1 Introduction

  • situation calculus, global
  • event calculus, local events and time periods

2 A Simplified Example

  • hire, left, update, knowledge
  • figure
    image
  • sometimes, the end of a period after(e) (e is a certain time instant) is undifined but might be determined by means of additional information later
  • end of lectureship, begin of professorship
  • proof after(\(E_1\)) = before(\(E_4\)), i.e. combine "*-- and --*" to "*--*"?
    • *--*--* structure, "complete"
    • derived by a rule expresses ... 'cannot': negation, non-monotonic
      image
  • another rule
    image
    • aim to find Mary's rank at a time instant using the mentioned rules.
  • general characteristics of the EC approach
    • add, not delete ("no longer"? we explicitly show the end)
      • in contrast, conventional database systems, arbitrary.
      • EC: an extra level of semantic structure (inductive bias)
    • Past and Future are treated symmetrically. independent of the order. incomplete, with new knowledge, with hypothetical possibilities.
      • conventional: integrity constraint, statement
    • relative ordering, numerical? not essential.
    • concurrent
    • duration, now pictured
      image
      • not wholly contained
    • equivalence preserving, PROLOG
    • situation calculus, global, frame problem, computationally inefficient. EC: local, different names even if they have the same duration
    • periods rather than on time instants
    • sacrifice conciseness for expressiveness, modal logic

3 The Promotion Example in Detail

  • an exemplar rule: Rank(person, position, after(instant)) if Hire(person, position, instant)
  • assume complete now (even though an incomplete description would be more natural)
  • general rule, same rank, <, not disjoint, "Temp" means domain specific.
    • axiom: same relationship: identical or disjoint
    • disjoint: < or =!
  • rule for the notion of disjoint
  • end, start, same, sabattical (so the notion of disjoint contains '=')
  • interpret the negative condition?

4 Execution of the Promotion Example

  • PROLOG, infinite
  • try to solve..., so try to solve..., loop
  • intelligent solver or transformation techniques

5 Incompleteness and Incorrectness of Start and End

  • Negation by failure is a form of the closed world assumption, that the "knowledge base" is complete:

not p is judged to hold if all ways of showing p fail

  • start, end, incomplete, incorrect
  • some event, not reported, occured
  • rely on the completeness of 'start' and 'end', incorrect
  • additional rule, solve the incorrectness before incompleteness

12 Conclusion

  • events, primitive
  • conflict, non-monotonically withdrawn
    • conventional: reject. only appropriate when ordered
    • extension, "Mary was a professor when Jim was promoted"