DynMaps

Master's degree project: Dynamic Construction of context-maps for Netscape's Open Directory Project

Background

Context-maps are a mix of mind-maps, UML-diagrams, concept-maps, topic-maps and other related graph-like presentations of knowledge. A special tool - a concept-browser called Conzilla - that supports navigation through an atlas of context-maps has been developed by the KMR-group over the last 3 years. Conzilla is intended to provide a user-friendly clean/extendable/personalizable environment. Originally we developed the idea of a concept-browser from a learning perspective, but it has proved to be very useful within other areas such as e-commerce, e-administration, organizational modeling etc. One of the problems with the current web is that its information is largely structured purely for human consumption. Hence this information is hard to extract from web pages overloaded with graphics and layout. W3C and the original creator of the web, Tim Berners Lee, strongly advocate RDF and what nowadays is called the semantic web as a solution to these problems. Even though RDF was suggested already in 1997, it is first within the last two years that it has gained real momentum. It now appears likely that there will be a "second-generation web" dealing more with meta-data, where search-engines and personal agents will thrive. We strongly believe that this web of well-structured, machine-readable information will have many other important impacts on society besides smarter agents, perhaps mostly noticeable within the learning community. Conzilla can work directly with the semantic web - both as a presentational tool and as an editor, which (to our knowledge) is a unique capability. Context-maps are loaded and saved as RDF expressions and can be thought of as presentational information added on top of any other information expressed in RDF. This means that in a not too distant future, when large parts of all information is accessible in RDF-format, we can present and extend this information in a multitude of different ways.

Currently all our Conzilla-maps are created manually, and since construction of maps is a time consuming activity, not that many maps have been made outside our own favorite areas. To remedy this situation, we suggest a server side creation facility for of maps based on some accessible knowledge.

The project

We would like a server side solution for dynamic creation of context-maps. In more detail:

• The information from which to create the context-maps can reside either in a local database on the server or elsewhere on the net.
• The server-solution should include a reasonably 'easy' configuration method for new knowledge sources of known types. The solution need not cover more than knowledge sources of RDF-type, but it would be nice with code that is easily extendable.
• For graph layout, existing packages such as GraphViz (AT&T) can be used.
• Netscape's Open Directory Project provides a nice RDF-based catalog structure, which is the world's largest human-edited index of the web (over 40 000 voluntary editors). This is an excellent knowledge source that would bring much value if it were accessible in the form of context-maps. Since other large search-engines (Google, Netscape search, Lycos, etc) use the same catalog-structure, it is very probable that these sets of context-maps can provide entry points to very precise searches (see our other suggested project: Using searches to fill concepts with content).

Project environment:

• We strongly believe in the open source concept and we would therefore like the produced code to be licensed under GPL (GNU Public License).
• For reasons of compatibility with earlier work, we prefer that the code is written in Java.
• Most of our experience is with Linux environments, and the server environment we can provide is a Linux machine running Debian. Hence we would prefer some experience in Linux.