clinicians, are frequently rushed and would not tolerate many non-relevant results.
6. They created “a framework that would allow librarians or end users to develop specialized search tools with little understanding of the mechanics of the [application internals].”
7. The search system post-pends user query terms with search filters that have been developed by a reference librarian with input from a physician in the emergency department.

The article is not very long, just a handful of screens. But, in that limited space it packs in a good amount of detail. For those considering building their own federated search applications, especially in the emergency medicine field, the JMLA article provides good food for thought. What is missing, though, is detail about how they implemented the federated search technology, i.e. the back-end stuff that is hidden from the user. Also, I think the article should make the point that doing what AHSL did was not easy; it’s not something I’d try at home.

Tags: federated search