Scientific Data Management in a Grid Environment

Managing scientific data is by no means a trivial task even in a single site environment with a small number of researchers involved. We discuss some issues concerned with posing well-specified experiments in terms of parameters or instrument settings and the metadata framework that arises from doing so. We are particularly interested in parallel computer simulation experiments, where very large quantities of warehouse-able data are involved, run in a multi-site Grid environment. We consider SQL databases and other framework technologies for manipulating experimental data. Our framework manages the outputs from parallel runs that arise from large cross-products of parameter combinations. Considerable useful experiment planning and analysis can be done with the sparse metadata without fully expanding the parameter cross-products. Extra value can be obtained from simulation output that can subsequently be data-mined. We have particular interests in running large scale Monte Carlo physics model simulations. Finding ourselves overwhelmed by the problems of managing data and compute resources, we have built a prototype tool using Java and MySQL that addresses these issues. We use this example to discuss type-space management and other fundamental ideas for implementing a laboratory information management system.     

Parallel adaptive wavefront algorithms solving Lyapunov equations for the Cholesky factor on message passing multiprocessors

The order of the matrices involved in several algebraic problems decreases during the solution process. In these cases, parallel algorithms which use adaptive solving block sizes offer better performance results than the ones obtained on parallel algorithms using traditional constant block sizes. Recently, new parallel wavefront algorithms solving the Lyapunov equations for the Cholesky factor using Hammarling's method on message passing multiprocessors systems have been designed. In this paper, new parallel adaptive versions of these parallel algorithms are described, and experimental results obtained on an SGI Power Challenge and a SUN UltraSparc cluster are presented. Copyright © 1999 John Wiley & Sons, Ltd.     

Fault diagnosis for a class of rearrangeable networks

Two methods for fault diagnosis on a class of rearrangeable networks using the same fault model for blocking multistage networks are developed. The first method uses minimal two-bit test vectors and uniform switch settings in two phases to detect and locate a fault. In most cases, the number of tests needed is independent of the network size. The second method uses (n + 1)-bit test vectors and nonuniform switch settings in two test phases. Speedup of the process is obtained in many cases. Again, the number of tests needed is independent of the network size in most cases. A combination of these two methods is proposed to achieve greater efficiency.     

Tuning PI/PID controllers for integrating processes with deadtime and inverse response by simple calculations

A simple calculation method of a practical PI/PID controller tuning for integrating processes with deadtime and inverse response based on a model is presented in this study. First, analytical expressions for PI/PID controller settings based on the model using a direct synthesis method for disturbance rejection (DS-d) are employed. Next, optimum tuning parameter (λ) for DS-d based on the model and minimum IAE criterion are obtained via the golden-section searching technique. These optimum λ data are then empirically correlated into two equations. Thus, PI/PID controller settings for the model can easily be obtained from the parameter λ using DS-d formulas. The advantage of the proposed method is that DS-d PI/PID settings could be expediently sought by simple calculations using these equations without any tedious design. Simulation results have demonstrated that the proposed tuning technique can perform better for load/disturbance changes than other available methods in the literature.     

A ‘benign addition’? Research on ICT and pre‐school children

Abstract This paper reviews the international research evidence on the ways in which information and communication technologies (ICT) are used in both formal and informal pre-school settings. The review addresses the debate over the value and desirability of young children using computers and computational toys; the relationship of these technologies to a media environment which encompasses television, video, books and magazines; the literacies involved in using these media; and interface design and interactivity.     

A multilevel information fusion approach for visual quality inspection

For visual quality inspection systems to be applicable in industrial settings, it is mandatory that they are highly flexible, robust and accurate. In order to improve these characteristics a multilevel information fusion approach is presented. A first fusion step at the feature-level enables the system to learn from an undefined number of potential defects which might be segmented from the images. This allows for the quality control operators to label the data at the image-level and the sub-image-level, and use this information during the learning process. Additionally, the operators are allowed to provide a confidence measure for their labelling. The additional information obtained from the increased flexibility of the operator inputs allows to build more accurate classifiers. A second fusion step at the decision-level combines the classifications of different classifiers, making the system more accurate and more robust with respect to the classification method chosen. The experimental results, using various artificial and real-world visual quality inspection data sets, show that each of these fusion approaches can significantly improve the classification accuracy. If both information fusion approaches are combined the accuracy increases even further, significantly outperforming each of the fusion approaches on their own.     

