Parts-of-speech tagging determines (i) POS per word by sentence meaning (ii) POS per word by sentence structure (iii) All POS for a specific word (iv) All of the above ✅ Explanation: Combines contextual, structural, and lexical analysis.
Resolving ambiguous word meanings uses (i) Fuzzy logic (ii) Shallow semantic analysis (iii) Word sense disambiguation ✅ (iv) All of the above Explanation: Specifically selects context-appropriate word meanings.
Decision support programs help managers with (i) Budget projections (ii) Visual presentations (iii) Business decisions ✅ (iv) Vacation schedules Explanation: Core function is data-driven decision support.
Best approach for game playing (i) Linear approach (ii) Heuristic approach ✅ (iii) Random approach (iv) Optimal approach Explanation: Balances efficiency and solution quality.
Not represented by propositional logic (i) Objects (ii) Relations (iii) Both objects and relations ✅ (iv) None Explanation: Requires predicate logic for relational knowledge.
Knowledge-based agents infer hidden states (i) True ✅ (ii) False Explanation: Combine knowledge + percepts to deduce unseen states.
Inference algorithm completes only if (i) Derives any sentence (ii) it can derive any sentence that is an entailed version (iii) Truth-preserving (iv) it can derive any sentence that is an entailed version and it is truth preserving ✅ Explanation: Must derive all entailed truths without errors.
What are the two basic types of inference (i) Reduction to propositional logic, manipulate rules directly ✅ (ii) Reduction to PL + modus ponens (iii) Modus ponens + rule manipulation (iv) Convert Horn clauses + reduction Explanation: Two fundamental FOL inference methods.
Expert system components (i) Inference engine (ii) Knowledge base (iii) Both ✅ (iv) None Explanation: Core architecture requires both.
Semantic networks use (i) Undirected graph (ii) Directed graph ✅ (iii) DAG (iv) Directed complete graph Explanation: Directional edges represent relationships.
2021 Paper
Parts-of-speech tagging determines (iv) All of the above ✅ (Same as 2020)
Programming a robot by moving it physically is (i) Contact sensing (ii) Continuous-path control ✅ (iii) Robot vision (iv) Pick-and-place Explanation: Also called “lead-through programming”.
Production rules consist of (a) A sequence of steps ✅ (b) Rules + steps (c) Arbitrary representation Explanation: Condition-action sequences.
Not represented by propositional logic (iii) Both objects and relations ✅ (Same as 2020)
Not a property of knowledge representation (i) Representational verification ✅ (ii) Representational adequacy (iii) Inferential adequacy (iv) Inferential efficiency Explanation: Verification isn’t a standard KR property.
Lifted inference uses (i) Existential instantiation (ii) Universal instantiation (iii) Unification ✅ (iv) Modus ponens Explanation: Finds substitutions for identical expressions.
Autonomous Q/A systems are (iv) All of the above ✅ (i) Expert systems (ii) Rule-based systems (iii) Decision tree systems Explanation: May incorporate multiple approaches.
Semantic networks are (i) Knowledge representation ✅ (ii) Data structure (iii) Data type (iv) None Explanation: A graphical KR method.
Semantic network inference methods (iii) True ✅ (i) Intersection search (ii) Inheritance search (iv) False Explanation: Inheritance and intersection are primary methods.
Utility functions use (iv) Linear weighted polynomial ✅ (i) Linear polynomial (ii) Weighted polynomial (iii) Polynomial Explanation: Weights features (e.g., board position).
2022 Paper
LISP function removing first list element (Options corrupted – answer is cdr) ✅ Correct: cdr Explanation: car returns first element; cdr returns rest.
Artificial Intelligence is (i) Putting intelligence into computers (ii) Programming with intelligence (iii) Making machines intelligent ✅ (iv) Playing games Explanation: Core definition.
Best game-playing approach (iii) Heuristic approach ✅ (Same as 2020/2021)
Face Recognition uses (ii) Applied AI approach ✅ (i) Weak AI (iii) Cognitive AI (iv) Strong AI Explanation: Solves specific real-world problems.
Not common AI language (iv) Perl ✅ (i) Prolog (ii) Java (iii) LISP Explanation: Python/LISP/Prolog dominate AI.
Not in propositional logic (iii) Both objects/relations ✅ (Consistent across years)
Inference algorithm requires (iv)✅ (Same as 2020/2021)
Blind search is (i) Uninformed search ✅ (ii) Simple reflex search (iii) All mentioned Explanation: Uses no domain knowledge.
Utility functions in games (iv) Linear weighted polynomial ✅ (Same as 2021)
Semantic networks use (ii) Directed graph ✅ (Same as 2020)
2023 Paper
Intelligent agent is (ii) System perceiving environment to maximize success ✅ (i) Rule-following device (iii) Routine software (iv) Human-like robot Explanation: Russell-Norvig definition.
Heuristic-prioritized search (- answer is A algorithm)* ✅ Correct: A* Explanation: Uses heuristic f(n)=g(n)+h(n).
Simulated Annealing is (ii) Heuristic search ✅ (i) Uninformed search Explanation: Probabilistic heuristic method.
Exploration problem when (ii) Agent doesn’t know states/actions ✅ (i) Knows states/actions (iii) Knows only actions (iv) None Explanation: Agent lacks environment knowledge.
Tokenization purpose (ii) Splitting text into words/phrases ✅ (i) Removing data (iii) Translation (iv) Sentiment analysis Explanation: Fundamental NLP preprocessing.
Making logical expressions identical (iv) None (Unification is correct) ✅ (i) Lifting (ii) Inference Explanation: Unification enables FOL inference.
Handling vague information (iv) Fuzzy Logic ✅ (i) Functional Logic (ii) Boolean Logic (iii) Human Logic Explanation: Models partial truth (e.g., “cheap”).
Neural network activation function (ii) Introduces non-linearity ✅ (i) Initializes weights (iii) Combines inputs (iv) Normalizes data Explanation: Enables complex pattern learning.
“Epoch” means (iv) One full training pass ✅ (i) Network layers (ii) Hidden nodes (iii) Dataset size Explanation: Fundamental training cycle term.