A TraceLab-based Solution for Creating, Conducting,and Sharing Feature Location Experiments - Online Appendix

This web page is a companion to our 20th IEEE International Conference on Program Comprehension (ICPC 2012) submission entitled "A TraceLab-based Solution for Creating, Conducting,and Sharing Feature Location Experiments"

Dit, B., Moritz, E., and Poshyvanyk, D., "A TraceLab-based Solution for Creating, Conducting, and Sharing Feature Location Experiments", in Proceedings of 20th IEEE International Conference on Program Comprehension (ICPC'12), Passau, Germany, June 11-13 2012, pp. 203-208 (25% acceptance ratio for short papers) [pdf][slides]

A beta version of TraceLab is available by request. TraceLab will be released officially in July 2012. After the release, this online appendix will be updated (datasets, components and experiments) to reflect the latest version of Tracelab.

We make available datasets, TraceLab Templates for feature location techniques (FLTs) and TraceLab Components. TraceLab is a framework for creating, running and sharing experiments using a visual modeling environment. In order to use the TraceLab templates and components, a version of TraceLab needs to be installed from here.

Datasets

We make available as input for TraceLab the datasets of five software systems. These datasets contain gold sets (i.e., set of methods related to a feature) at method level granularity, execution traces, queries, and software corpora at method level granularity. More details about these datasets and their format are presented here.

The datasets for the five software systems are:

TraceLab Templates

Use the following links to download the templates for the VSM, VSM+Dyn, LSI, LSI+Dyn feature location techniques (note that TraceLab has to be installed first):

TraceLab Components

Download the library that contains components that are used in the design of the VSM, VSM+Dyn, LSI, and LSI+Dyn FLTs. This library needs to be copied in the TraceLab components folder. The components contained in this library are for:

  • importing data

  • computing results for FLTs using the effectiveness metrics

  • comparing results using box plots, descriptive statistics, and statistical tests – these components currently are under development and should be ready by April 2012. A screenshot of the box plot components is presented here:

    Box Plots Component

Participants


We gratefully acknowledge financial support from the NSF on this research project.