Regret bounds for sleeping experts and bandits

We study on-line decision problems where the set of actions that are available to the decision algorithm varies over time. With a few notable exceptions, such problems remained largely unaddressed in the literature, despite their applicability to a large number of practical problems. Departing from previous work on this “Sleeping Experts” problem, we compare algorithms against the payoff obtained by the best ordering of the actions, which is a natural benchmark for this type of problem. We study both the full-information (best expert) and partial-information (multi-armed bandit) settings and consider both stochastic and adversarial rewards models. For all settings we give algorithms achieving (almost) information-theoretically optimal regret bounds (up to a constant or a sub-logarithmic factor) with respect to the best-ordering benchmark.     

Coreference resolution: an empirical study based on SemEval-2010 shared Task 1

This paper presents an empirical evaluation of coreference resolution that covers several interrelated dimensions. The main goal is to complete the comparative analysis from the SemEval-2010 task on Coreference Resolution in Multiple Languages. To do so, the study restricts the number of languages and systems involved, but extends and deepens the analysis of the system outputs, including a more qualitative discussion. The paper compares three automatic coreference resolution systems for three languages (English, Catalan and Spanish) in four evaluation settings, and using four evaluation measures. Given that our main goal is not to provide a comparison between resolution algorithms, these are merely used as tools to shed light on the different conditions under which coreference resolution is evaluated. Although the dimensions are strongly interdependent, making it very difficult to extract general principles, the study reveals a series of interesting issues in relation to coreference resolution: the portability of systems across languages, the influence of the type and quality of input annotations, and the behavior of the scoring measures.     

Handling redundant processing in OBDA query execution over relational sources

Redundant processing is a key problem in the translation of initial queries posed over an ontology into SQL queries, through mappings, as it is performed by ontology-based data access systems. Examples of such processing are duplicate answers obtained during query evaluation, which must finally be discarded, or common expressions evaluated multiple times from different parts of the same complex query. Many optimizations that aim to minimize this problem have been proposed and implemented, mostly based on semantic query optimization techniques, by exploiting ontological axioms and constraints defined in the database schema. However, data operations that introduce redundant processing are still generated in many practical settings, and this is a factor that impacts query execution. In this work we propose a cost-based method for query translation, which starts from an initial result and uses information about redundant processing in order to come up with an equivalent, more efficient translation. The method operates in a number of steps, by relying on certain heuristics indicating that we obtain a more efficient query in each step. Through experimental evaluation using the Ontop system for ontology-based data access, we exhibit the benefits of our method.     

Scheduling a single vehicle in the just-in-time part supply for a mixed-model assembly line

This paper focuses on the scheduling of a single vehicle, which delivers parts from a storage centre to workstations in a mixed-model assembly line. In order to avoid part shortage and to cut down total inventory holding and travelling costs, the destination workstation, the part quantity and the departure time of each delivery have to be specified properly according to predetermined assembly sequences. In this paper, an optimisation model is established for the configuration that only one destination workstation is involved within each delivery. Four specific properties of the problem are deduced, then a backward-backtracking approach and a hybrid GASA (genetic algorithm and simulated annealing) approach are developed based on these properties. Both two approaches are applied to several groups of instances with real-world data, and results show that the GASA approach is efficient even in large instances. Furthermore, the existence of feasible solutions (EOFS) is analysed via instances with different problem settings, which are obtained by an orthodox experimental design (ODE). An analysis of variance (ANOVA) shows that the buffer capacity is the most significant factor influencing the EOFS. Besides this, both the assembly sequence length and distances to workstations also have noticeable impacts.     

