Unravelling Federated search

Some sort of knitting image!
Martin White, who is well known for being an expert in the fields of search, information management and intranets has written a short but very useful piece on federated search.

For anyone not familiar with the concept, Federated search is a technology that allows the simultaneous searching of multiple resources. So a user will enter a single query request into a search engine, this is then sent to the search engines which are part of the federated search. Finally the federated search then aggregates the results that are received from the search engines.

Federated search has been used for many years by law firms and other professional services firms as like most organisations they will have multiple information resources (databases) that users need to access, but don't want to have to search these resources individually. This is one of the main reasons Federated search has proved popular, although Federate search is not without its own issues.

In Martin's article which have been published to the CMS Wire site he looks at some of the challenges associated with Federated search and how to approach them. So the major issue with Federated search, at least in my opinion, is how you present a significant number of results to users in a away that will make sense to them. Martin outlines two ways in which this can be achieved, which are as follows:

  • The first option is to crawl and index a number of individual search engines within a single application. This is often the easiest way to faciliate Federated search, but it does mean as Martin says that users are often presented with very long lists of results, which may make no sense to them.
  • The second option outlined by Martin is to use a search application to manage the query entered by the user. This sends the query out to other search applications and results are then either integrated, or presented in a number of different sections on the results page. However this option requires the use of connector, which conver queries into a readable format for earch of the search applications and returns the results. This is the more complex approach to Federated search.

Martin then looks at some of the challenges of managing Federate search using the two options outlined in his article, as well as a 3rd option, which is to not use Federated search at all, but to improve the search tools that exist within the organisation. This is a very interesting article, which if you're looking at Federated search is well worth reading to give you an ideas as to some of the challenges you're likely to face.

I have also previously written about Enterprise search in a blog post called "Choosing and using an enterprise search tool" and "A short introduction to Enterprise search"