You are on page 1of 15
Logical Agents and Knowledge Representation using Propositional Logic Syllabus Logical Agents, Knowledge-based agents, The Wumpus world, Logic, Propositional logic, Propositional theorem proving, Effective propositional model checking, Agents based on propositional logic. Contents 54 Characteristic of Propositional Logic ........ Summer-20- 5.2 Drawbacks of Propositional Logic 53. Syntax for Propositional Logic 5.4 Semantics for Propositional Logic 55 Reasoning Patterns in Propositional Logic 8.6 Forward and Backward Chaining 5.7 Effective Propositional Inference 58 Local Search Algorithms for Inferencing in Propositional Logic 59 Knowledge based Agents + Mark 1 5.10 University Questions with Answers sl . f2 ops 9 Pe ma Propositional Logic Ce EEXM characteristic of Propositional Logic 1. Propositional logic is declarative (pieces of syntax corresponds to facts) 2. Propositional logic allows partial /disjunctive/negated inference (unlike moy a structures and databases) 3. Propositional logic (Meaning of By, 5 « compositional 4 fs deried from meaning of By, and meaning of py 4. Meaning in propositional logic 1s context independent (unlike matural angus where meaning depends on context). E24 Drawbacks of Propositional Logic 1. Propositional logic has very limited expressive power (untke natural languag) Example : Cannot express “pits causes breezes in adjacent squares. For such statement one need to write, one sentence for each square. 2. Propositional logic represents statements about the world without reflecting be structure and without modeling these entities explicitly. 3. Therefore some knowledge is hard or impossible to encode in the proposition! logic. 4. Two cases that are hard to represent : 1) Statements about similar objects and their relations 4) Statements about similar objects and relations needs to be enumerated b) Because of this knowledge-base grows large. ca We need to represent many rules to allow inferences. The solution fo this problem is introduce variables. Case li) Statements referring to groups of objects 4) Statements referring to groups of objects require exhaustive ‘enumeration of objects. ») Solution to this problem is to allow quantification in statement which is done in first order logic}. TECHOUCAL PUSUCATIONS® «An up & fe nage ose 5.3 Krone e for Propositional Logic syntax redefines allowable sentences In the model 57 onal calculus symbols : poral ca one erry a popeonl clas ak S~ . 1 rade symbol: Tae wl epee a atom For example = if we want to represent 2 sentence lik, "Dipu reads book’ tue wl use the predicate symbol "READS" The simple predicate will be, READS (Dipu, Book). say Constant symbols: A constant symbol used to represen objets or entities in 2 Gomnin. These objects or entities may be physical objects, people, concepts oF spything that i tbe named. for evample: In the above formula Dipu and Book are constant symbols sy Variable symbols: Variables Ike x o y are also symbols, They allow the users to indefinite about which entity s to be reffered to example READ (x,y) 4) Function symbols: They denote functions in the domain of discourse For example : “Ram's mother is marred to Ram's father, the atomic formula would be, MARRIED [MOTHER (Ram), FATHER (Ram)} where MOTHER, FATHER and MARRIED are function symbols. 2 Truth symbols True, False. 