Logo image
An approach to mining call-usage patterns with syntactic context
Conference proceeding

An approach to mining call-usage patterns with syntactic context

Huzefa Kagdi, Michael Collard and Jonathan Maletic
Proceedings of the twenty-second IEEE/ACM international conference on automated software engineering, pp.457-460
ASE '07
11-05-2007

Abstract

call usage patterns mining software repositories
An approach to mine frequently appearing ordered sets of function-call usages, taking into account their proximal control constructs (e.g., if-statements), in the source code is presented. These ordered sets are termed as call-usage patterns. Additionally, variant usages, such as those with missing or out of order calls, are automatically identified along with their specific contextual location. The approach uses lightweight source code analysis and frequent sequential pattern mining. The hypothesis is that these call-usage patterns embody latent programming rules that developers commonly reuse, for example standard usages of API calls. The variants are an indicator of future changes such as the elimination of non-standard usages and/or bugs

Metrics

15 Record Views

Details

Logo image