Large-scale information retrieval in software engineering - an experience report from industrial application

Software Engineering activities are information intensive. Research proposes Information Retrieval (IR) techniques to support engineers in their daily tasks, such as establishing and maintaining traceability links, fault identification, and software maintenance. We describe an engineering task, test case selection, and illustrate our problem analysis and solution discovery process. The objective of the study is to gain an understanding of to what extent IR techniques (one potential solution) can be applied to test case selection and provide decision support in a large-scale, industrial setting. We analyze, in the context of the studied company, how test case selection is performed and design a series of experiments evaluating the performance of different IR techniques. Each experiment provides lessons learned from implementation, execution, and results, feeding to its successor. The three experiments led to the following observations: 1) there is a lack of research on scalable parameter optimization of IR techniques for software engineering problems; 2) scaling IR techniques to industry data is challenging, in particular for latent semantic analysis; 3) the IR context poses constraints on the empirical evaluation of IR techniques, requiring more research on developing valid statistical approaches. We believe that our experiences in conducting a series of IR experiments with industry grade data are valuable for peer researchers so that they can avoid the pitfalls that we have encountered. Furthermore, we identified challenges that need to be addressed in order to bridge the gap between laboratory IR experiments and real applications of IR in the industry.     

Nanoparticle‐embedded polymer‐dispersed liquid crystal for paper‐like displays

Abstract— A polymer‐dispersed liquid‐crystal (PDLC) matrix template embedded with nano/microparticles can be backfilled/infiltrated with a dye‐doped liquid crystal for a paper‐like reflective display. In this way, a desirable degree of diffusion can be realized to reduce the viewing‐angle dependency of a gain reflector and metallic glare without changing other electro‐optical properties.     

Classification-based objective functions

Backpropagation, similar to most learning algorithms that can form complex decision surfaces, is prone to overfitting. This work presents classification-based objective functions, an approach to training artificial neural networks on classification problems. Classification-based learning attempts to guide the network directly to correct pattern classification rather than using common error minimization heuristics, such as sum-squared error (SSE) and cross-entropy (CE), that do not explicitly minimize classification error. CB1 is presented here as a novel objective function for learning classification problems. It seeks to directly minimize classification error by backpropagating error only on misclassified patterns from culprit output nodes. CB1 discourages weight saturation and overfitting and achieves higher accuracy on classification problems than optimizing SSE or CE. Experiments on a large OCR data set have shown CB1 to significantly increase generalization accuracy over SSE or CE optimization, from 97.86% and 98.10%, respectively, to 99.11%. Comparable results are achieved over several data sets from the UC Irvine Machine Learning Database Repository, with an average increase in accuracy from 90.7% and 91.3% using optimized SSE and CE networks, respectively, to 92.1% for CB1. Analysis indicates that CB1 performs a fundamentally different search of the feature space than optimizing SSE or CE and produces significantly different solutions. Editor: Risto Miikkulainen     

Mining the change of customer behavior in fuzzy time-interval sequential patterns

Comprehending changes of customer behavior is an essential problem that must be faced for survival in a fast-changing business environment. Particularly in the management of electronic commerce (EC), many companies have developed on-line shopping stores to serve customers and immediately collect buying logs in databases. This trend has led to the development of data-mining applications. Fuzzy time-interval sequential pattern mining is one type of serviceable data-mining technique that discovers customer behavioral patterns over time. To take a shopping example, (Bread, Short, Milk, Long, Jam), means that Bread is bought before Milk in a Short period, and Jam is bought after Milk in a Long period, where Short and Long are predetermined linguistic terms given by managers. This information shown in this example reveals more general and concise knowledge for managers, allowing them to make quick-response decisions, especially in business. However, no studies, to our knowledge, have yet to address the issue of changes in fuzzy time-interval sequential patterns. The fuzzy time-interval sequential pattern, (Bread, Short, Milk, Long, Jam), became available in last year; however, is not a trend this year, and has been substituted by (Bread, Short, Yogurt, Short, Jam). Without updating this knowledge, managers might map out inappropriate marketing plans for products or services and dated inventory strategies with respect to time-intervals. To deal with this problem, we propose a novel change mining model, MineFuzzChange, to detect the change in fuzzy time-interval sequential patterns. Using a brick-and-mortar transactional dataset collected from a retail chain in Taiwan and a B2C EC dataset, experiments are carried out to evaluate the proposed model. We empirically demonstrate how the model helps managers to understand the changing behaviors of their customers and to formulate timely marketing and inventory strategies.     