3. Connectives: (23 Fropositional symbols denote propositions, or s Tuy be either true or false, such as “the car is ed” or “water is wet’ tatements about the world that EEX Propositional Calculus Sentence |) Every propositional symbol and truth symbol is sentence For example : True, P, Q R are sentences aan Paucar i Pate mee ‘oper ~ = am 2 eo x sete opel igo scr ' . oo 5 aprosontaton wat Propastone! Le 81 Atom sentence: I is inisle (nomcompeste) IEE demon, ee of proposition symbol. Each such symbol stands for 4 Proposition ye IF connec’ ‘dal of Propstn ys «a an pws 9 epeset hen tenet, ee onample! Complex sentence : connectives. ED Logical sentences ave alo called EEE] connectives ‘Atomic formulas can be converted to well forme ae oarned fom spt tints SSR aye tock en cri pice epi suvs (simu, Frock) = COLOUR (rock, Fink) “iepipa is caught in the ran then she uses umbrel. {GHT IN (Dipu, Rain) = USES (Dipu, Umbrella) by connecting twa formulas with‘! ie called as implication. The left lication is called the antecedant and the right hand side is called the 1s well-formed formulas or WF 3)" caut [A formula built and side of an irl formulas By combining thea, consequent sythe BIGONDITIONAL connective : <= 1 (2) Q shows that it is true whenever both P=» Q and Q.=> P are true, In English ten written as "P" if and only if°Q" or “P* iff "Q" {) The AND connective : » The *y” connective is used to represent compound statement Hike “Nima wey dress. WEARS (Nim, Dress) « COLOUR (Dress, Pink) ‘Where the predicate WEARS gives the relation between person and object an prediate COLOUR gives the relation between object and colour Formulae built by connecting other formulae by 1's are called conjunctions and ns of the component of the formula is called a8 conjunct. this i oft For example : -vimu becomes happy iff Nimu eats icecream” This sentence can be represented 25, BECOMES (Nimu, Happy) = EATS (Nim, Ieecream) «Truth table for five logical connective b)The OR connective : v ‘The symbol ‘V" is used to represent inclusive ‘or’. ¢)The NOT connective ; — i. aca . oe . represent the NOT connective. It is used to negate B mula. anges the value of sentence from T to F and vice vets Dipu did not read book, can be represented as : Following are Some of the Legal Sentences (WFF's) in Propositional Logie 1). The negation of a sentence is a sentence. For example : —P and ~ false are sentences. 2), The conjunction, ‘and’, of two sentences is a sentence, For example + Pa 4 Pisa sentences. 3). The disjunction, ‘or’, of two sentences is a sentence, A formula with ain front of i in front of itis called a negation, ©. TECHNCAL PUBLICATIONS? Anup tat for inowtodge Se Logica Roprosentation using seed Art cil Intogence AD For example : Pv Pisa sentence. 4) The implication of one sentence from For example : P+ Qis a sentence. 5) The equivalence of two sent For example + Pv Q= Risa sentence. another is a sentence. tences is a sentene- [EEE] crouping of Symbols in Propositional Logic sional calculus, the symbols () and [ in propos pre .ir order of evaluation and me subexpressions by which the For example = (Pv Q) = Ris different from Pv (Q= RD [EEE] Forming a Legal Sentence in Propositional Logic ‘An expression of legal symbols through some seq (PaQ>R=aPvaQvR js a well fomed sentence in the propositional calculus because 4). P, Qand R propositions and thus sentences i) PQ the conjunction of two sentences. yuence of these rules. It is a sentence. ii) (P 4 Q) > Ris the implication of a sentence for another. Tt is a sentence. iv) +P and Q the negations of sentences are sentences. v) +Pv~Q the disunction of two sentences, is a sentence. vi) Pv +Qw R the disjunction of two sentences, is a sentence. vii) (P 4 Q)>R)#4PvQVR The equivalence of two sentences is a sentence. FEET Formal Grammar for Propositional Logie ABNF (Backus Naur « j peerless Form) grammar of sentences in propositional logit 8 ® TECHCAL PUBLICATIONS® - An up trust fr knowiedg ‘ge 1 are used 10 group the s AG eaning are controled, ie a sentence of propositional calculus if and only if i can be for ts ara Kroniet® Lapel beats 03 ee > Atomie sentence | Complex sentence. sentene sentence — True | False | Symbol. atomic Snbol 7 P1QI RL Comples sentence -9— Sentence | (entence » Sentence) | (Sentence v Sentence) | (Sentence = Sentence) | (Sentence <= Sentence). sce that grammar very sit about parentheses. Every sentence constructed with weg Ceanectves must be closed in parentheses 4 order of precedence in propositional loge i fom Nghest to lowes The recedence order, are shown below, a operators with veir pt Hence, the sentence, axvYnZaw is equivalent to, (anv aD) We [EEG Inference in the Knowledge Base ‘A Iogical inferencing means to decide whether KB Fa for some sentence General inferencing algorithm (truth table enumeration algorithm) 1) This algo is used for deciding entailment in propositional logic. 