Inference and Resolution Questions and Answers

Q: What is the cut in Prolog?

The cut commits the search to the choices made so far, which improves efficiency and destroys the declarative reading, since the program's meaning now depends on the search order. Clause removal, depth limits and syntax are different notions. A construct that buys performance by breaking the correspondence between what a program says and what it does is a genuine engineering compromise, and the debate over its use has run for decades. Naming the cost of a choice alongside its benefit is what turns a fact into a design decision.

Q: Why does the cut damage the declarative reading?

With alternatives discarded, whether a goal succeeds depends on which clauses were tried first, so the logical reading no longer predicts the behaviour. Programs using cut must be read procedurally. Negation is a separate feature, some cuts do have logical readings, and no clauses are deleted. A feature whose cost is the loss of the language's principal claim is worth using sparingly and documenting carefully wherever it appears.

Q: What is negation as failure in a logic programming system?

The rule succeeds for not p exactly when the attempt to prove p fails finitely, which implements the closed world assumption operationally. It is sound only when the program is complete about p. The third option overstates it slightly by ignoring the finiteness requirement, and the others misdescribe it. An operational rule standing in for a semantic assumption must be used only where the assumption holds, which is a condition rarely checked in practice. The quantity that varies and the quantity that does not are worth naming before any substitution is attempted.

Q: Why must the failure be finite for negation as failure to be sound?

If the attempt to prove p runs forever, the system has established nothing, and concluding not p would be unjustified. Implementations that time out and report success for the negation are unsound in exactly this way. Memory, goal identity and syntax are not the issue. A soundness condition that depends on the behaviour of a search, rather than on the logic, is easy to violate and hard to notice.

Q: What is the practical summary of inference methods for definite clauses?

Both chaining directions are sound and complete for definite clauses and run in time linear in the knowledge base, which is why this fragment underpins so much practical work. Resolution is also complete but the specialised procedures are far more efficient. The pessimistic options are wrong. A fragment combining adequate expressiveness with linear inference is the sweet spot the whole expressiveness against tractability discussion has been circling. The shared properties explain why these belong to one family, and the differences explain why more than one exists.

Q: What is the overall lesson of the inference topic?

Resolution supplies one rule sufficient for refutation, and the strategies, orderings and redundancy elimination around it are what turn an unusable search into a working tool. Tractable fragments and satisfiability solvers cover much of the rest. Neither pessimism nor unqualified optimism is warranted. That the rule is simple and everything around it is engineering is a fair summary not only of this topic but of much of the field.

Q: What does a satisfiability solver return when a formula is satisfiable?

The solver returns a model, which can be checked independently in linear time, so a positive answer is easy to verify. A negative answer requires a refutation proof, which good solvers can also emit. Counting models and eliminating variables are different problems, the first being considerably harder. That a yes answer is cheaply checkable while a no answer needs a proof is an asymmetry worth exploiting whenever a tool's output must be trusted. The quantity that varies and the quantity that does not are worth naming before any substitution is attempted.

Q: How can a negative answer from a solver be checked?

Modern solvers can emit a resolution proof that an independent checker verifies, so trust in a complex solver reduces to trust in a simple checker. Cross checking with a second implementation is weaker, and enumeration is infeasible. Providing a certificate that can be checked by a smaller and simpler program is the same de Bruijn principle that governs interactive provers, applied to an automatic tool.

Q: Why is emitting a proof valuable for an automatic solver?

A satisfiability solver is tens of thousands of lines with aggressive optimisations, and no one can audit it, whereas a proof checker is a few hundred and can be verified. Speed is unaffected and often reduced, regulations vary, and the proofs are far too large to read. Concentrating the assurance in the smallest component that can supply it is a principle that applies to any complex tool whose output must be relied upon.

Q: What is a certificate in the context of a solver's output?

A satisfying assignment or a refutation proof lets the answer be confirmed without trusting the tool that produced it, which is what a certificate means. Testing, resource records and vendor guarantees are quite different assurances and much weaker. Designing a system to emit checkable evidence, rather than asking to be believed, is an idea whose applicability extends far beyond automated reasoning.

71.

What is the cut in Prolog?

A.A limit on the depth to which the search may proceed
B.A separator between the head and body of a clause
C.A control construct discarding alternative solutions
D.A rule that removes a clause from the knowledge base

Answer: C

The cut commits the search to the choices made so far, which improves efficiency and destroys the declarative reading, since the program's meaning now depends on the search order. Clause removal, depth limits and syntax are different notions. A construct that buys performance by breaking the correspondence between what a program says and what it does is a genuine engineering compromise, and the debate over its use has run for decades. Naming the cost of a choice alongside its benefit is what turns a fact into a design decision.

Discuss in Forum

72.

Why does the cut damage the declarative reading?

A.The cut introduces negation into an otherwise positive program
B.Cuts cannot be given any logical interpretation at all
C.The cut removes clauses that the declarative reading requires
D.The program's meaning becomes dependent on the search order

Answer: D

With alternatives discarded, whether a goal succeeds depends on which clauses were tried first, so the logical reading no longer predicts the behaviour. Programs using cut must be read procedurally. Negation is a separate feature, some cuts do have logical readings, and no clauses are deleted. A feature whose cost is the loss of the language's principal claim is worth using sparingly and documenting carefully wherever it appears.

Discuss in Forum

73.

What is negation as failure in a logic programming system?

