[PlanetKR] Advice on Knowledge Representation approach
Edmon Begoli
ebegoli at gmail.com
Sat Dec 8 01:29:35 EST 2012
Dear Colleagues,
I need your advice on Knowledge Representation (KR) approach.
Let me first offer background, then ask for advice:
Background
-----------------
I am attempting to model a domain consisting of a human and intelligent agent.
Intelligent agent is initiating behavior of a human participant, unsolicitedly.
Upon the response from a human, agent might provide a positive or
negative feedback
based on the outcome.
Intelligent agent might provide cues to suggest a particular outcome
while expecting it.
Environment is highly restricted with ideally no external distractions
or content.
These sessions are time restricted. If outcome does not occur within a
particular amount of time,
agent should consider this to be a negative outcome.
Initial KR Thoughts
----------------------------
I see this problem space as connection situation and event calculus. I
also observe that there
is a need to formalize this relationship of acting on observed outcome
and to formalize how is the outcome observed.
I am considering to model hypothesizing observed outcome as an abduction.
Would you please, based on what I described, offer any advices in
regards to KR formalization of this problem.
Thank you in advance,
Edmon
Sessions have to be repeated until the desired outcomes happen
regularly (e.g. n positive outcomes of last m trials).
More information about the PlanetKR
mailing list