Ontiki: an ontology-aware wiki

Ontiki is an experimental effort to create an ontology-aware wiki, being developed by Rich Morin as a tool for Model-based Documentation. Rich floated the idea, initially, in a weblog entry. This page continues the topics introduced there, in the same (highly speculative) style.

Motivation

Ontiki is expected to have many characteristics in common with semantic wikis, but emphasize aspects of ontology creation, editing, and use. In particular, it is intended to be populated mechanically, based on harvested information, while allowing users to add their own comments, questions, etc.

Most semantic wikis are based on Resource Description Framework (RDF), which use binary relationships. This is both buzzword-compliant and convenient. If the current and target pages are interpreted as the subject and object of an RDF triple, all that needs to be added is the predicate. Unfortunately, this approach has some drawbacks:

So, Ontiki will allow N-ary relationships (ala Conceptual Graphs and Object-Role Modeling) in which each entity plays a specified role. As "scaffolding", each class or instance of an entity or relationship will have its own page.

For convenience, certain common idioms (e.g., attributes, typed links) will also be supported. This combination should allow both simple complex concepts to be expressed in an efficient and natural manner.

The initial test case for Ontiki will be one or more Unix-based operating systems, such as FreeBSD, Linux, Mac OS X, or Solaris. Thus, it will be a continuation of work that Rich has been doing on Morinfo and the Meta Project.

Overview

Ontiki will have all of the capabilities of conventional wikis: convenient editing, simplified markup, support for collaboration, rollbacks, etc. It will also borrow ideas from existing semantic wikis: backlink tracking, ontology support, typed links, etc.

It will differ from previous efforts, however, in two ways. First, it will try to strike a balance between the freedom of wikis and the complexity and rigidity of (most) ontology editors. Second, it will have strong support for simplified mechanical generation of pages.

Because Ontiki's definitions are stored in a knowledge base, derived information can be used to increase its capabilities:

Ideally, Ontiki would excel at providing:

  • Content

    Information could be entered by users, derived from the knowledge base, and/or harvested from an external source.

  • Context

    Relationships could be exploited to provide the user with contextual information and navigation aids (e.g., links and clickable diagrams with explanatory tooltips).

  • Contact

    Comments and questions could serve as "messages" in a "structured forum", allowing users (and programs) to monitor areas of interest.

Infrastructure

Currently, Ontiki is using only a few technologies:

The next technologies to be integrated will (probably) include:

In addition, the following components look very promising: