Visual Pattern Mapping

So this prototype project is turning out to be a very fun exercise. The basic structure is to use a relation db such as MySQL to store the coordinates of the SVG based visual map. The actual patterns that these images represent are to be stored as XML files following the PLML DTD (a suggestion I got from Yishay).

I haven't decided if I'm going to use a native XML database such as eXist, but for the time being I think I'm going to stick with a basic file-based storage system and use MySQL to deal with the visual mapping application. Note: I hear PostgreSQL would be better for this type of application.

Each primary component of a pattern and the language itself will comprise both a basic image associated with the pattern, forces and problem statement. Following Joanne Tippet's work with DesignWays, I think starting off with either a design goal or collaboratively defined problem statement that allows users to construct a pattern language will be an important piece of software. Hopefully, her work as well as the work taking place with the Public Sphere Project can benefit.

Now, this is not new idea per se, though I have been interested by the implications of such a project, but a similar project promoted and tested at the London Knowledge Lab has sought to promote a similar paradigm. However, I want this project to be the first step that will eventually support the use of past patterns. The system will query a pattern database based upon a search criteria. By using PLML, patterns will link each other together and in a sense help auto populate a basic starting pattern language to address a particular community problem.

Now, with the basic pattern mapping application already complete users could then amend and extend this generated pattern language and replace it with a more community generated one.

This is important I think, as the shear number of patterns that have emerged over the past few years has severely hampered the ability of pattern users to reuse components of knowledge. If we could search and provide semantic relationships between community developed knowledge it is hoped that we could be more able to use that knowledge in future projects.

Question: Why not define and embed an Ontology within your Content?

As some of you might have noticed, I haven't made a post in a few days. This has been for a couple of reasons: 1.) the holidays, 2.) the end of the semester, and 3.) I have been focusing on research questions.

On the topic of research I have migrated back to the idea of creating NRM centered patterns and pattern languages versus creating a pattern language for NRM based systems design. The reason for this move back is due to the inescapable perception that patterns provide a fundamental model for structuring knowledge. I have thought this to be so for several years now, and as I have delved deeper into the literature of Knowledge Management Systems in the Natural Resource Sciences, I have found myself thinking of patterns as increasingly relevant components for developing community based KM systems.

However, the issues identified in my master's thesis still exist. This issues are: 1.) how do we visually express not only an individual pattern, but more specifically, how do we express a network of patterns?, and 2.) how do groups search and access interrelated knowledge within a system of patterns?

While the first question seems to interest my colleagues the most, it seems to me that the issue of traversing a pattern knowledge-base should be the first issue to solve, with the visual display coming in after the underlying mechanisms for calling relevant patterns have been addressed. This I think is especially relevant when considering the growing scope of pattern languages and the vast number of patterns that have been authored over the years.

This assumption and observation has led me to consider approaches to solving this issue. As a result, I have been doing an extensive investigation into the role of ontologies and taxonomies for describing relationships between patterns. This brought me to a paper written by Scott Henninger from the University of Nebraska who has describe similar thinking. The paper entitled, "Disseminating Usability Design Knowledge through Ontology-Based Pattern Languages" argues for need to develop ontologies for patterns so that groups might be better able to search and draw upon previously codified knowledge. His approach was to use OWL (Web Ontology Language) as specified by W3C.

This discussion on OWL sparked an intensive search into OWL and related technologies. However, the complexity of the technology and how it is applied has made it very difficult for me to get my head around. I understand the notion of a Semantic Web and I think the vision of such a web is extremely interesting.

However, I have been looking at the Pattern Language Markup Language (PLML) as suggested by Yishay and I just started to wonder, "Why not embed these ontological rules between patterns/problem/solution/forces directly within the PLML marked-up patterns. Meaning, why not just extend the DTD for PLML to include ? In this way the PLML document could be processed with any programming language that has support for XML.

Now, I understand that this might negate the idea of a global Semantic Web based upon the standards set forth by the W3C. But it just seems to me that this approach is so much simpler (except for the fact that pOWL and Protege-OWL attempt to make ontology creation easier).

I don't know, but RDF and OWL confuse me.

Someone, please explain why I couldn't or wouldn't want to embed simple ontological statements in my XML doc? pOWL seems interesting, but why?