Google SoC 2008 Ideas
Idea 1: Conxilla: A web-based GUI for collaborative conceptual browsing
How could a concept browser like Conzilla work together or be embedded into a traditional web browser? There is a basic version of a web-based Conzilla available, called Conxilla. However, this is just a proof of concept for showing and embedding already published (from within Conzilla) into web pages. Conxilla does not have any UI or specific integration into browsers yet.
The idea of this project is to first investigate existing web based conceptual mapping solutions to get an idea of the requirements and possibilities of such a web-based GUI. The focus should lie on usability and user-interface design.
Realization and results
The outcome of this project should be a functional prototype of Conxilla using widgets of an appropriate AJAX toolkit (e.g. Dojo, this has been used for a first and very basic proof-of-concept implementation).
The GUI should have the following features:
- Toolbars or buttons to easily access frequently used functionalities
- Colors and/or icons to indicate special kinds of concepts (links, attached content, reused in a different context, etc)
- List of available contributions
- Metadata presentation (this can be done using the AJAX version of the SHAME framework)
- Conzilla concept browser
- Collaborilla collaboration server
- SHAME metadata editor and presenter, AJAX version
Other slightly related projects can be used for getting ideas for a user interface, e.g. Zitgist, CmapTools, Compendium.
Idea 2: Connecting up Conzilla and DBpedia
Increasing amounts of semanticised data can be found on the Internet. One popular and open example is DBpedia, a project to provide semantic data in addition to Wikipedia articles. This repository could be integrated into Conzilla in a way so that DBpedia identifiers are reused and the concept descriptions (the metadata) are fetched from the DBpedia system.
An integration into Conzilla could look like this:
- A basic search interface for finding DBpedia concepts which are then turned into Conzilla concepts and used for conceptual modeling.
- DBpedia metadata is loaded and displayed inside Conzilla (this can be done using the SHAME framework).
Idea 3: Conceptual indexing for building global "contextual-neighbourhoods"
A semantic index can be defined as a directory of identifiers which are related to each other. A basic version could keep track of identifiers (URIs) and the models (RDF graphs) that are referring to it for building a database of relationships. An advanced version could also include the type of such relationships. Such an index would be used within the concept browser "Conzilla" (see below) for finding out whether and where a specific concept is reused (within other context-maps). However, this project would not be restricted to context-maps only, as it would provide very important information for semantic applications.
Example: A URI is sent to the service and it answers with a list of URIs which identify e.g. context-maps where the URI in question is used. This should happen asynchronously. E.g. if an RDF document is published and should be indexed, the URL is sent to the service which replies immediately with another URL where the status of the indexing can be requested and followed.
This project would be used as a fundamental part within the next version of the Collaborilla framework (Open Source). The programming should be done in Java and follow the REST design principles for resource access.
Idea 4: Domain-specific conceptual modeling using style sheets
The concept browser Conzilla allows for conceptual modeling using symbols gleaned from popular modeling languages like UML. This works very well for technical topics, but limits the capability of expression within other domains, since UML was designed for the software engineering world. There is a systems modeling language which has recently been standardized: SysML. SysML is based on the UML standard and can be seen as a generalization of it. An integration of SysML into Conzilla would be a great improvement for the whole project.
To make this idea sustainable, it would be a huge improvement to be able to import/use/exchange whole sets of symbols in Conzilla, not just SysML. This would improve the domain-specific conceptual modeling capabilities greatly.
To make this idea work, support for stylesheets has to be introduced in Conzilla. To make it as user-friendly as possible, the Conzilla GUI should have support for creating and importing such style sheet packages. It would contain a set of symbols (preferably in some vector format, e.g. SVG) and a description of how to use it.
Do you have your own ideas? Or do you want to discuss the ideas from above?
Just talk to us, we are open for all kinds of ideas. You can contact the Conzilla team either via the mailing lists or simply drop by in the research group's chat: channel #kmr on the server kmr.nada.kth.se, port 6667 (irc://kmr.nada.kth.se:6667/#kmr).