Rule3: After each time a belief is used in inference, the term-link accessing it will be rewarded by a value r that is the maximum of the priority of the derived task (or the achieving level of the processed task). This reward can be considered as indicating the usefulness of the belief, and is used to increase the quality of the link from q to or(q, r), so the link will remain active longer. Another possibility is to relate q to the weighed average of the achieving levels of the tasks processed by the belief.
More Info
The Budgets of Beliefs are rewarded. I don't think TermLinks are rewarded. TermLink budgets are updated on line 395 of GeneralEngine.py, though it seems to follow a different procedure than described here.
Technical Report
More Info
The Budgets of Beliefs are rewarded. I don't think TermLinks are rewarded. TermLink budgets are updated on line 395 of GeneralEngine.py, though it seems to follow a different procedure than described here.