SMART is an implementation of the vector-space model of information retrieval proposed by Salton back in the 60's. The primary purpose of SMART is to provide a framework in which to conduct information retrieval research. Standard versions of indexing, retrieval, and evaluation are provided.
A secondary goal is to have an actual low-medium end information retrieval system available to users. I'm not sure how to define low end; the largest collection I've worked with was about 1200 Mbytes of newswire/other. Indexing speed was reasonable, from 120 - 250 Mbytes per hour. Retrieval speed was a second or two on a Sparc 1 for simple queries. It uses natural language queries.
SMART suffers from the advantages and disadvantages of most academic research software. It's designed to be extremely flexible (as long as you know what you're doing!). But it is correspondingly not strongly optimized for any one particular use. The code is relatively straightforward and should run (with possible minor modifications) on most UNIX systems of sufficient size. Our standard setup is a SPARC 1 with 12 Mbytes of memory. (For the large collection above, it was a Sparc ELC with 24 Mbytes). A standard disadvantage of much academic software is documentation, and SMART is no exception. It needs more.
SMART versions 10/11 are a complete rewrite of the mid and top levels of SMART version 8 (the last version we released). About 5 It's much more flexible and easily modified, and the various modules have much more potential for interaction than the old version. (Getting access to more information at retrieval time was a major reason for the rewrite).
Version 11 is not yet as fully featured as I would like, but it seems reasonably solid. New features are still being added, but we've now reached the point where the existing features are stable. Features still missing include Pnorm retrieval and clustering.
(end excerpt)
The first step in trying to use SMART would be to install it. If you are lucky enough to have a system administrator do this for you, happily skim trough this section and continue reading either the Technical documentation if you intend on using and modifying SMART to suit your own use, or the User documentation if you only intend on indexing documents and using SMART as it is now.