基于BP-GamysBoost的乳腺癌诊断模型

针对乳腺癌数据存在的不平衡性问题,对标准的Adaboost算法进行改进,即首先引入BP神经网络,然后融合模拟退火遗传算法（SA-GA）较强的全局寻优能力和较快的收敛速度,最后通过权重的合理分配,提出BP-GamysBoost算法。同时为验证所提出的新算法BP-GamysBoost的合理性,从UCI机器学习知识库中获取WBCD数据库,比较BP-GamysBoost算法模型与BP模型、BP-GA模型、BP-Adaboost模型的稳定性、准确率、漏诊率、灵敏度等性能指标。最终结果表明,BP-GamysBoost模型在乳腺癌数据库中运行良好,并优于其他3种算法模型。     

Computing planar swept polygons under translation

A translational sweep is the translating of a polygon, called the generatrix G, around another polygon, called the directrix D, under two conditions: (1) G is always in contact with D; and (2) the interiors of G and D do not intersect. Three classes of translational sweep are studied, including the case in which both G and D are convex; the case in which G is convex, D monotone; and the case in which both are monotone. Efficient algorithms for computing the trajectory and the swept area as well as geometric and computational properties are presented for each class. A notion called the inverse generatrix, which reveals a duality between the trajectory and the swept polygon, is introduced to reduce complexity.     

Weathering wikis: Net-based learning meets political science in a South African university

Wikis represent flexible tools functioning as open-ended environments for collaboration while also offering process and group writing support. Here we focus on a project to innovate the use of wikis for collaborative writing within student groups in a final-year undergraduate political science course. The primary questions guiding our research were in what ways could wikis assist collaborative learning in an undergraduate course in political science and how we could support educators’ in the effective use of wikis? Curiously, wikis may serve as a mediating artifact for collaborative writing even among students who are reluctant to post online drafts. The paper raises questions concerning the nature and limits of lecturer and tutor power to deliver transformative educational innovations in relation to the capacity of students to embrace, comply with, or resist such innovation. In analysing the negotiation of the use of wikis in the course by and among the lecturer, tutors, and students, we draw on two principles in activity theory, which Yrjö Engeström argued are central to his model of expansive learning: multi-voicedness and contradictions [Engeström, Yrjö. (1987). Learning by expanding: An activity-theoretical approach to developmental research. Helsinki: Orienta-Konsultit; Engeström, Yrjö. (2001). Expansive learning at work: Toward an activity theoretical reconceptualization. Journal of Education and Work14(1), 133-156.]. We add a third principle, transparency, to more fully capture what we observed.     

A hybrid model for a multiproduct pipeline planning and scheduling problem

Brazilian petrobras is one of the world largest oil companies. Recurrently, it faces a very difficult planning and scheduling problem: how to operate a large pipeline network in order to adequately transport oil derivatives and biofuels from refineries to local markets. In spite of being more economical and environmentally safer, the use of a complex pipeline network poses serious operational difficulties related to resource allocation and temporal constraints. The current approaches known from the literature only consider a few types of constraints and restricted topologies, hence they are far from being applicable to real instances from petrobras. We propose a hybrid framework based on a two-phase problem decomposition strategy. A novel Constraint Programming (CP) model plays a key role in modelling operational constraints that are usually overlooked in literature, but that are essential in order to guarantee viable solutions. The full strategy was implemented and produced very adequate results when tested over large real instances.     

Error Analysis for Image Inpainting

Image inpainting refers to restoring a damaged image with missing information. In recent years, there have been many developments on computational approaches to image inpainting problem [2, 4, 6, 9, 11–13, 27, 28]. While there are many effective algorithms available, there is still a lack of theoretical understanding on under what conditions these algorithms work well. In this paper, we take a step in this direction. We investigate an error bound for inpainting methods, by considering different image spaces such as smooth images, piecewise constant images and a particular kind of piecewise continuous images. Numerical results are presented to validate the theoretical error bounds. Tony F. Chan received the B.S. degree in engineering and the M.S. degree in aerospace engineering in 1973, from the California Institute of Technology, and the Ph.D. degree in computer science from Stanford University in 1978. He is Professor of Mathematics and currently also Dean of the division of Physical science at University of California, Los Angeles, where he has been a Professor since 1986. His research interests include mathematical and computational methods in image processing, multigrid, domain decomposition algorithms, iterative methods, Krylov subspace methods, and parallel algorithms. Sung Ha Kang received the Ph.D. degree in mathematics in 2002, from University of California, Los Angeles, and currently is Assistant Professor of Mathematics at University of Kentucky since 2002. Her research interests include mathematical and computational methods in image processing and computer vision.