Knowledge Mapper

Produce alphabetical index for document repository using SWISH-E. Index files are analysed with WordNet to produce a theme list, which is used for searches to find documents. Theme words in documents are automatically hyperlinked to a list of references.

This project expands the capabilities of the SWISH-E search engine to include theme extraction or categorisation. This combines the best features of handcrafted catalogues with the benefits of a search engine. Users do not have to build and maintain complex knowledge bases of their own in order to establish relationships between topics, nor do they have to formulate complex queries to find information that may be worded differently in the documents being searched.

A number of spin-offs also came out of this project. The most important are:

• a CGI interface for SWISH-E written in C that does hit highlighting
• a lightweight HTML parser for converting HTML to plain text
• a web crawler that is used for downloading all web pages in the result set of a Google search
• an SGML parser for the Reuters-21576 document collection

Things on the to-do list:

• improve the accuracy of SWISH-E by adding synonyms for search terms from the WordNet thesaurus
• dynamic hyperlinking of documents