2) It works like backtracking search algorithm. 5) It performs a recursive enumeration of a frte space of assignments variables. |) This algorithm is sound, because it implements directly the definition of entailment. 5) This algorithm is complete, because it works for any knowledge base and and always terminates (As there are finite models to check). Time complexity of algorithm is 02") (assuming m symbol in all, 2” models). 7). Space complexity of algorithm is O(n), 4 then there are because the enumeration is depth first. Every known inference algorithm for propositional logic has ‘worst case time complexity which is exponential in the size ofthe input. TECHNICAL ‘PUBLICATIONS® - An up thrust for knowledge = revs tt EBL Proporitcnatentatment is Ce : Is CONP-complote {CO-NP stands for Complement « Nonpolynomial time, where as CO-NP-complete means hardest problem in CON?) Tho functions for truth table enumoration algorithm + Functlon TT Entalls (KB, w) Returns True or F Input « (KD, (a sentence tn propositional loge, rence in propositional loge. ¢ knowledge base) the query, a ‘Symbols « A list of the proposition symbols in KB and « return TT CHECK: Function TT:CHECK-AL MLEMPTY (symbole th {PLeTRUE (KE, modal) thon roturn PL-TRUE (, model) return true LL (KB, spb 1) (Ki, c, symbols, model) runs true or fase P + FIRST (eymbol REST (ym) toturn T-CHECK ALL (KB, a, rost, EXTEND (P, true, model) and TE-CHECK-ALL(KE,«,tost, EXTEND (P, fase, todo) Note that, 1) In truth-table enumeration algorithm for, deciding propositional entailment stands for truth Table 2) PLLTRUE returns true; if a sentence holds within a model. 43). The variable model represet Ze a ‘presents a partial modelan assignment to only some of 4) The function call EXTEND (P, retums: 1, : P has the value true, nee) vm sOwveU PusuicanoNs® say ww Seen partial model in | a 1 define rules for determining the truth cf a sentence with respect 10 pe seman rhe fede. In propositional fg, & model snply tes the truth value tor OF sat ey pontine a semantic for propositional loge should ypectly how to compute the tet value he een, given # model, TRS is done scurvy All gentences are const fay vere aentences and the five connectives from 0 Intorprotation of Propositions ‘an Interpretation of a set of propositions is the assignment of a truth value, ‘iiher T or F, to each propositional symbol, 4) The symbol tru always assed T, 4) The ayrmbo false Is assigned F. 4) Propositions = 1). A proposition is a statement which in English would be declarative sentence 1 Every proposition Is ether true o false. jositions are sentences, elther true or false but not both. » 1i) Props ju) Sentence is smallest unit in propositional logic. | 4) Ia proposition is true then its truth value is "rue" i vi) Ifa proposition is false then its truth value is "False" EZED Computation of Truth Valuos 1) The truth assignment of negation, P, where Pis any propositional symbol, is F, if the assignment to P is T, and T ifthe assignment to P is F. ‘The truth assignment of conjunction, », is T only when both conjuncts have truth value T; otherwise it is F. 3) The truth assignment of disjunction, v, is F only when both disjuncts have truth value F, otherwise it is T. 4) The truth assignment of implica 2 ion, -» is F only when premise or symbol before th value of the consequent or symbol after the : ‘ the implication is T and the tru 3 implication is F; otherwise it is T. 2 5) The truth assignment of equivalence,» is T only when both expressions Fave the 7 same truth assignment forall possible interpretations; otherwise itis Fs ae TECHNICAL PUBLICATIONS® - An up Sus or owes AtiicialIntligence ‘or exampl Sentence 1) "Grass is green? W2+ses | H) "Close the door” | 19) "Is thot out side ? | PX > 2 where X is variable | wo-x aa | | 5.10 ‘Truth value “Te “False” FEED truth tabio 1) The teh assignments of compound propositions are often described py « tables. 