Maurizio Pighin

An empirical analysis of fault persistence through software releases

This work is based on the idea of analyzing the behavior all over the life-cycle of source files having a high number of faults at their first release. In terms of predictability, our study helps to understand if files that are faulty in their first release tend to remain faulty in later releases, and investigates the ways to assure a higher reliability to the faultiest programs, testing them carefully or lowering the complexity of their structure. The purpose of this paper is to verify empirically our hypothesis, through an experimental analysis on two different projects, and to find causes observing the structure of the faulty files. As a conclusion, we can say that the number of faults at the first release of source files is an early and significant index of its expected defect rate and reliability.

A predictive metric based on discriminant statistical analysis

The purpose of this paper is to put forward a methodology based on discriminant statistical analysis, which, by evaluating a series of structural parameters of a program, is able to predict its risk level, namely how prone it is to containing faults. The metric was constructed in an experimental context in which the high number of available observations (almost 350,000 lines of code) allowed us to divide the body of data into two parts, one for the effective creation ofthe model, the other as an objective, statistical means by which the proposed methodology could be evaluated. The conclusions we have reached allow us to assert that the basic assumption holds true, and that this particular type of analysis can be used in a fixed environment, during the release and testing sf software as a predictive metric for the early identification of dangerous programs.

Software Fault Prediction using Language Processing

Accurate prediction of faulty modules reduces the cost of software development and evolution. Two case studies with a language-processing based fault prediction measure are presented. The measure, refereed to as a QALP score, makes use of techniques from information retrieval to judge software quality. The QALP score has been shown to correlate with human judgements of software quality. The two case studies consider the measures application to fault prediction using two programs (one open source, one proprietary). Linear mixed-effects regression models are used to identify relationships between defects and QALP score. Results, while complex, show that little correlation exists in the first case study, while statistically significant correlations exists in the second. In this second study the QALP score is helpful in predicting faults in modules (files) with its usefulness growing as module size increases.

Increasing diversity: Natural language measures for software fault prediction

While challenging, the ability to predict faulty modules of a program is valuable to a software project because it can reduce the cost of software development, as well as software maintenance and evolution. Three language-processing based measures are introduced and applied to the problem of fault prediction. The first measure is based on the usage of natural language in a programs identifiers. The second measure concerns the conciseness and consistency of identifiers. The third measure, referred to as the QALP score, makes use of techniques from information retrieval to judge software quality. The QALP score has been shown to correlate with human judgments of software quality. Two case studies consider the language processing measures applicability to fault prediction using two programs (one open source, one proprietary). Linear mixed-effects regression models are used to identify relationships between defects and the measures. Results, while complex, show that language processing measures improve fault prediction, especially when used in combination. Overall, the models explain one-third and two-thirds of the faults in the two case studies. Consistent with other uses of language processing, the value of the three measures increases with the size of the program module considered.

A new methodology for component reuse and maintenance

Component reuse and maintenance requires the development and utilization of specialized tools. In order to be correctly used, any software component needs to be properly understood, engineered and catalogued. Different information about components have to be organized, developed and retrieved during the process. The article discusses a methodology based on information retrieval techniques for automating the cataloguing of existing software components. We describe with an experiment the utilization of the system with prototype examples of reuse and maintenance and finally we evaluate the results of the experimental phase.

A Methodology Supporting the Design and Evaluating the Final Quality of Data Warehouses

The design and configuration of a data warehouse can be difficult tasks especially in the case of very large databases and in the presence of redundant information. In particular, the choice of which attributes have to be considered as dimensions and measures can be not trivial and it can heavily influence the effectiveness of the final system. In this article, we propose a methodology targeted at supporting the design and deriving information on the total quality of the final data warehouse. We tested our proposal on three real-word commercial ERP databases.

Reducing corrective maintenance effort considering module's history

A software package evolves in time through various maintenance release steps whose effectiveness depends mainly on the number of faults left in the modules. The testing phase is therefore critical to discover these faults. The purpose of this paper is to show a criterion to estimate an optimal repartition of available testing time among software modules in a maintenance release. In order to achieve this objective we have used fault prediction techniques based both on classical complexity metrics and an additional, innovative factor related to the modules age in terms of release. This method can actually diminish corrective maintenance effort, while assuring a high reliability for the delivered software.

A formative evaluation of information retrieval techniques applied to software catalogues

Software catalogues are crucial ingredients of any development process based on reuse. In fact, to be reused, software components must be first properly catalogued and then easily found and understood. In this paper we discuss how information retrieval (IR) techniques can be utilized to handle a catalogue derived from an existing software package and how their effectiveness can be assessed. An empirical evaluation of a prototype system handling an industrial-level software package is described and its outcomes are discussed. The conclusion of the formative evaluation is that such techniques are effective and that both expert and non-expert users find them useful and satisfying.

Fault-Threshold Prediction with Linear Programming Methodologies

This paper presents a new experimental methodology that operates on a series of programs structural parameters. We calculated some simple metrics on these parameters and then we applied linear programming techniques on them. It was therefore possible to define a model that can predict the risk level of a program, namely how prone it is to containing faults. The new system represents the software files as points on an n-dimensional space (every dimension is one of the structural attributes for each file). Starting from this model the problem to find out the more dangerous files is brought back to the problem to separate two sets in Rn. A solution to this linear programming problem was achieved by using the MSM-T method (multisurface method tree), a greedy algorithm, which iterative divides the space in polyhedral regions till it reaches an empty set. The classification procedure is divided in two steps: the learning phase, which is used to tune the model on the specified environment and the effective selection. It is, therefore, possible to divide the n-dimensional space and find out the risk-regions of the space, which represent the dangerous files All the process was tested in an industrial application, to validate the soundness of the methodology experimentally. A comparison between linear programming and other risk definition techniques was provided.

Program risk definition via linear programming techniques

The paper defines an innovative experimental metric which operates on a series of structural parameters of programs: by applying linear programming techniques on these parameters it is possible to define a measurement which can predict the risk level of a program. The new proposed model represents the software modules as points in a dimensional space (every dimension is one of the structural attributes for each module). Starting from this model the problem to find-out the more dangerous files is brought-back to the problem to separate two sets. The classification procedure is divided in two steps: the learning phase which is used to tune the model on the specified environment, and the effective selection which is the real measure. Our engine was built using the MSM-T method (multisurface method tree), a greedy algorithm which iteratively divides the space in polyhedral regions till it reaches a void set. It is thus possible to divide the n-dimensional space and find out the risk-regions of the space which represent the dangerous modules. All the process was tested in an industrial application, to validate experimentally the soundness of the methodology.