Toward a constraint-satisfaction framework for evaluating program-understanding algorithms

Quilici, Alex; Woods, Steven G.

doi:10.1109/wpc.1996.501121

Cited by 13 publications

(10 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our approach is to treat program plan recognition as a constraint satisfaction problem (CSP) [25,35,36,38,39]. In particular, our recognition engine, Layered MAP-CAP, represents plan components as CSP variables and the possible values of these components as the variable domains.…”

Section: Recognizing Plansmentioning

confidence: 99%

“…• The plan recognizer uses our constraint-based approach [25,35,36,38], with the added benefit that the interplay between plans and the types recognized by the static date analyzer (such as year or month) may help to improve its performance. Type information can be used in the node consistency propagation phase of the constraint solver used in the recognition engine [36,Chapter 3], potentially reducing the engine's search space significantly.…”

Section: An Environment For Plan-recognition Experimentsmentioning

confidence: 99%

See 1 more Smart Citation

Program plan recognition for Year 2000 tools

Deursen

Woods

Quilici

Proceedings of the Fourth Working Conference on Reverse Engineering

Self Cite

View full text Add to dashboard Cite

There are many commercial tools that address various aspects of the Year 2000 problem. None of these tools, however, addresses the closely related leap-year problem. In this paper, we provide experimental results that suggest that the leap-year problem can be addressed by plan-based techniques for automated concept recovery. In particular, we provide representative code fragments illustrating the leap-year problem, and we show the results of an empirical study that provides evidence that a plan-based approach can efficiently recognize both correct and incorrect leap-year computations and that the needed plan library is likely to be tractable in size. This paper furthermore argues that plan-based techniques are in fact mature enough to make a significant contribution to the Year 2000 problem itself, despite none of the existing tools making any documented use of these plan-based techniques.

show abstract

Section: Recognizing Plansmentioning

confidence: 99%

Section: An Environment For Plan-recognition Experimentsmentioning

confidence: 99%

Program plan recognition for Year 2000 tools

Deursen

Woods

Quilici

Proceedings of the Fourth Working Conference on Reverse Engineering

Self Cite

View full text Add to dashboard Cite

show abstract

“…In the past, we have performed a variety of experiments to help us understand the performance characteristics of our constraint-based approach [22,12,201. All of these experiments worked with artificially-generated programs and searched for plans that relied on locality constraints (e.g., nearness and containment).…”

Section: An Experiments With "Real" Programsmentioning

confidence: 99%

“…This paper and our previous efforts [12,201 are not the only empirical studies of program understanding algorithms. Many of the previous studies, however, have been devoted to non-efficiency issues, such as plan library completeness for plan instances on a group of similarly sized programs for performing a particular task [5, 71.…”

Section: Related Workmentioning

confidence: 99%

“…In earlier work, we showed how program understanding could be viewed as a constraint satisfaction problem (CSP)', and we performed some initial studies on the efficiency of several constraint-based implementations of existing program understanding algorithms [ 12,201. These experiments involved generating a set of test programs, applying a particular constraint satisfaction approach to locate all instances of a given plan, and measuring the number of constraints evaluated in the process.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

New experiments with a constraint-based approach to program plan matching

Quilici

Woods²,

Zhang³

Proceedings of the Fourth Working Conference on Reverse Engineering

Self Cite

View full text Add to dashboard Cite

In earlier work, we presented some preliminary empirical scalability results for a constraint-basedprogram plan matching algorithm. Those initial experiments had several important shortcomings: they worked with U collection of art$cially generated programs, and they applied a particular; general-purpose constraint satisfaction approach. This paper reports the results of a collection of new experiments that begin to address these dejiciencies. In particulal; we have begun experimenting with programs based on realworld C code, and we have begun exploring new constraint satsfaction algorithms that take advantage of the particular characteristics of the program understanding problem. While not dejinitive, these new experiments provide further support for our earlier results, and they have led to a new approach that provides signiJcant improvements in the scalability of our plan matching algorithm.

show abstract