2) A teh bl sts all possible rth vale assignments to the atomic stan expression and gives the truth value of the expression for cach pe 3) Thus, a tuth table enumerates all posible words of interpretation tht may sven to an expression. For exampl ‘The truth table for P-» Q, lists truth values for each possible truth assignment nt aie a aatan tad Gccton ero a ivalnce oar defined in a Pele "acne ioe aware fer T F T 1 fet. F F F 7 a F T T t 7 oan [eae F Tr Tr T ye 4) Equivalence of two sentences in propositional logic : Two expressions in the same value under all trut By demonstrating that two di ‘identical truth tables, we can expressions P, Q and R. TECHNIC = WNCAL PLBLCATONS® an wp ons nowmape ¥ ‘+ Truth Table demonstrating the similar fashion, Yes Yes No No No (Since X fs not defined) No {ais unknown that what i Seer ir i equal to alr or wate” has no mening). equivalence of P - Q and Pv Q. % and “Water ig Propositional calculus are equivalent if they have ith value assignments, ’ a | ferent sentences in the propositional calculus bit Prove the following equivalences for z geal ats on cst a (af) #P (Double-negation elimination) wy ey HPQ iy The conteapostive ow : @7QeOQaar) tw) De Morgan's Law : (PV Q)8(PATQ) and 5 (P 4Q) e(aPvQ a 1) The commutative law 7 6)" QAP and PV QeQvr) sy The associative law : (@ADAR BCA Qa) si) The associative law + (ev QV Rv QvR) ii) The distributive law: Py QaRaPvQavR ix) The distributive law + PAQvRePaQy aR 3) Implication elimination : © Qa(aPvQ ai) Biconditional elimination ¢ @ 2 Qe(PFAAQ>P) ERZA vatiaity [A sentence is valid if itis true in all models. " For example : (Parrot is green v Parrot is not green) is valid, one of the two holds. Valid sentence is called Tautology. They are necessarily true. EZ satisfiability A sentence is satisfiable if it is true in some model. If a sentence § is true in model m, then we say that m satisfies. $ or m is a model ofS. TECHNICAL PUBLICATIONS® - An up thrust for knowedze Knowiody® LogcalAgants and = s-19 Reorsataon uns sn aid charactorization : : er peerinistie Artic Love! Koa na, Inttigonce 512 sec ry TERED 4 conic eanpn ot rovege usd Age ‘A wumpus world agont es omes exacly speci Wompus World is a cassie atciat — [peo es yes outcomes 2 *P intelligence problem, which is used to 4 | stench a sratic Gemonstate various apes ech [Bite B Fes - Wampus and Pits donot mave Simulation, as well as other Al concepts. Gents] HD la] jserete Wampus was an early computer ° [ee a ao game in which an agent had to explore an cave made up from a series of , /g6«?? iis, 4) Sele % eee ey) ee ea ne hoa Bie, Yes Wampus is essentially a natural feature. rooms in the cave, there was a Wampus rally observable Which would kill the agent if it entered rm » Ne only Local perception. that room. Some rooms contained pit and the agent would die if it entered any of those rooms too. The agent had + 2 3 7 fone arrow with which it could kill the Fig, 6.4.1 A wumpus world agent and its Wumpus. The goal was to locate the Ca) poring te wurmpus world £ See he knowledge base iil contains the rules of he environment. 