SARA (SGML Aware Retrieval Application) is a software system developed specifically for the British National Corpus. The first release of the corpus includes only a part of the SARA system, specifically a pre-compiled word index to the corpus, and the SARA server program. This program is designed to receive queries from another program (a client program), and uses the word-index to identify parts of the corpus which satisfy them. The query results are then passed back to the client program, for any further processing required. The language in which queries from the client and responses by the server must be expressed is known as the protocol; a definition of the SARA protocol is included with the documentation. A competent programmer should have no difficulty developing a SARA client for any modern computing environment.
The server program distributed with the first release of the BNC contains a number of system dependencies, which make it unsuitable for use on systems other than SunOS. A new version of the server which has been tested on a wider range of machines is now available to all licence holders free of charge.
Instructions on installing and configuring a SARA server are now available. Please read these before attempting to use the distributed SARA release.
The present release includes a very simple client program called solve. A more ambitious and user-friendly client program for the MS Windows environment is currently on beta test.
You can also read a brief technical description of the SARA system.