Back ... ... ... ... ... ... ... ... ... ... ... Send comments to review committee. ... ... ... ... ... ... ... ... ... ... ... Forward
This is a note from John Scott to BCN Group Founding Committee Nov. 1.
Paul,
A method is proposed to map a bead game into a Bayesian network using category policies.
Starting at an entangled bead (or a free bead) a choice is made by the user to enter another bead. The user has a category policy present before entering the chosen bead and we assume this predisposes the behavior.
The free bead is entered.
The category policy is used as a means for reference twice, one policy creates the new category policy with the new information (a union combining the present info. and existing). The other action is a category policy of dis-union that compares the original category policy with the new information, but leaving out all categories included in the new info (1 minus the combined category policy).
From this a predictor could be generated to anticipate the next state the user would move onto.
This is a work in progress, but the core process would use Bayesian networks and inference to predict the next step.
The problem I am having is do we have to force the category policy into a number system, or can we perform some other function to manipulate category policies.
Questions
2. And how can we allow a user to follow a line of argument discussion on a focused topic, should those beads be laid out in a path or be left the way they are presently set-up?
- Js