Calculating fuzzy inverse matrix using fuzzy linear equation system

Linear equation systems play a very important role in engineering, mathematics, statistics and other disciplines. Fuzzifying either parameters or variables or both in these systems has been one of the research areas in the fuzzy literature since these kinds of systems are encountered in many applications. These systems are generally called fuzzy linear equations. Various types of these models have been examined for a decade. The solution procedures of these systems depend on different methods such as extension principle and interval arithmetic. Also, the method which is often used in computing inverse of a matrix in real case could be extended to fuzzy case, which employs linear equation system and identity matrix. For this purpose, we propose a new method which includes some new definitions which are fuzzy zero number, fuzzy one number and fuzzy identity matrix. Based on these definitions, direct computation of fuzzy inverse matrix is done using fuzzy arithmetic and fuzzy equation system. Actually, this simply extends the notion used in real case to fuzzy case. Calculation is realized with two different settings. While the first one is called direct numerical solution, the other is obtained by choice of decision maker. It is noted that the uniqueness of the calculated fuzzy inverse matrix is not guaranteed.     

VCKSM: Verifiable conjunctive keyword search over mobile e-health cloud in shared multi-owner settings

Searchable encryption (SE) is a promising technique which enables cloud users to conduct search over encrypted cloud data in a privacy-preserving way, especially for the electronic health record (EHR) system that contains plenty of medical history, diagnosis, radiology images, etc. In this paper, we focus on a more practical scenario, also named as the shared multi-owner settings, where each e-health record is co-owned by a fixed number of parties. Although the existing SE schemes under the unshared multi-owner settings can be adapted to this shared scenario, these schemes have to build multiple indexes, which definitely incur higher computational overhead. To save bandwidth and computing resources in cloud servers and guarantee the correctness of search results, we present a secure cryptographic primitive, namely verifiable conjunctive keyword search over mobile e-health cloud scheme, in the shared multi-owner settings by utilizing multisignatures technique. Formal security analysis proves that our scheme is secure against the keyword guessing attacks in standard model. Empirical study using a real-world dataset justifies that our scheme is efficient and feasible in practical applications.     

Learning through online peer discourse: Structural equation modeling points to the role of discourse activities in individual understanding

Although learning through discourse activities seems well-documented, it is unclear which mechanisms and behavioral variables are involved. What exactly contributes to learning when two or more learners interact in online learning environments? To analyze interrelations between central discourse activities and individual learning outcomes at the level of constructs, we applied structural equation modeling to data collected from 160 dyads engaging in written online learning discourses within a series of homogeneous experiments. We analyzed three theory-based indicators of conceptual elaboration activities during online discourse: the number of questions asked to receive information and expand knowledge, the number of explanations formulated to express individual knowledge, and the amount of on-task discourse. Individual conceptual understanding was represented by objective learning parameters that varied in each particular experimental task. These measured general understanding of the topic addressed and particular understanding of conceptual terms, complemented by the gain in individual self-assessed knowledge. Results of structural equation modeling revealed a strong effect of dyadic conceptual elaboration on individual understanding at the construct level, demonstrating that dyadic elaboration fosters the development of an elaborated individual understanding of specialist concepts and general content knowledge. Moreover, conceptual elaboration was best measured by the number of explanations during discourse. Implications regarding which features of collaborative learning settings promote mutual conceptual elaboration are discussed.     

基于遗传算法和极限学习机的Fugl-Meyer量表自动评估

为实现脑卒中上肢居家康复评定的自动化和定量化,针对临床上最常用的Fugl-Meyer运动功能评定（FMA）量表,利用极限学习机（ELM）建立了FMA量表得分自动预测模型。选取FMA肩肘部分中的4个动作,采用固定于偏瘫侧前臂和上臂的两个加速度传感器采集24名患者的运动数据,经预处理和特征提取,基于遗传算法（GA）和ELM进行特征选择,分别建立单个动作ELM预测模型和综合预测模型。结果显示,该模型可对FMA肩肘部分得分进行精确的自动预测,预测均方根误差为2.1849分。该方法突破了传统评定中主观性、耗时性的限制及对康复医师或治疗师的依赖性,可方便用于居家康复的评定。速度传感器采集24名患者的运动数据,经预处理和特征提取,基于遗传算法（Genetic Algorithm, GA）和ELM进行特征选择,分别建立单个动作ELM预测模型和综合预测模型。结果显示,该模型可对FMA肩肘部分得分进行精确的自动预测,预测均方根误差为2.1849分。该方法突破了传统评定中主观性、耗时性的限制及对康复医师或治疗师的依赖性,可方便用于居家康复的评定。     