2 sant Location = [1, 1] the cave and return to the start without getting killed. is 2a [34 [sa B= Agent © Performance measure 3 ae G = Gitar gold car mem ale aren ee os ss - ‘Squares adjacent to pit are breezy oe = Glitter iff gold is in the same square. + Shooting kills Wumpus if you are facing it It screams, = Shooting uses only arrow. Fig, 5.4.2 Exploring wumpus world * Grabbing picks up gold if in same square, Percept : {- Stench, ~ Breeze, ~ Glitter, Bump, ~ Scream} Action : Move to safe cell [2, 1} 3) Location : [2, 1] Percept : [- Stench, Breeze, — Glitter, + Bump, ~ Scream Infer : Breeze indicates that there isa pit in [2, 2} oF [3,1 Action ; Return to [1, 1] to try next safe cell ~ Releasing drops the god in same square, = You bump if you walk into a wall + Actuators: Lett tur, right tum, forward, gab, release, shoot © Sensors: " Stench, breeze, gtr, ump, sere, TECH PUBLICA 18° A st oon TECHNICAL PUBLICATIONS - Anup tt for kowiedse Seer i: . 543 Exploring wumpus world «IT 4) Location : [1,2] Percept : (Stench, ~Breeze,~ Gitter, Infer : Wumpus in [1,3] or [2-2] — Bump, ~ Scream, YET — not in [1 1} ‘Thus. not in [2,2] or stench would have been detected in [2, 1] : “Thus . Wumpus isi [1 3) [2.2] i safe because of lack of breeze in 1,3) Tras — pit in 33] ‘Action : Mave to next safe cell [2 2) rape B = some © = Breeze fas pas ee a i G = citer gots I on = Sehsquare aay fe fz] e = me leo S = Stench i fete pasar] (Mane we [yo = vous ox | id Fig. 544 Exploring wumpus world - IIL EEG Reasoning Pattems in Propositional Logic ‘There are standard pattems that can be appli ‘an be applied to derive chains of conclusions thst lead to the desied goal. These pattern of inference are called inference rules. oo ——— TECHCAL PUBLICATIONS? - Anup wnt or inontodte resesy mete tat tle in cn te at were eae ieee are found in the Knowles tase The conciancn of the roe Tet Fre eegatdlest of what ee te cat aoe inference Rules jy Mods Fonens is represen’ 2 oaks oe sy mesns that eda are BER por example ~ wo statements, Cigna Pole Ahead » Signal Red) = Sep fp (Gigeal Pole Ahead » Sigal Red) are sven then “Stop” can be inferred. ‘pAndatimination : It ays that from conjinction any cones can be tere nis represented 25, an spenever any sentences of the form = then the sentence P can be inferred. For example - From the sentence, (Signal Pole Ahead Signal Red) The inference, "Single Red”. can be drawn, 3) Unit resolution (Pv Q, Qt PI: \ terals and a literal and produce : Unit resolution rule takes a clause - a disjunction of lit Aaprcetaon tg haa eee Tk vi 1VTigiv where each 1 i literal; and m are complementary literals. Ht means that ong, negation of other} 4) Resolution : The unit resolution rule can be generalized t0 the Fill esolution yy By ov Iyer V oo Mn TT ov my where {and my ate complementary literals. If we ate dealing only With cag length two we could write this 8 ih “That is, resolution takes two cutes and produces anew clause containing at gy iterats ofthe peo orginal awses exept the vo complementary itera Tis single inference rule tat ils a complete inference algorithm when ou with any complete serch got. CConjunctive normal form 1) The resolution rue applies only to disunctons of literals, soit Would’seem tb relevant only to knowledge bases and queries consisting of such disjnctions, 2) Every sentence of propositional logic is logically equivalent to a conjunction of dlisjunctions of literals. 3) A sentence expressed as a conjunction of dsjanctions of literals is said to be CConjuncive Normal Form (CNP) 4) 1s used for representing sentences ‘Steps for converting propsitional logic sentences into CNF 1) Bliminate 2, replacing a = B with (496) Bra) : For example - Consider following statement, ay Pa Alter eliminating ein above sentence, 1,19 2 Pa (Ph2 Pa), 1) 2) Eliminate +, replacing @ +6 with + av © Bay Pav Fa) ACIP 2)v Pav By 1) TECHCAL PUBLCATIONS® Anup tt ohm sand Howie? br Logeat agers and Know ee eee equites ~ t0 appear only in literals, so we “move ~ inwards" by repeated 9 ean oe lowing sand equ ie tin tg 1 4B) #4 YP) De Morgan) in the