Research Talks

Frank Schweitzer, ETH Zurich
OSS Projects: Software Structure, dynamics, communication

Archives of Open Source Software (OSS) not only preserve information about the development of the product, i.e. the software, over time but also information about the interaction of the producers, i.e. the developers and users. What do they reveal? Our current project (see www.sg.ethz.ch) sets out to answer this question in an integrated, highly interdisciplinary approach. It addresses the specifics of the two different system levels -- product and production -- and the relation between them, as key to a comprehensive insight into OSS in general, and successfull OSS products in particular. Using a highly data driven approach, we analyse the community dynamics in more than 100 projects, the dependency structure and the change records of 35 Java projects and the evolution of the dependency network. Our investigations show remarkable regularities in the structure and dynamics of OSS which can be reproduced by simple mathematical models, this way challenging established paradigms in software engineering. Understanding the statistical laws of software evolution may help developers to steer development towards favorable architectures. Understanding the link between architecture and project organization may enable new management principles or provide tools for smoothening the interface between software, developers, and users.
 

Martin Robillard, McGill Univ
Why Do We Mine? Understanding Development Contexts with Qualitative Studies

Mining software archives to provide developers with accurate and useful recommendations requires us to understand the context in which archives are created and evolved, and the context in which recommendations would be used. This “context” is a complex, human-centric phenomenon rich in details and variations. In this talk I will present four qualitative studies we recently conducted to better understand the development contexts that we aim to support with software archive mining.
 

Tudor Girba, Univ Bern and Sw-eng. Software Engineering Gmbh
Tooling your way through data

What if it would take you 10-20 minutes to build an interactive tool dedicated to your data? In this demo, we show exactly that. We build various views and visualizations of a data set to expose its inner structure. We also link these views in an interactive browser. At the end, we take a step back and observe some implications of fast prototyping on the process of research.
www.tudorgirba.com/presentations/assessment-through-exploration
 

Jim Whitehead, UC Santa Cruz
As We Might Mine: Implications of Web-based Development Environments on Empirical Software Engineering Research
 

Abraham Bernstein, Univ of Zurich
On time and concept drifts
 

Adrian Kuhn and Oscar Nierstrasz, University of Bern
Software GPS -- consistent positioning for software maps

It is our vision that developers should be able to refer to code as being "up in the north", "over in the west", or "down-under in the south". We want to provide developers, and everyone else involved in software development, with a *shared*, *spatial* and *stable* mental model of their software project. We aim to reinforce this by embedding a cartographic visualization in the IDE. In this talk, we discuss the challenge of establishing cardinal directions on a cartographic software visualization; in particular across many projects or even over the entire universe of all software.
 

Anita Sarma, Univ of Nebraska, Lincoln
Interactive Exploratory Data Analysis
 

Claus Lewerentz, BTU Cottbus
Urban Development in Software Cities

The talk introduces a new approach to represent a software system’s development history in using an urban metaphor. The main contributions are (1) the use of time based hierarchical layout strategies of the city map and (2) the use of ground elevation levels to represent the creation time of software components. The approach creates incrementally adaptable and stable landscapes and city structures. This is necessary to preserve a city’s overall morphology throughout the structural evolution of the visualized software system. The urban landscape layout generation is part of a systematic approach to construct and use these visualizations by adopting the three‐staged cartographic modeling chain. This allows for creating a multitude of uniform and consistent visualizations supporting different analysis scenarios based on development process information as well as structural system properties.
 

Serge Demeyer, Ahmed Lamkanfi, Emanuel Giger, Bart Goethals
Predicting the Severity of a Reported Bug

The severity of a reported bug is a critical factor in deciding how soon it needs to be fixed. Unfortunately, while clear guidelines exist on how to assign the severity of a bug, it remains an inherent manual process left to the person reporting the bug. In this paper we investigate whether we can accurately predict the severity of a reported bug by analyzing its textual description using text mining algorithms. Based on three cases drawn from the open-source community (Mozilla, Eclipse and GNOME), we conclude that given a training set of sufficient size (approximately 500 reports per severity), it is possible to predict the severity with a reasonable accuracy (both precision and recall vary between 0.65-0.75 with Mozilla and Eclipse; 0.70-0.85 in the case of GNOME).
 

Tudor Girba, Univ Bern and Sw-eng. Software Engineering Gmbh
Demo-driven research

Research is less about discovering the fantastic, as it is about revealing the obvious. The obvious is always there. It needs no discovery. It only needs us to take a look from a different perspective to see it. Thus, the most important challenge is not the fight against nature, but against our own assumptions. One way of fighting against our own assumptions is to expose them to other people. I advocate demo-driven research, a way of doing research that puts emphasis on presenting the state of research with any given chance and to any audience willing to listen.
www.tudorgirba.com/research/demo-driven-research
 

Tom Zimmermann, Microsoft Research
Should I fix this bug report?

Tracking bugs is an important activity in software development. In this talk, I will present a study which characterized factors that affect which bugs get fixed in Windows Vista and Windows 7, focusing on bug report edits and relationships between people involved with handling bugs. The findings are reinforced with survey feedback from 358 Microsoft employees and we built a model to predict whether a bug will be fixed. This is joint work with Philip Guo, Nachi Nagappan, and Brendan Murphy.
 

Gregorio Robles, Universidad Rey Juan Carlos
Replication and Mining Software Archives
 

Ahmed Hassan, Queens Univ
Automated Verification of Load Tests
 

Stephan Diehl, Univ Trier
JCCD: The Java Code Clone Detection API
 

Zhou Minghui, Beijing Univ
JBOSS, JOnAS and Apache Geronimo: a comparison of open source middleware development

To understand the participation of developers in open source projects, and the contrasts between them and the commercial projects, we investigate three major middleware projects - JBOSS and JOnAS. We found that during years with more reported defects the median time to resolve a defect is longer, and the longest times to resolve bugs increase because some open bugs accumulate from earlier years and are probably not going to be closed. It appears that based on the developer participation of these three projects are more similar to commercial projects than to Apache or Mozilla analyzed by Mockus et al [1].
 

Daniel German, Univ Victoria
Intellectual Property and Mining Software Repositories
 

Shivkumar Shivaji, UC Santa Cruz
Predicting Software Bugs Using Humans and Machine Learning
 

Sung Kim, Hong Kong Univ
Reproducing crashes using stack traces