Abstract

As the number of Internet users grows, the problem of indexing and retrieval of electronic information resources becomes more critical. A number of search systems are currently available for this purpose on the Internet; examples are Lycos, Yahoo, Web Crawler, etc. However, they offer uneven search results, namely, many of them produce mishits or miss existing resources. This is due to the fact that they attempt to match the specified search terms without context as to where the words appear in the target information resource. This calls for a proper cataloging to avoid such uneven results. This thesis is concerned with a metadata-based indexing system proposed to describe the semantic content of information resources. The metadata description is called Semantic Header, and its main intent is to include those elements that are most often used in the search for an information resource. A variety of fields have been included in Semantic Header for the indexing and retrieval of resources. This will considerably reduce the aforementioned unpredictable results. The system also distributes the expertise of a librarian to help users choose appropriate subject terms, during indexing and searching, from an associated thesaurus. A prototype has been developed based on this proposal for indexing and discovery of information resources on the Internet. The prototype is composed of two main subsystems viz. Graphical User Interface (the client) and Database (the server). This thesis presents the design and implementation of the Database subsystem. Object Modelling Technique (OMT) has been employed for the analysis and design of this subsystem. Object Database and Environment (ODE) is the database system used for indexing and retrieval of Semantic Headers. Communication between the two subsystems has also been implemented using the TCP/IP protocol.