Introduction

The World Wide Web (WWW) [6] [5] is an open public database. As part of the Internet, it contains information on many topics, ranging from technical matters to entertainment and news. With the Web increasing in size every day, it is becoming a difficult task to find specific information.

Database services have been set up to provide a solution to Web users. They include the following :

1. Meta-documents divide the WWW into section, grouped by topic and providing different types of search operations on selected databases.
2. Robot-based searchable indexes cover almost the entire WWW.

These services are not problem free. The Web is a highly distributed system, but these database services are centralised at one or a few sites on the Web, therefore creating a speed and access problem. A distributed database system would be more appropriate.

This paper introduces a new indexing and searching system to the Web. Called "Rumour", the system includes tools to create local indexes and collect indexes from different servers, and provides a new search process. The interesting feature of this system is that it is distributed and local to the related server.

Rumour was designed to be installed on a currently running Web server. It can also be incorporated into a Web server package, although it needs the support of a proxy Web server. It is capable of updating itself regularly by running the update program, and does not waste Internet resources whilst it is updating.