boundary, diagram, expression, form, knowledge, language, machine, network, organization, parasite, relationship, representation, topology, wikipedia

Expressive Networks

expressive networks

towards a new diagrammatic model for the abstraction and representation of relational knowledge

How can we apply distributed network theory to knowledge representation? In this paper, we advance a new hypothesis regarding the role of the network topology in information science. In particular, we argue for the need (and significant advantage) of thinking in terms of a parasitic or “counter-network” topology.

While networks are certainly good at representing many things, we need to recognize the significant limitations of this image of knowledge. What does this mean? That the network structure itself must be deformalized, made “molecular” and placed in constant pragmatic variation. The network topology is the most questionable “paradigm” today — despite, or in a sense, because — it has rendered the old hierarchical models obsolete. We find evidence of an uncannily deterministic (and even political) character of the network topology in terms of the protocol or prescriptive communicative rules ‘in force’ throughout the network space. But what if we were to consider a system where all the rules are optional?

One of our most important hypotheses here is that structure itself is a metaphor, something which ‘stands in’ for a formal gap in our understanding. This idea can indeed be made rigorous and even schematic. Information science stands in need of a proper diagram of the parasite. We present here a kind of counter-structural “structure” for knowledge representation, on the hypothesis that a “parasitic” conductor can operate as an atomic unit of relationality. Accordingly we shall also provide an illustrative application (eNAML,) as well as notes towards a new kind of programming language.

Let’s briefly consider the numerous protocol-oriented networked systems in existence today. Such systems are characterized by (a) an open and networked database of information (which is nevertheless essentially “flat” and even planar, though cross-linked,) (b) many distributed agents operating over a continuous flow of information, and (c) organized but non-linear and multi-dimensional growth. Wikipedia and Wordnet are prominent examples of such systems.

In consideration of the wide array of applications already making excellent use of these databases, our strategy here is to rather take a step back in order to ask the formal meta-question: where is the organization topology which best supplements (or supplants!) this networked form of information? Let us define a “network” for our purposes here as a series of singular elements combined a function which encodes certain kinds of differential relationships between elements. In this context a “parasitic” organization could be represented by a tertiary network which provides a transversal mapping between two disparate networks.

The most critical problem with the network model is that it oversimplifies the “edge” or boundary of the system. But what would it take to place these boundaries themselves in constant variation? A parasitic topology “breaks apart” the simplicity of network elements and relationships in order to reorganize the structure itself along entirely new lines. It is in this sense that a parasite is really just a collection of lines along which a network can be mixed with other kinds of networks. (Knowledge itself is a parasite — it infects a system from within and without, bursting this very duality asunder in its rampant infestation.)

The parasitic structure or “threat” to structure comes not only from outside the network, but the formation of counter-networks within the dominant structure. Thus our question here will be: how can we take advantage of minor formations? We have built an abstract model of a networked system whose content will be drawn from the network of relationships between pages in Wikipedia. The goal of these networks are to produce abstract, diagrammatic mappings between concepts. Particular sub-networks encodes relationships according to a relational model or topology defined in turn by another network or diagram. Now it would seem this recursion has to ‘bottom out’ somewhere; where is the basic formal model for our diagrammatic system? It is precisely here where we discover the distinct theoretical advantages of the counter-network topology…

(more soon–)

Standard

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s