Abstract
This paper studies how logic-based reasoning about actions and change (RAC) with its problems of temporal projection and qualification can be formalised in terms of argumentation. In particular, we extend earlier work of translating the language for RAC into a logic-based argumentation framework, by introducing new types of arguments for (i) backward persistence and (ii) persistence from observations. This forms a conservative extension of the language
that gives a semantic meaning to domains that cannot be interpreted under
thus addressing further the frame and (exogenous) qualification problems. As such the paper strengthens the link between argumentation theory and RAC in artificial intelligence.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. PFP resp.
stands for Positive Forward
resp. Backward
Persistence, NFP
resp.
stands for Negative Forward
resp. Backward
Persistence,
resp.
stands for Positive Forward
resp. Backward
Generation,
resp.
stands for Negative Forward
resp. Backward
Generation, PO
resp.
stands for Positive
resp. Negative
Observation and finally PA
resp.
stands for Positive
resp. Negative
Assumption.
2. if and only if
and there does not exist
such that
.
3. An initiation or termination point for a fluent f in an admissible extension E is defined as in Definition 2.2 where now the preconditions C of the c-proposition are satisfied at T in E when the corresponding HoldsAT conclusions are derived by E.
4. For simplicity of presentation we are assuming that there can only be one generating (initiating or terminating) action of a given fluent at any time point.
5. When we have the other case of ‘L holds-at ’ the extended domain will have admissible extensions with an exogenous qualification of persistence of L in
.
6. Note that can be both an initiation and a termination point for f in M but any given model the fluent f will have either the value true or false (see case (iv)).