  |
The VAN methods offers a number of analytic capabilities inherent in the VAN data organization and, consequently, unavailable in the other techniques. One unique capability is called Landslide Cueing, where the analyst is alerted regarding the forthcoming radical changes in the overall situation based on the preponderance of accumulated evidence composed of apparently disconnected local developments occurring in different locations and within different time periods.
The Landslide Cue (LC) manifests as multiple significant changes throughout the entire structure of interlocked hierarchies resulting from the accumulation of small local weight adjustments which, being introduced separately and individually, did not cause tangible local re-organizations. Since the entire organization is re-computed each time any adjustment is made, a sequence of such adjustments can leave the organization intact, until the last one in the series results in major changes in both the content and boundaries of the entities, and inter-entity relationships. In a sense, LC indicates potential “paradigm shift” in the domain of observation, such as re-constitution and re-alignment of the structural entities, wherein some entities can dissolve and new ones emerge.
There exists a family of other unique capabilities entailed by the dual integration/partitioning process in VAN. Some of these capabilities are referenced below.
Assertion of the underlying connection between the objects of observation, warranted by the entire history of observation. An entity can hold within its bounds a number of objects with no direct links between them (the linking criteria applied in the course of observation has never been fully satisfied). However, the unobserved connections can still be inferred from the entire history of observation. That is, there can be no single fact pointing at the existence of the connection. Still, the inference can be strongly warranted by the preponderance of indirect evidence suggesting that some object is unlikely to be influenced by any other one except the neighboring objects within the boundaries of the entity where the object resides.
Hypothesizing unobserved structural variations consistent with the observed structure and the history of its development. VAN offers two classes of procedures for hypothesis formation: topologic manipulations and landscape (energy-based) manipulations. Topologic manipulations involve concatenation of the subnets comprised by selected entities. Concatenation operations such as union, intersection, ring sum, produce hypothetical constructs combining some of the elements of their constituents while discarding some other elements. For example, ring sum of two subnets preserves their nodes while keeping only those links that appear in one or the other subnet but not in both. In landscape manipulations, the user adjusts the height of energy barriers bounding some selected entities, and observes structural changes such adjustments entail. Barrier adjustment is equivalent to changes in the factors causing co-variation of the observed objects. In response to such selective adjustment, the VAN process returns hypothetical re-organizations consistent with and afforded by the surrounding structure, permitting "what if" analysis with global comprehension.
Self-diagnosis and bias detection. The VAN system keeps track of the requests made by the user (what entities are requested and in what order, including the objects inside the entities). Selective focusing on some entities without attending to others can be detected and reported back to the analyst. More significantly, tracking data can be converted into additional weights assigned (temporarily) to the co-variation links (e.g., requesting object B after object A causes weight increment on the AB link). Accumulation of “attentional weights” can cause re-organization demonstrating to the analyst the impact of excessive fixation (if the original and “attended” organization turn out to be different).
|
|
