ORB Visualization
(soon)
Procedure for removing the central linkage from an ORB
In this research note, we discuss a procedure for removing the central linkage from an ORB, thus allowing the categoricalAbstraction prime structure to be more easily seen.
It is first important to see why it is necessary to our purposes.
Figure 1: The cA compound for the word occurrence “then”
In Figure 1 we see the local region (Subject Matter Indicator neighborhood) for the word “then” (this is in the FCC 1998 ORB). We also see the linkage from this neighborhood to other neighborhoods, including one word linking “then” to a large number of other neighborhoods. (The labeling does not allow us to see the name of the word, but one can see the fan of connectivity from the upper left of the neighborhood graph.)
We want to remove neighborhoods that act to provide this high linkage, because in most cases these neighborhoods are formed around subjects that express quite differently in each of the occurrences. So these neighborhoods require a high degree of disambiguation. Perhaps this disambiguation could be developed using the Method for disambiguation, but our purpose is to create a 200-word subject list. The subjects that we want to find are more like Figure 5 – 8 in the previous research note.
Fortunately, the way in which the ORB is encoded allows us to immediately remove all neighborhoods having high connectivity. The ORB internal architecture is made public in the Notational Paper, and using the CCM patent. We are thus able to talk directly about the internal architecture while retaining ownership over the internal process.