A new kind of search engine
MKSearch is a research project to develop a metadata search engine. The system is composed of two linked systems; an indexing Web crawler and a public query interface. The indexing component extracts Dublin Core metadata from Web documents and stores them in RDF format. The query interface matches documents in the index using an RDF query language and can return the results in a variety of formats including standard HTML and as a standing RSS feed.
Project status
The project has completed an alpha proof of concept stage and is approaching a first beta release. The primary system components have been developed, integrated and tested from end to end using test data, and indexing this site. Some final refinements are being made for the system to complete test deployment on Fedora Core 4.
The MKSearch system is being developed using the Java programming language and is licenced under the GNU General Public Licence. All software is compiled and tested using both the Sun and GNU Java compilers. All project source material is available through the public MKSearch Subversion repository.
To contact project members, please join the MKSearch mailing list. The MKSearch Java documentation is periodically updated in the project's Subversion repository.