Search suggestions (aka autocomplete or autosuggest) are a way of instantly suggesting to the user potentially interesting queries. However most forms of suggest are simply based on previous user queries; VirtualWorks constructs the terms in the suggest lists directly from the content of document database and thus shows the user what he can find even before the query is completely typed. An interesting example of this form of AutoSuggest can be found on www.jameda.de.
VirtualWorks has developed a framework called “Front Page Suggest” with which the complete content of millions of documents can be made “visible” in the form of a very flexible suggest module. This module can handle very large suggest lists together with a powerful approximate search functionality which can handle erroneous input very efficiently. There are also many ways to configure the presentation and the ranking of the suggest terms to fit a large variety of application types. Combined with VirtualWorks extraction capacities, it is possible to set up a suggest module on the basis of the content of the indexed pages in a matter of minutes even for very large data sets. Contrary to just about all other suggests, the suggest terms are dynamically extracted from the document content. For example, the construction of a suggest index for a large scientific document database comprising around 60 million documents is performed in less than two hours on a single machine!
VirtualWorks has also devised a technology which allows for the suggestion of combined suggest terms thus allowing the effect of drill-downs even during querying!