This is an old revision of the document!
We have dedicated paths to curated "official" knowledge silos in the form of the links to DIG's docs.evergreen-ils.org, the website's various FAQs pages, and links to developer-oriented documentation on Launchpad. All of these are somewhat "structured" in the sense that they are categorized in some way to help users find specific pieces of knowledge quickly.
We have dedicated paths to community-generated "unofficial" knowledge silos in the form of the wiki, the mailing list archives, and the IRC archives. Of these, the wiki has a loosely structured approach, being divided into "sections" including "New features," contributed component docs and resources, committee working documents, advocacy and outreach materials, etc.
Given the technical nature of the material on the site, the fact that most of our silos are unstructured, and the sheer amount of material available, many users rely on the site's search feature.
Our Website Recommendations include the creation of "content structure[s] and curated content" for specific audiences/roles (e.g., potential implementers, system administrators, consultants, etc.) and for specific topics (e.g., circulation, acquisition, translations, etc.). In practical terms, we want to help visitors in a specific audience/role or those searching for a specific topic by directing them to an easy-to-find list of relevant, high-quality information.
One possible implementation of this system is that of a "knowledge base" - a generic term that can mean a lot of different things depending on context. In general, we want a way to create, manage, and retrieve pieces of knowledge, which can be a fix for a specific error message, a HOW-TO guide, a best practices document, or any other info on the site useful to multiple people.
In one sense, the existing website with its ability to search multiple silos of information is already a knowledge base. However, to make it more effective, there are some technical features and policies that would need to be implemented.
Features common to "knowledge base" implementations
Some sample open source applications dedicated to knowledge base management (these would need to be integrated into the existing web ecology OR replicated with existing systems):