How does the Knowledge Base work?

The Indiana University Knowledge Base (KB) is a custom-built content management and workflow system managed by the IU Knowledge Management (KM) team to collect, maintain, and publish online answers to information technology questions at IU.

The heart of the KB is its staff, including both the KM team and the KB Development team.

The KM team writes the documents that make up the KB and publishes them to the web. To support the IU community, team members work with information technology providers throughout the IU enterprise to publish both internal and external information in the KB.

The KB Development team maintains the infrastructure of the KB and ensures its availability as a mission-critical system. Team members also design and implement improvements to the KB to take advantage of new technology.

In terms of resources and infrastructure, the KB is made up of the following elements:

• More than 14,000 files (each one a KB document), most consisting of a question and an answer, formatted in Knowledge Base Markup Language (KBML), a proprietary text markup language described using SGML. KBML is similar to HTML, but with some customizations and special features for the needs of the KB.
  
  
• A proprietary full-text search engine written in Perl 5
  
  
• Debian Linux
  
  
• A cluster of load-balanced HP rack servers broken into a backend tier handling the data and logic, and a presentation tier handling the web interface
  
  
• An Apache web server
  
  
• Several MySQL databases containing document metadata
  
  
• Unix Revision Control System maintaining the revision and editing history on all documents
  
  
• GNU Concurrent Versions System (CVS) for version control of all source code
  
  
• Tools and utilities for KB usage reports and text maintenance. The KM team uses Emacs with proprietary Lisp extensions to edit documents. The KB also has web and command-line tools for intensive document searches, and a web-based workflow system called the Workshop for collaborative document editing. Additional utilities include a tool to study the search strings actually submitted by users, and tools for adding and removing links to other KB documents and for changing metadata. KB documents are written in KBML but translated into XML before being stored in the database and delivered to the presentation system.

KB documents

Thousands of KB documents have been written by the telephone, email, and walk-in consultants who work for the University Information Technology Services (UITS) Support Centers in Bloomington and Indianapolis. Many documents originate from questions asked of consultants by the computing community. Also, as technologies change and new computing tools are introduced at IU, UITS system and service administrators and others from departments across the entire IU system revise and contribute information to help prepare for changing support requirements.

As KM staff add documents, they supply metadata to help with document maintenance. The text of a document can also contain hidden information. Some of this information, such as background or maintenance information for staff, does not affect the way the document is used. Other hidden words, however, can be embedded in the document to affect searches. For example, in each document that contains the text "email", KM staff create a hidden area of the document in which they also include the term "mail". Including common alternate spellings of such terms increases the chance of a search returning all relevant documents.

KB documents regarding technologies no longer used at IU (such as Pegasus Mail) or services no longer provided at IU can be archived. By default, archived texts are not returned in searches, but you can access them by checking Include archived documents when you submit your search.

When staff make changes to documents, words that were once contained in documents may be removed, and new words added. These changes may not be reflected when searching the KB until the index is rebuilt each night. Likewise, new documents are not searchable until the day after they are added to the KB.

Also see:

• What is the history of the Knowledge Base?
• May I reuse Knowledge Base content, or have the Knowledge Base code?
• How do I use the Knowledge Base?
• What's new with the Knowledge Base?