Structure of Rumour

Indexes involved in Rumour:

Local Index (Lindex) - which contains the location of documents that are accessible on the Web server, with their keywords.

Referring List - which is a collection of Cindexes from other servers. This Referring list will provide "suggestions" for the search program.

Collectible Index (Cindex) - which is a shorter version of the Lindex. It includes such things as the keywords that are used in Lindex, and the Web server name.

Programs in Rumour

Search Program - This is an individual Web program, that has a common gateway interface (cgi). This program is designed to search the Lindex and the Referring List from the same package.

Jack - This is a robot program that traverses the local file system to get pre-defined keywords in the .html or .htm files in prescribed directories or home directories that exists in the password file.

Collector - This program will take a file of Web servers and request the Cindex from each of these Web servers.

Get Index - This program is used to request documents on the WWW.

Search form - This is an interface form that invokes the search program when it is completed.

The Rumour package should be accessible from the server's document root, therefore the user can access the Rumour Search form using http://server.name.edu/Rumour/. That means either the Rumour package is installed at the server document root, or there is a link from the document root to another directory. This is done to unify access to Rumour.

Most of Rumour's components should be contained in the Rumour directory, with the exception of the Search Program which should be copied into the server's cgi-bin directory because of security considerations.