example, We require just one application of the ast rule. eB Pav Pad AG (P.24 Pry By) et caaatcnes eee eed 1 soba i ong es eet Go Br¥ Pi2¥ Pa) AGP GPav Biv By 2) ee ce now in pee we efutation completeness : 4 ears that resolution can always be used to ithe confi or refute a eentence but a rey be used fo enumerate true sentences. a0 esoition algorithm ) takes input as knowledge ase [a sentence in propositional logic] and a (i {uery = which sa sentence in propositional lg. 2 ts output is true or false means tha Knowledge base do zesolves a. It shows thar knowledge base [> a, for this it willbe shown that (KB 1 =<) 4) Steps are as follows, ) (KB 44) is first converted in to CNF i) Resolution rule is applied to the resulting clauses. ii) Each pair that contains complementary literals is sesolved to produce now clause. iv) This new clause is added to set if itis not already present. ¥) Step (i) to (iv) are repeated until one of the two following situations occurs. no new clauses that ean be added, in which case ct does not entail B OR 2) There are 8) An application of the resolution rule drives the empty clause, in which ease a entails TECHNICAL PUBLICATIONS - An up thrust for knowiedye Logical Agen, Ati tince ganar Pa iy east Completoness of resolution ans that it Will Urey Produces The algorithm of resolution is complete. It me able from them. It finds checking all clauses in KB and all the lauses dev clauses set which has all derivable claus ‘query against the closure set. Ground resolution theorem It states that ~ “Tf a set of clauses is unsstls clauses contains the empty clause” “The ground resolution theorem is called as com Forward and Backward Chaining [EEE The Horn Clause 1) Knowledge base contains simple Hor clause is a disjunction of literals of which at mé ppeteness theorem for resolution, rere 2) Hom clause : = Ithas atmost one positive literal = Horn clause with exactly one position literal are calle = The positive literal is called HEAD of the clause. tral is called as BODY of the clause. ith no negative literals is called as fact. sd definite literal. = The negative li =A definite clause Horn clause with no positive iteals can be written as an implication whose conclusion is the literal false, Such horn clause can be used to represe, integrity constraints (conditions) in the databases. 3) Inferencing with HORN clause - For inferencing with Horn clause forward and backward chaining methods are used. EERE] Forward Chaining Method The concept : 1) tis general concept of datacriven ress c 1s gral concept of dated soning in which the focus of attention 2) It can be used within an agent to derive conclusions from incoming percepts. cued by te nl caf Hanae eerstion spb! Ce a) 4) I initially stars from known facts (positive literals) in the knowledge base TECHNICAL PUBLICATIONS® - An up thrust for knowledge room exiting CAUSES set. It checks yee s Pon Jiable, then the resolution closure tay oo Legal Agents and Krowied2® "proposer Loe — Papas ie OPE se premises of 20 impliatin ate known then is cocusion i ded "0 wal cing ager £ fod aT 7, ae ‘peturns true OF false uno DA, sislon symbol c Fetes; Count, a table, indexed by clause italy the numberof premises Inferred, a table, indexed by symbol, each entry intaly false. ‘agenda, alist of symbols, inital the symbols known to be true is KB. fats not empty do ue agenda i DOE OEDEY pe PoP (egenda) rod tpl do aioe infor (Pl saferrod [pl U9 or each Hor clause (In whose premise tof propositional Hor clauses, q. the quer, boeal ‘p appears do decrement count (el) if count fe] = 0 then do if HEAD [e] = a then rerum true PUSH (HEAD [el, agenda) return false. Properties of forward chaining algorithm "Forward chaining algorithm is sound it means that every inference is essentially an application of modus ponens. 