From FSL
Jump to: navigation, search

jMiner is a specification mining framework that provides a generic and effective approach for mining parametric specifications from execution traces.

A parametric specification is a formal specification that carries parameters that are bound to concrete object instances at runtime. Parameters are crucial because they enable monitoring systems (such as JavaMOP) to distinguish among different object interactions.

jMiner brings the benefits of using parameters to mining. At the heart of jMiner is the trace slicer, which extracts a set of independent object interactions from the given execution trace. Since each extracted interaction is merely a list of strings (e.g., method names), one can simply use existing learners (such as PFSA learners) or easily develop new ones.

Publications on Mining

Below is the list of all our publications on jMiner.

Mining Parametric Specifications 
Choonghwan Lee, Feng Chen and Grigore Rosu
ICSE'11, ACM, pp 591-600. 2011
PDF, Slides(pptx), ACM, ICSE'11, jMiner, BIB
Personal tools