A.Concluding a negated goal when the positive goal fails
B.Reporting failure when a negated goal cannot be proved
C.Treating every unproved goal as false in the knowledge base
D.Negating each clause before adding it to the knowledge base

Answer: A

The rule succeeds for not p exactly when the attempt to prove p fails finitely, which implements the closed world assumption operationally. It is sound only when the program is complete about p. The third option overstates it slightly by ignoring the finiteness requirement, and the others misdescribe it. An operational rule standing in for a semantic assumption must be used only where the assumption holds, which is a condition rarely checked in practice. The quantity that varies and the quantity that does not are worth naming before any substitution is attempted.

Discuss in Forum

74.

Why must the failure be finite for negation as failure to be sound?

A.Finiteness is required by the syntax of the language
B.An infinite search neither proves nor disproves the goal
C.Infinite failures consume more memory than is available
D.The negation would otherwise be applied to the wrong goal

Answer: B

If the attempt to prove p runs forever, the system has established nothing, and concluding not p would be unjustified. Implementations that time out and report success for the negation are unsound in exactly this way. Memory, goal identity and syntax are not the issue. A soundness condition that depends on the behaviour of a search, rather than on the logic, is easy to violate and hard to notice.

Discuss in Forum

75.

What is the practical summary of inference methods for definite clauses?

A.No complete procedure exists for definite clauses at all
B.Model checking is the only feasible method for this fragment
C.Forward and backward chaining are sound, complete and linear
D.Only resolution is complete for this fragment of the logic

Answer: C

Both chaining directions are sound and complete for definite clauses and run in time linear in the knowledge base, which is why this fragment underpins so much practical work. Resolution is also complete but the specialised procedures are far more efficient. The pessimistic options are wrong. A fragment combining adequate expressiveness with linear inference is the sweet spot the whole expressiveness against tractability discussion has been circling. The shared properties explain why these belong to one family, and the differences explain why more than one exists.

Discuss in Forum

76.

What is the overall lesson of the inference topic?

A.Inference is always intractable and must be avoided in practice
B.Model checking should be preferred to proof in every situation
C.Complete procedures exist for every logic of practical interest
D.A single rule plus strategies makes mechanical proof feasible

Answer: D

Resolution supplies one rule sufficient for refutation, and the strategies, orderings and redundancy elimination around it are what turn an unusable search into a working tool. Tractable fragments and satisfiability solvers cover much of the rest. Neither pessimism nor unqualified optimism is warranted. That the rule is simple and everything around it is engineering is a fair summary not only of this topic but of much of the field.

Discuss in Forum

77.

What does a satisfiability solver return when a formula is satisfiable?

A.A satisfying assignment
B.A proof that no contradiction can be derived from the clause set
C.The number of distinct assignments satisfying the whole formula
D.A simplified version of the formula with the variables eliminated

Answer: A

The solver returns a model, which can be checked independently in linear time, so a positive answer is easy to verify. A negative answer requires a refutation proof, which good solvers can also emit. Counting models and eliminating variables are different problems, the first being considerably harder. That a yes answer is cheaply checkable while a no answer needs a proof is an asymmetry worth exploiting whenever a tool's output must be trusted. The quantity that varies and the quantity that does not are worth naming before any substitution is attempted.

Discuss in Forum

78.

How can a negative answer from a solver be checked?

A.A negative answer cannot be checked by any practical means
B.By verifying the refutation proof it emits
C.By running a second solver and comparing the answers obtained
D.By enumerating the assignments and confirming none satisfies

Answer: B

Modern solvers can emit a resolution proof that an independent checker verifies, so trust in a complex solver reduces to trust in a simple checker. Cross checking with a second implementation is weaker, and enumeration is infeasible. Providing a certificate that can be checked by a smaller and simpler program is the same de Bruijn principle that governs interactive provers, applied to an automatic tool.

Discuss in Forum

79.

Why is emitting a proof valuable for an automatic solver?

A.Regulations require every solver result to be documented
B.The proof can be shown to a user who will find it readable
C.Trust reduces to a small checker rather than a large solver
D.The proof allows the solver to run more quickly next time

Answer: C

A satisfiability solver is tens of thousands of lines with aggressive optimisations, and no one can audit it, whereas a proof checker is a few hundred and can be verified. Speed is unaffected and often reduced, regulations vary, and the proofs are far too large to read. Concentrating the assurance in the smallest component that can supply it is a principle that applies to any complex tool whose output must be relied upon.

Discuss in Forum

80.

What is a certificate in the context of a solver's output?

A.A statement that the solver has been tested and approved
B.A record of the resources the solver consumed while running
C.A guarantee from the vendor about the solver's correctness
D.Evidence a third party can check independently

Answer: D

A satisfying assignment or a refutation proof lets the answer be confirmed without trusting the tool that produced it, which is what a certificate means. Testing, resource records and vendor guarantees are quite different assurances and much weaker. Designing a system to emit checkable evidence, rather than asking to be believed, is an idea whose applicability extends far beyond automated reasoning.

Discuss in Forum

Loading…

Topics

Other Subjects

Tags: Inference and Resolution Questions and Answers || Inference and Resolution MCQ Questions and Answers || Inference and Resolution GK Questions and Answers || Inference and Resolution GK MCQ Questions || Inference and Resolution Multiple Choice Questions and Answers || Inference and Resolution GK || GK on Inference and Resolution || Artificial Intelligence Questions and Answers || Artificial Intelligence MCQ Questions and Answers || Artificial Intelligence GK Questions and Answers || GK on Artificial Intelligence

Inference and Resolution Questions and Answers

Topics

Other Subjects

About Us

Stay informed about all important Dates & Notifications