Rule Based Systems

1. Rule Based Systems

2. Rule Based Systems - Introduction • Well formed formulae (wff) in propositional or first order logic represent assertional knowledge. • Such wffs are divided into two categories. - Rules : assertions given in implicational form. - Facts : assertions that represent domain specific knowledge.

3. Rule-Based Systems • In Logic we represent knowledge in a declarative, static way - as some facts and rules that are true. • Rules in logic say what is TRUE given some conditions. • Rule-based systems are based on rules that say what to DO, given various conditions. • IF <this is the case> THEN <do this> • A special interpreter controls when rules are invoked.

4. Rules and Logic • Simple examples are very similar to rules in logic. • However, in rule based systems we consider: • Other kinds of actions (apart from adding facts). • Degrees of certainty associated with facts. • Various different control schemes (not necessarily related to idea of logical proof).

5. Example Rules • If it rains today the roads will be wet today Rains(today)  Wet-road (today) If Rains(today) Then Wet-road(today) If Rains (today) and Not-covered (roads) Then Wet-road (today) Which are facts? Facts are also generated

6. Recall Modus Ponens P  Q P --------------- Q Q  R --------------- R

7. Rule Based (Production) Systems • A system whose knowledge base is represented as a set of rules and facts is called a Rule Based System. • A rule-based system consists of a collection of IF-THEN rules, a collection of facts, and some interpreter controlling the application of the rules, given the facts.

8. Rule Based System - Rules • Rules are represented in the following form : IF <antecedent> THEN <consequent> • When the antecedent part is NULL the Rule becomes a fact. • Rules are normally represented as Horn Clause.

9. Rules are normally represented as Horn Clauses P  Q ¬ P V Q P  Q  R ¬ P V ¬Q V R P  Q  R V S ¬ P V ¬ Q V R V S Horn clauses can have at most one non-negative literal

10. Rule Based Systems – Triggered and fired rules • A rule is triggered when all the antecedents evaluate to true. • A rule is fired when the action stated in the consequent part or the inference related to the consequent part is inferred / taken.

11. Rule-based system architecture Control Scheme (Interpreter) Condition-Action Rules R1: IF hot AND smoky THEN ADD fire R2: IF alarm_beeps THEN ADD smoky R3 IF fire THEN ADD switch_on_sprinklers Database of Facts alarm_beeps hot

12. Conflict Resolution Match Execute Goal Check Inference Machine • A machine that implements strategies to utilize the knowledge base and derive new conclusions from it. Automata of Inference machine Fact and Rule Base R’

13. Inference Machine • The execute state fires the rules once all its antecedents match. • Essentially, the function of the execute state can be thought of as searching a path to the goal in a search space.

14. R5 R1 Goal Node

15. Backtracking • If the inference machine reaches a dead end ( no new rule is enabled and the goal is not met) then backtrack to the earlier node • Good conflict resolution strategy reduces backtracking.

16. Conflict Resolution • The objective is to decide which of the triggered rules in a particular stage should be fired. • Can use various strategies :-- • FCFS (Rule Ordering) • Specificity Ordering • Fire all • Heuristic measures (distance from the goal)

17. Conflict Resolution : Contd.. • Other strategies used for the purpose are:-- • Refractoriness: Rules once fired will not be fired later. • Meta-rules : Rules about rules embedded within the inference machine that provide the information about which of the rules apply under what conditions. • 90% of computing time is spent in the match phase, hence minimizing the same is necessary.

18. Searching the space • Start from the given facts try to arrive at the goal • Start from the goal and try to prove the goal using the given facts

19. Reasoning Mechanism in Rule Based Systems

20. Forward Chaining Mechanism • Starting from the start state applying rules one by one to arrive at the goal state. • Also known as Data Driven Search. • Forward chaining may lead search to a dead-end. In such cases backtracking is necessary. • Backtracking strategies can be : • Chronological • Intelligent

21. Forward chaining • In a forward chaining system: • Facts are held in a working memory • Condition-action rules represent actions to take when specified facts occur in working memory. IF condition THEN action. • Typically the actions involve adding or deleting facts from working memory.

22. Forward chaining • Control cycle called recognise-act cycle. • Repeat: • Find all rules which have satisfied conditions given facts in working memory. • Choose one, using conflict resolution strategies. • Perform actions in conclusion, probably modifying working memory. • Until no rules can fire, or “halt” symbol added to working memory.

23. Example Control Scheme (Interpreter) Condition-Action Rules R1: IF hot AND smoky THEN ADD fire R2: IF alarm_beeps THEN ADD smoky R3 IF fire THEN ADD switch_on_sprinklers Database of Facts alarm_beeps hot

24. Example • Simple “fire” example from earlier: • Working memory initially contains: • alarm_beeps • hot • Following the algorithm: First cycle.. • Find all rules with satsified conditions : R2 • Choose one: R2 • Perform actions: ADD smoky. • Working memory now contains • alarm_beeps, hot, smoky

25. Example continued • Next cycle: • Find all rules with conditions satisfied : R1 • Choose one and apply action: ADD fire • Working memory now contains alarm_beeps, hot, smoky, fire. • Then • Rules with conditions satisfied: R3 • apply action: ADD switch_on_sprinklers.

26. Forward chaining applications • Forward chaining systems have been used as: • a model of human reasoning • basis for expert systems - various expert system shells based on this model, such as CLIPS. • Practical forward chaining systems support pattern matching • Example CLIPS rule:(defrule fire-alarm (temperature ?r1 hot) (environment ?r1 smoky) => (assert (fire-in ?r1)))

27. Backward Chaining • Same rules/facts may be processed differently, using backward chaining interpreter. • This allows rather more focused style of reasoning. (Forward chaining may result in a lot of irrelevant conclusions added to working memory.) • Start with possible hypothesis. Should I switch the sprinklers on? • Set this as a goal to prove • Similar to Prolog which uses a backward chaining style of reasoning.

28. Backward Chaining • Basic algorithm: • To prove goal G: • If G is in the initial facts, it is proven. • Otherwise, find a rule which can be used to conclude G, and try to prove each of that rule’s conditions.

29. Control Scheme (Interpreter) Condition-Action Rules R1: IF hot AND smoky THEN ADD fire R2: IF alarm_beeps THEN ADD smoky R3 IF fire THEN ADD switch_on_sprinklers Database of Facts alarm_beeps hot

30. Backward Chaining Example Should we switch on the sprinklers? Set as a goal. • G1: switch_on_sprinklers • Is it in initial facts? No. Is there a rule which adds this as a conclusion? Yes, R3 • Set condition of R3 as new goal to prove: • G2: fire. • Is it in initial facts? No. Rule? Yes, R1 • Set conditions as new goals: G3: hot, G4: smoky.

31. Example continued • Try to prove G3: hot. In initial facts. • Try to prove G4: smoky. Conclusion of rule so.. • G5: alarm_beeps. • In initial facts, so all done… • Proved hypothesis switch_on_sprinklers.

32. Examples of RBS • Available expert systems:-- • CLIPS • NEXPERT • Languages facilitating development of Rule Based Systems by incorporating within its structure certain advanced storage techniques. • PROLOG

33. Expert Systems • Systems acting in a particular domain and behaving like human experts therein. • Normally expert systems are applied to problems for which no algorithmic solution exist. • Guided by domain specific rules and should be able to explain its actions.

34. Expert Systems Applications • Rule Based systems have been widely used in expert systems. • E.g., medical systems, where start with set of hypotheses on possible diseases - try to prove each one, asking additional questions of user when fact is unknown.