The project involved:
Searchable Relational Database: more specific searches are possible, compared to web or wiki tools.
Browsable: Search by hierarchical category scheme, as well as by keyword. The database does not attempt to decree one universal ontology, like most systems. It also does avoids the pitfalls of many "folksonomies" or tag systems, which average away individual insights. It recognizes that different users will understand different schemes, and in fact users can create their own category schemes. This will allow users to not just find information, but organize it, and even help researchers expose gaps in existing knowledge.
Linked Search & Browse: When a record is selected it automatically pops up in the category schemes--thus, finding one record by keyword can lead to finding a whole category of records which are related but do not share keywords. This helps users find things they do not know the words for, or search across discipline boundaries.
Modularity and Interrelation: Records are split into challenges, strategies, people, organisms, products, and citations. Records are connected to each other through category schemes and permanent links.
Interactivity / Collaboration: Records include space for comments, which can become bulletin-boards for discussion; qualified users can also add, edit, and link records like a wiki. Each user has their own page, which they can use to blog research or ideas.
I belive the novel structure of this system would make it a useful tool for any cross-discipline knowledge portal.
The proof-of-concept prototype of the database is online at http://database.biomimicry.org. A full-fledged version awaits further planning and funding.