Navigating relational information spaces
From Knowledge Federation
Contents |
Relational Information Spaces
Let us begin the process of defining relational information spaces. There is lots of room to expand on such definitions.
From Wikipedia, a relational space is composed of relations between objects. Nicholas Rashevsky created a lens through which to view notions of relational spaces in his paper Topology and life [1] in which he investigated the complexities of a single-celled creature trying to understand the principles of life. His observation was a simple one: we can open up a living creature and count all the components found inside (to the limits of our instruments), but we could not put it back together; something is missing from our understanding. The outcome of that work became known as Relational Biology, the elevator pitch for which is that the action resides in relations between components and between components and their environment. Living creatures, social systems, the ecosystem in which we exist, are all instances of relational spaces.
Returning to the root concept that a relational space is composed of relations between objects, and using Rashevsky's observations and conclusions as a hint, it follows that a relational information space should be one that grants priority (some might suggest citizenship) to the relations in that space at least equal to that metric as granted to the components (actors) in that space. Nouns and verbs, in other words, are equals.
Organizing Relational Information Spaces
I suggested that Nouns and Verbs are equal. To organize them, we must represent them. A familiar approach is that of nodes and arcs, where nodes are containers of information and arcs (usually labeled) represent the relations between nodes. Collections of nodes and arcs form a graph structure. Let us examine the level of citizenship granted to nodes and to arcs.
Nodes are containers. They have identity. This means they contain information about the entity (actor) they represent. We will soon see that they stand as a proxy for a topic.
Arcs are typically just labeled links between two containers (nodes). They typically do not have identity; they typically do not serve as containers for other information. It's reasonable to say that arcs do not share the same level of citizenship as nodes. We seek another approach. The nodes + labeled arcs graph is known as a concept map.
The next level of representation is known as a topic map. In some topic maps, relations are represented as nodes just as are the nodes which are actors in a relationship. They are containers with identity, which means they can serve as actors in other relationships. Consider a trivial example:
Actor A: Id="123", Name="Carbon Dioxide"
Actor B: Id="945", Name="Climate Change"
Relation 1: Id="946", Name="Causes"
With those three nodes, one serving as a relation, we now have a graph that can represent this statement: Carbon Dioxide causes Climate Change
Since people may disagree with that statement, they have a convenient node in the graph to anchor an argument. It is the relation node, the explicit meaning of which is precisely that statement.
That's the representation component. For the organization component, consider this: a topic map is a type of map. This is knowledge cartography [2] at work. A map is an organizing tool that facilitates this organizational heuristic: whatever is knowable to the map about a particular topic is to be found at one and only one place in the map. In terms of software implementations of topic maps, there is no rule that says that all that is known about a topic is physically co-located; the heuristic is about views provided by the map. The terminology applied to topic mapping refers to a node (from the graph model) as a proxy. A proxy represents a topic, any topic. Since we can represent relations, those can include taxonomic relations.
Topic maps are not suggested to be the only approach to the represenation of relational information spaces; they are listed here to suggest a rich starting point for a larger conversation on the nature and fabric of relational information spaces.
A rationale for crafting relational information spaces with topic maps is that the knowledge federation component is built in. The topic map serves as a federation engine; it continuously maintains the subject-colocation aspect of a well-organized information space.
Navigating Relational Information Spaces
Navigating relational information spaces is a rich space in which the concept of US - user experience design is a topic of growing importance. We will only touch on the topic very lightly here, leaving lots of room for expansion. Let us consider just one particular relational information space that is appropriate to journalism: conversation spaces. The blogosphere is one such space. An emerging but similar space is that of structured conversations. Issue-based Information Systems (IBIS) are one of several approaches to structuring conversations using graph structures. Examples are the well-known Debategraph, Cohere, Climate Collab, Rationale, and some experimental platforms, for instance an IBIS extension to MediaWiki, an experimental IBIS server, and many others. Those are cited as indications of approaches to user experience design for structured conversations.
References
[1] Topology and life: In search of general mathematical principles in biology and sociology. Bulletin of Mathematical Biophysics 16 (1954): 317–348
[2] Knowledge cartography: http://www.knowledgecartography.org/ , http://books.kmi.open.ac.uk/knowledge-cartography/
