========================================================================== MODELS -------------------------------------------------------------------------- Ontologies Models (in general) are representations of objects, relations, actions, or processes. Examples include a photograph of you, an architect's scale model of a building, an engineer's drawing of an electronic circuit. Models in AI too are representations of an agent's world. Representing knowledge of something requires a language. A knowledge representation language consists of an ontology, i.e., a set of primitives, that form the vocabulary of the language. It also consists of a grammar that defines how primitives may be combined into larger structures. -------------------------------------------------------------------------- Representing Time The simplest representation of time is called a "time-line." Different timelines may represent "events" at different temporal granularity. Sept Sept Sept Sept 13 14 15 16 |--------------------|--------------------|--------------------|--- Patient Patient Patient enters undergoes is hospital surgery discharged | | | V ------------------------------------------ | | | | | | V V Sept 14 7:00 8:00 9:00 am am am |--------------------|--------------------|--- Patient Patient Patient undergoes is given undergoes medical anesthesia surgery examination An event refers to something that happens in time. Multiple events in a given time duration form an "event cluster." Different timeslines may be organized in an abstraction/aggregation hierarchy as indicated above. -------------------------------------------------------------------------- Representing Space A "point" is a location in space; a "region" is a set of points. A "direction" (in 3D) is a point on a unit sphere; (in 2D, direction is a point on a unit circle). In an "adjacency matrix" representation, adjacent cells represent objects (or parts of objects) and adjacency represents spatial relations among objects: ---------------- | X | | X | ---------------- | X | | X | ---------------- | X | X | X | ---------------- In a standard quadtree representation, space is partitioned into quadrents around X (or West-East) and Y (or North-South) axes: NW | NE | | ------------ | | SW | SE As the name suggests, a quadtree organizes spatial representations in a tree. Your projects 3 and 4 use a quadtree spatial representation. -------------------------------------------------------------------------- Representing Causality Causal processes are represented as directed acyclic graphs (called DAGs). state1 | | transition1-2 V state2 | | transition2-3 V state3 /\ tran3-4 / \ tran3-5 / \ state4 state5 Consider the following electrical circuit: Battery Switch | / ----||------------/ ----- | | | | | | ^^ | ----------||-------------- Bulb The two causal behaviors of this circuit may be represented as follows: Behavior 1: State 1 subs loc prop val electricity battery voltage 1.5 volts | | Under-Condition State 2 of Behvaior 2 | Serially Connected Battery, Switch, Bulb V State 2 subs loc prop val electricity bulb voltage 1.5 volts | | Using Function CreateLight of Bulb | Relation Intensity == Voltage**2 (== means proportional to) V State 3 subs loc prop val light bulb intensity 6 lumens Behavior B2 State1 comp(onent) mod(ality) switch off | | Due-to-ExternalStimulus | V comp mod switch on --------------------------------------------------------------------------