SMART

This file is part of the course Information Retrieval from the faculty of arts of Tilburg University.

Tutorial

A step-for-step Smart tutorial is included for first-time users and other curious people. I hope Chris Buckley doesn't mind... :-)

Interfaces

Various attempts & scripts to access Smart by HTML

Here be dragons! Enter at risk.

Warning! Unexploded bugs!!!

Notes

There is good news and bad news. I'll start with the good news. SMART has been and still is the most interesting Information Retrieval system that has ever been built. You can use it to compute most variations on frequency-based term weighting. It incorporates the vector space model for document-document and query-document comparisons. It has an interactive search mode, field control, stemming and stopwords. SMART also is the most publicized IR system by far. All IR scientists of the last twenty years have used SMART in one form or another and have used it and still are using it for weighting and comparing words and vectors. The bad news is that the SMART system is difficult to use and that the documentation is scetchy. The documentation that comes with Smart consists (among others) of the following files:

overview: A description of the goals and organisation of the SMART system.
admin: A very concise description, based on complicated examples of the administration of the system. Good luck!
app.spec: Specification files.
defaults: Default values for various system parameters.
examine: How to examine the output and system files of SMART
retrieve: How to make an experimental retrieval run within SMART