2) Forward chaining is complete. = Tt means that every entailed atomic sentence will be derived. Algorithm constructs inferred table for final state. In this table each symbol inferred during process has true value and false for all other symbols. 3) For example : Consider a simple knowledge base of Ho 1. ASQ 2 BACSA 1m clause, as given below. TECHNICAL PUBLICATIONS® - An up trust fr owed? Artic cial Inteligonco aa 3. DaBac 4 EAASB 5. ExD=B ee 7D AND - OR graph for above Hom clause knowledge bate 1) Multiple inks joined by are indicate 2 conjunction. 2) Multiple links without an arc indicate @ disjunction 3) ‘The links should be proved. 4) Inforce propagates up the graph. 5) Before proceeding ahead, from conjunction point, the propagation waits until all the conjuncis are kaown, + The order of clauses considered by forward chaining algorithm for inferencing query Q: 1) E and D are given true. 2) E and D implies B also B and A implies B. Fig. 56.1 AND. orm clause a? 3) D and B implies C 4) Band C implies A. 5) A implies Q 6) Therefore Q is tue. EEE] the Backward Chaining Algorithm The concept 1) 1k works backwards from the query ie if query Q is known to be true thet work i needed 2) Algorithm finds those implications inthe knowledge base that conclude Q IF all the premises of one of those impli se implications can be proved true (by backet chaining) then Q is tre. ® va 3) Backward chaining works doven the graph starting at it amd Joon chat fom te Bn for apes nS AN 4) Backward chaining isa form of goairected reasoning. 1 °. "CHNCAL PUBLICATIONS? An up thst or knowledge an Lege Agate ard rowing” 5:2 Rapnsataten vary Prepontcrd Pe ier ga for answering question sch as “what scion should do now”? re packard chaining 1 equal too tess thin linear in the se of cs! {As the process touches only relevant facts) co dedge BE weer ample ~ (Refering 1 same knowledge base considered in forward chaininé) es considered by backward chaining algorithm for inferecing 1 of claus (is implied By A. ‘ais implied by B and C is implied by E and A as well as E and D. is implied by E and D. f and D ae given tre in knowledge base from Q we found implications sequence that conclude Q ued ‘Therefor Bi Effective Propositional Inference ences we need efficent algorithm which are based on model checking. vung approach can be backracking such or hiklimbing search. ‘The algorithms snoring 1 suse a for checking sasha. EBB complete Backtracking Algorithm [Davis Putnam Logmann-Loveland ‘Algorithm-{OPLL)] or better infer the concept : 1) It takes input as a sentence in conjunctive normal form, 2) Ie do recursive depth first enumeration of possible models, rropositional logic sentence (in CNF) is satisfiable. which is a set of clauses. 3) It determines if input p: DPLL- A Improved algorithm This algorithm is improvement ov opesition logic (The TT ~ Entails algorithm). There are three improv table enumeration (TT - entails algorithm) as described below, er ealier general inferencing algorithm for rements over trutl 4) Early termination 1) A clause is true if any literal is true. i) A sentence is false if any clause is false TECHNICAL PUBLIGATIONS® - An up thst for knawieda® women even N 2) Puro symbol nour ee i hep ae 1)” Pare symbol : Pure symbol always appears with the same “sign in ay onl FIND-PURE-SYMBOL is a function which retums pure symbol (with value) Example i, (? ent. Inthe tee aes (Av ~B) (mB OH EY AYA EB Pac jp HINDUNFF-CLAUSE i fron which eit ase (vith vals 4i) Make a pure symbol literal true or ull Local Search Algorithms for Inferencing in Propositional Loge ch algorithm like Fill Uap A (> Forward Bump!) ( - ard tic acing Down'”! 4 For ath) gt!» orward™ 1) 11 ci (igo (acing bat Fig. £92 The clteul for datermining if the agent Is at [1, 1], Each fo ‘antation vegister has a similar circult attachod A prelate to circuit base agent eat wi be the initial truth value contained in Bs, 5,

You might also like