19 May - Thursday
Session 3
2:00 PM

  Research Papers   Keynotes & Invited Talks
  Experience Reports   Panels
  Education & Training Reports   Research Demonstrations
  Awards & General Remarks   Meetings

Empirical Software Engineering
19 May @ 2:00 PM

St. Louis Ballroom D [Floor Plan]
Session Chair: Alistair Sutcliffe

> Use of Relative Code Churn Measures to Predict System Defect Density
Nachiappan Nagappan and Thomas Ball
> Main Effects Screening: A Distributed Continuous Quality Assurance Process for Monitoring Performance Degradation in Evolving Software Systems
Cemal Yilmaz, Arvind Krishna, Atif Memon, Adam Porter, Douglas Schmidt, and Aniruddha Gokhale
> Effort Estimation of Use Cases for Incremental Large-Scale Software Development
Parastoo Mohagheghi, Bente Anda, and Reidar Conradi


19 May @ 2:00 PM

St. Louis Ballroom E [Floor Plan]
Session Chair: Constance Heitmeyer

> Automatic Discovery of API-Level Exploits
Vinod Ganapathy, Sanjit Seshia, Somesh Jha, Thomas Reps, and Randal Bryant
> Sound Methods and Effective Tools for Model-based Security Engineering with UML
Jan Jürjens
> Improving Software Security with a C Pointer Analysis
Dzintars Avots, Michael Dalton, V. Benjamin Livshits, and Monica Lam


Requirements & Testing
19 May @ 2:30 PM

St. Louis Ballroom C [Floor Plan]
Session Chair: Stefania Gnesi

> Developing Use Cases and Scenarios in the Requirements Process
Neil Maiden and Suzanne Robertson
> Observations and Lessons Learned from Automated Testing
Stefan Berner, Roland Weber, Rudolf K. Keller


Michael Twidale (U. of Illinois at Urbana-Champaign)
Silver Bullet or Fool's Gold: Supporting usability in open source software development
19 May @ 2:00 PM

St. Louis Ballroom A & B [Floor Plan]
Session Chair: Hausi Muller

Biography: Michael Twidale is an Associate Professor of the Graduate School of Library and Information Science, University of Illinois at Urbana-Champaign. Before that he was a faculty member of the Computing Department at Lancaster University, UK. His research interests include computer supported cooperative work, computer supported collaborative learning, user interface design and evaluation, information visualization, museum informatics, how people cope with computers, scenario based design and the application of ethnographic methods to computer systems design and evaluation.

All these involve the use of interdisciplinary techniques in order to better understand the needs of end users and their difficulties with existing computer applications as part of the process of designing more effective systems. Current projects include over the shoulder learning, an investigation into collaborative techniques for improving data quality in databases, and the usability of open source software.

Abstract: At first glance it can look like Open Source Software development violates many, if not all, of the precepts of decades of careful research and teaching in Software Engineering. One could take a classic SE textbook and compare the activities elaborated and advocated in the various chapters with what is actually done in plain sight in the public logs of an OSS project in say SourceForge. For a Professor of Software Engineering this might make for rather depressing reading. Are the principles of SE being rendered obsolete? Has OSS really discovered Brooks' Silver Bullet? Or is it just a flash in the pan or Fool's Gold?

In this talk I will mainly look at one aspect of Open Source Development, the 'problem' of creating usable interfaces, particularly for non-technical end-users. Any approach involves the challenge of how to coordinate distributed collaborative interface analysis and design, given that in conventional software development this is usually done in small teams and almost always face to face. Indeed all the methods in any HCI text just assume same-time same-place work and don't map to distributed work, let alone the looser mechanisms of OSS development. Instead what is needed is a form of participatory usability involving the coordination of end users and developers in a constantly evolving redesign process.

Peter Ayton (City University, London)
How Software Can Help or Hinder Human Decision Making (and vice–versa)
19 May @ 2:00 PM

St. Louis Ballroom A & B [Floor Plan]
Session Chair: Hausi Muller

Biography: Peter Ayton is a professor of Psychology in the Department of Psychology at City University, London, which he joined in 1992. He holds a PhD in Psychology from University College London, (1988). His research has been concerned with judgmental forecasting, human judgement of uncertainty and human choice. Applied research on decision making has been a particular interest and he has been a collaborator on multidisciplinary research projects funded to investigate expert reasoning with toxicological risks, public perceptions of food risk, convicted prisoners' perceptions of recidivism risks and software reliability.

He was a contributing author to the 2001 Assessment Report of the Intergovernmental Panel on Climate Change. He has published numerous papers in international journals and is a member of the International Institute of Forecasters, the Society for Judgment and Decision Making, the European Association for Decision Making and the Experimental Psychology Society.

Abstract: Developments in computing offer experts in many fields specialised support for decision making under uncertainty. However, the impact of these technologies remains controversial. In particular, it is not clear how advice of variable quality from a computer may affect human decision makers. Here I review research showing strikingly diverse effects of computer support on expert decision-making. Decisions support can both systematically improve or damaged the performance of decision makers in subtle ways depending on the decision maker's skills, variation in the difficulty of individual decisions and the reliability of advice from the support tool. In clinical trials decision support technologies are often assessed in terms of their average effects. However this methodology overlooks the possibility of differential effects on decisions of varying difficulty, on decision makers of varying competence, of computer advice of varying accuracy and of possible interactions among these variables. Research that has teased apart aggregated clinical trial data to investigate these possibilities has discovered that computer support was less useful for - and sometimes hindered - professional experts who were relatively good at difficult decisions without support; at the same time the same computer support tool helped those experts who were less good at relatively easy decisions without support. Moreover, inappropriate advice from the support tool could bias decision makers' decisions and, predictably, depending on the type of case, improve or harm the decisions.