Community-based linking of biological...

Community-based linking of biological network resources: Database, Formats and Tools

To understand your data, you do need to view it in context. Use the kind of network that will be relevant to your question.

How to find Networks? Want structured, dynamic, user-driven content. Got monolithic, unstructured, static content.

The solution? A colaborative web site - WiKi. Sounds a lot like you basically punt all  the hard work of dealing with the data by harnessing 'lots of monkeys' in the form of domain experts to deal with it all for you. Use on-line tools to make it as easy as possible for the domain experts.

Their model has Objects and Connections. Objects include databases, files, formats, tools. Connections describe relations between these e.g. Kegg can export in KGML and this can be read by VANTED. seems like lots of stuff on user-management, access control etc. to give a feeling of data quality by restricted/audited authoring.

How does it deal with conflicting information? What support is there for reasoning? Is it all just human-readable? Are the semantics all in our heads, or actually in the types of connections and objects?

Suports 'semantic-drill-down' of data.

It's very small right now. Probably has a worth that scales non-linearly with the size and activity of the user-base. Very human-facing.