Comp 307 Lecture 19 Week 3 Help

Comp 307 Lecture 19 Week 3 • Help desks – – – Mon 4 -5 CO 238 Tutor: Su Wed 1 -2 CO 238 Tutor: Urvesh Fri 11 -12 CO 238 Tutor: Bing • Lectures: Mon, Tue, 3 -4, HU 119 • Tutorial: Fri, 3 -4, HU 119

Comp 307 Knowledge based systems Reasoning Backward chaining Forward chaining Reasoning implemented in Prolog More on Knowledge representation More on reasoning Challenges

Comp 307 Simple Rule-based System in Prolog mammal: -f(hair). mammal: -f(milk). bird: -f(feathers). bird: -f(eggs), f(flies). carnivore: -mammal, f(meat). classify(tiger): -carnivore, f(tawny), f(striped). classify(puma): -carnivore, f(black). f(hair). f(black). f(meat). |? -classify(X). X=puma.

Comp 307 Lecture 19 An example • classify. pl – – Prolog Knowledge in a special rule language Expert system shell/tool Backward chaining • Available on lecture page • It will ask questions, just answer “yes” or “no”, can ask “why”.

Comp 307 Rule Language for the Knowledge. Base Defining the operators : : - op(511, op(505, op(503, op(501, xfx, : : ). fx, if). xfx, then). xfy, and). rule 4: : if eggs and flies then bird. : : (rule 4, if(then(and(eggs, flies), bird))).

Comp 307 Operators in Prolog • Examples of operators: : - ? - , ; + - * / is • Operator declarations e. g. : - op(1200, fx, ? -). : - op(1200, xfx, : -). : - op(1000, xfy, , ). : - op(700, xfx, is). : - op(500, yfx, [+, -]). : - op(400, yfx, [*, /]).

Comp 307 Operators in Prolog • Operator precedence is represented as a number 0 -1200 (A higher number means a lower precedence. ) Operator types: xfx xfy yfx fx, fy xf, yf binary, infix, not associative binary, infix, right-associative binary, infix, left-associative unary, prefix unary, postfix

Comp 307 Operator Precedence and Parse Trees : - op(1200, xfx, : -). : - op(700, xfx, is). : - op(500, yfx, [+, -]). add(X, Y, Z) : - Z is X + Y Internal (term) representation: : - (add(X, Y, Z), is(Z, +(X, Y))). : - (1200) add(X, Y, Z) is (700) Z + (500) X Y

Comp 307 Design your own rules in prolog : -op(800, fx, if). : -op(700, xfx, then). : -op(300, xfy, or). : -op(200, xfy, and). If hall_wet and kitchen_dry then leak_in_bathroom. if window_closed or no_rain then no_water_from_outside. fact(hall_wet). fact(bathroom_dry). fact(no_rain).

Comp 307 backward-chaining • Pick one hypothesis • Find the rule • Verify each condition in the body of the rule – Each condition as a new goal

Comp 307 Backward chaining in Prolog is_true(p): fact(p). is_true(P): if Condition then P, is_true (Condition). is_true(P 1 and P 2): is_true(P 1), is_true(P 2). is_true(P 1 or P 2): is_true(P 1) ; is_true(P 2).

Comp 307 Forward Chaining • Data driving, Data directed reasoning, bottom up – Search from facts to valid conclusions • Given database of true facts: Apply all rules that match facts in the database – Add conclusions to database – Repeat until a goal is reached OR – Repeat until no new facts added –

Comp 307 Design your own rules in prolog : -op(800, fx, if). : -op(700, xfx, then). : -op(300, xfy, or). : -op(200, xfy, and). If hall_wet and kitchen_dry then leak_in_bathroom. if window_closed or no_rain then no_water_from_outside. fact(hall_wet). fact(bathroom_dry).

Comp 307 Forward Chaining in Prolog : -dynamic fact/1. farward: new_derived_fact(P), !, assert(fact(P)), farward: -write(‘no more facts’). new_derived_fact(Conclusion): if Cond then Conclusion, + fact(Conclusion), composed_fact(Cond): -fact(Cond). composed_fact(Cond 1 and Cond 2): composed_fact(Cond 1), composed_fact(Cond 2). composed_fact(Cond 1 or Cond 2): composed_fact(Cond 1) ; composed_fact(Cond 2).

Comp 307 Knowledge Representation and reasoning • More Knowledge Representation – – – Decision trees Decision tables Belief nets Fuzzy logic Neural networks … • Reasoning – Case based reasoning • Knowledge: Cases and modification rules • Reasoning: find similar case and adapt – …

Comp 307 Decision Tables diet meat grass meat habitat color jungle striped house striped plains striped jungle brown animal tiger tabby zebra weasel Example Rules: tiger meat jungle striped jungle<-- meat / striped / tiger

Comp 307 Decision Tree Diet? grass zebra meat Color? striped brown Habitat? weasel house jungle tabby tiger

Comp 307 Knowledge based systems • Challenges and limitations – – – Knowledge acquisition Conflict resolution Uncertainty Common sense ……