Federated Search

Streamlining Information Access

Federated Search: MultiSource Indexing, MetaSearch, and MultiSearch

To streamline information access, organizations need to consider agile integration methods that include technologies such as federated search. Federated search includes the ability to query multiple search repositories on behalf of the user, normalizing, and deduplicating results in such a way that they appear to come from a single search index.

The term "federated search" is used in several different ways within the industry. Most generally, it simply means search that combines information from multiple systems into a unified result set. Figure 1 shows different mechanisms to accomplish this.

multiSource Indexing Diagram

Multisouce Indexing:

The most common way of accomplishing an aggregated result set is to index the results into a single centralized search index, sometimes called MultiSource Indexing.

However, in many cases, indexing is not always possible, especially if the target information source is not under the control of the organization (e.g., external information sources such as LexisNexis, Dialog) or it is much larger or changing much faster than indexing can pragmatically handle (such as google.com or Twitter). In these cases, a federation technique called MetaSearch is the only option available to provide a unified view of the information.

metaSearch Federator Diagram


This approach uses a query federator to translate the original query, transmit it to multiple search engines, and combine the results as they come back.

Metasearch is familiar to people from the web, where comparison travel sites like kayak and tripadvisor are common. Many people also use this technique for desktop search with software like Copernic. The same technique applies to enterprise search. With Metasearch, there is no local index and the query federating component is completely independent.

multi Search Federator Diagram


A third technique, called MultiSearch, is a combination of MultiSource Indexing and MetaSearch. Some information sources are indexed centrally, and results from this index are combined with results from other search engines. MetaSearch can also use information from a local index to determine what other search engines to query and to normalize the results. Both MetaSearch and MultiSearch use query federation to incorporate information that can't be indexed. Understanding the best federated search approach is important when planning a unified information access strategy.