细节决定成败：推荐系统实验反思与讨论

近些年来,随着互联网的迅速发展,用户在各种在线平台上接收到海量的信息,信息爆炸成为一个关键性问题。在此背景下,推荐系统逐步渗透到人们工作生活的各个场景,已成为不可或缺的一环。它不仅可以帮助用户快速获得想要的信息和服务,还可以提高资源利用效率,从而给企业带来更多效益。因此,个性化推荐算法不仅获得了工业界广泛的关注,也是科研领域的研究热点之一。在个性化推荐的研究中,受限于平台与效率等因素,研究者大多无法将算法部署到在线系统上进行评价,因此离线评价成为推荐领域研究的主要方式。然而个性化推荐涉及到的场景复杂,可获得的数据信息多种多样,用户行为多为隐式反馈且存在许多噪声,这使得推荐系统离线评价的实验设定复杂多变,存在大量易被忽视却十分重要的细节。比如在训练采样负例时,既可以仅从用户没有交互过的商品中采样,也可以将验证测试集的商品视作未知交互加入采样池。同样,从训练到测试在很多其他环节也涉及这样的实现细节（如数据集处理、已知负样本的使用、Top-N排序候选集范围等）。这些实验细节通常不会在学术论文中被显式提及,却潜在影响了模型效果的对比,还决定着实验的科学性,甚至会导致相反或错误的分析结论。本文从数据集处理、模型训练、验证与测试、效果评价等多个角度,系统地讨论与反思了推荐系统实验中的细节设定。对于每个环节,我们枚举了若干常见设定,并在真实数据集上验证了其中某些设定的实际影响。实验结果表明一些细节确实会导致关于模型优劣的不同结论。最终我们形成了关于推荐系统实验细节的指导性总结,包括可选、建议、必须的三类设定,希望帮助推荐算法研究者规避实现细节上的陷阱,更科学合理地设计实验。     

Context-aware service engineering: A survey

Context constitutes an essential part of service behaviour, especially when interaction with end-users is involved. As observed from the literature, context handling in service engineering has been during recent years a field of intense research, which has produced several interesting approaches. In this paper, we present research efforts that attempt mainly to decouple context handling from the service logic. We enumerate all context management categories, but focus on the most appropriate for service engineering, namely source code level, model-driven and message interception, taking also into account the fact that these have not been dealt with in detail in other surveys. A representative example is used to illustrate more precisely how these approaches can be used. Finally, all three categories are compared based on a number of criteria.     

A study of overheads and accuracy for efficient monitoring of wireless mesh networks

IEEE 802.11-based wireless mesh networks are being increasingly deployed in enterprize and municipal settings. A lot of work has been done on developing measurement-based schemes for resource provisioning and fault management in these networks. The above goals require an efficient monitoring infrastructure to be deployed, which can provide the maximum amount of information regarding the network status, while utilizing the least possible amount of network resources. However, network monitoring involves overheads, which can adversely impact performance from the perspective of the end user. The impact of monitoring overheads on data traffic has been overlooked in most of the previous works. It remains unclear as to how parameters such as number of monitoring agents, or frequency of reporting monitoring data, among others, impact the performance of a wireless network. In this work, we first evaluate the impact of monitoring overheads on data traffic, and show that even small amounts of overhead can cause a large degradation in the network performance. We then explore several different techniques for reducing monitoring overheads, while maintaining the objective (resource provisioning, fault management, and others) that needs to be achieved. Via extensive simulations and experiments, we validate the efficiency of our proposed approaches in reducing overheads, their impact on the quality of data collected from the network, and the impact they have on the performance of the applications using the collected data. Based on results, we conclude that it is feasible to make the current monitoring techniques more efficient by reducing the communication overheads involved while still achieving the desired application-layer objectives.