.png){kind=icon}
.png){kind=icon}
Return
Observation-based approximate dependency modeling and its use for program slicing
.png){kind=icon}
OA
AI
DOI：10.1016/j.jss.2021.110988.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
While dependency analysis is foundational to much program analysis, many techniques have limited scalability and handle only monolingual systems. We present a novel dependency analysis technique that aims to approximate program dependency from a relatively small number of perturbed executions. Our technique, MOAD (Modeling Observation-based Approximate Dependency), reformulates program dependency as the likelihood that one program element is dependent on another (instead of a Boolean relationship). MOAD generates program variants by deleting parts of the source code and executing them while observing the impact. MOAD thus infers a model of program dependency that captures the relationship between the modification and observation points. We evaluate MOAD using program slices obtained from the resulting probabilistic dependency models. Compared to the existing observation-based backward slicing technique, ORBS, MOAD requires only 18.6% of the observations, while the resulting slices are only 12% larger on average. Furthermore, we introduce the notion of the observation-based forward slices. Unlike ORBS, which inherently computes backward slices, MOAD's model's dependences can be traversed in either direction allowing us to easily compute forward slices. In comparison to the static forward slice, MOAD only misses deleting 0-6 lines (median 0), while excessively deleting 0-37 lines (median 8) from the slice. (C) 2021 Elsevier Inc. All rights reserved.
Keywords:
Dependency analysis
Program slicing
Model learning
MOAD
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
4.1
Papers:
5.4K
Citations:
8.4K
Organization
View more organizations