Motivation

In the latest years, and specially thanks to the huge availability of data about software development that can be obtained from libre (free, open source) projects, the research community is starting to produce, use and exchange large data sets of information. These data sets have to be retrieved, purged, described, and can be published for public consumption by other groups. Their availability allows for the decoupling of research activities (some groups can focus on data retrieval and preliminary analysis, which others can devote to more in-depth analysis without bothering with data retrieval), the reproducibility of research results, and even the collaboration (and competition) in the analysis of data.

All this activity is being presented in several workshops and conferences, but since they are not focused specifically on it, exchange of experiences and discussions are not as deep and fruitful as they could be. This workshop is once again (for the third year in a row) being such a place, hosting researchers in the field to discuss specifically about this kind of data sets, how they are retrieved, how can they be analyzed and mined, how they can be exchanged and complemented, etc.

Main goals

The goal of this workshop is to foster the analysis of public available data sources about software development, and the exchange of data between different research groups.

The workshop is aimed at three different kinds of studies (although other related studies could also be considered):

1. Analysis of specific projects (provided by the organizers, see below).
   The analysis should show a methodology to explore the projects, but also it should show explanations to ``odd'' things that could appear in the data set. For instance, a company-driven project can show different behavior than a community-driven project. The study can be in the field of software engineering, economics, sociology, human resources, and others.


2. Retrieval process and exchange formats of public available data collections about software development.
   The data collections presented should be publicly available, based themselves on public data (so that other groups could reproduce the data collection process), and be related to the field of software development. This includes, but is not limited to, data from source control systems, but tracking systems, mailing lists, websites, source and binary code, quality assurance systems, etc. Although any kind of data collection can be considered, those including information about a large amount of projects will be considered especially appropriate.


3. Data mining activities and new retrieval tools.
   Working with a huge quantity of data is always a complex problem. Data mining techniques are welcome in this section, provided that papers include some conclusions about a specific set of projects. Again, this analysis should show a methodology to explore the data and explanations about the whole process. Cross-analysis of datasets, and specially of those provided by the organizers (FLOSSMole and FLOSSMetrics databases) is specially welcome.
   Also, new tools developed to obtain data from several data sources, such as forums, wikis, bug tracking systems and others fit perfectly here.

Detailed description of data sources

Following the goals described above, the workshop will accept papers about two specific issues (not taking into account the development of new data mining tools):

1. Analysis of two data collections about libre software development: FLOSSMole and FLOSSMetrics
   These collections, already available to any researcher, are offered for analysis by third parties (see below). The studies submitted should detail how they have been used, which part of the information has been considered, how they have been validated or filtered and/or post-processed (if that is the case). The description should be detailed enough to let any other research group reproduce the study.


2. Studies about the data retrieval and preparation for public consumption of data sets in the same realm, which could be proposed for analysis in future editions of the workshop.

FLOSSMole

FLOSSMole (formerly OSSmole) is a set of tools for gathering data (metrics) about the development of free/libre/open source projects. The FLOSSMole project also publishes the resulting analysis about FLOSS projects, and accepts data donations from other research groups. It offers this workshop a complete set of data gathered from the SourceForge development platform and the Freshmeat announcement systems.

More information can be obtained from http://ossmole.sourceforge.net

FLOSSMetrics

At the end of FLOSSMetrics a huge database with data from thousands of projects. Nowadays, the project is already working on the retrieval of the data, with information already available for about 1,000 projects (mainly retrieved from CVS and SVN repositories, but also mailing lists). These results are publicly available at http://data.flossmetrics.org

More information can be obtained from http://flossmetrics.org

The target audience is composed by the research groups interested in empirical software engineering and quantitative studies of the software development processes and methods. This includes not only software engineers, but also researchers from other fields that might use the data for economic, social and other studies.

• Jesus M. Gonzalez-Barahona (Universidad Rey Juan Carlos, Spain)
• Megan Squire (Elon University, USA)
• Daniel Izquierdo-Cortazar (Universidad Rey Juan Carlos, Spain)

• FLOSSMole
• FLOSSMETRICS
• QUALOSS
• SQO-OSS
• QUALIPSO