Search

This paper explores website link structure considering websites as interconnected graphs and analyzing their features as a social network. Two networks have been extracted for representing websites: a domain network containing subdomains or external domains linked through the website and a page network containing webpages browsed from the root domain. Factor analysis provides the statistical methodology to adequately extract the main website profiles in terms of their internal structure. However, due to the large number of indicators, the task of selecting a representative subset of indicators becomes unaffordable. A genetic search of an optimum subset of indicators is proposed in this paper, selecting a multiobjective fitness function based on factor analysis results. The optimum solution provides a coherent and relevant categorization of website profiles, and highlights the possibilities of genetic algorithms as a tool for discovering new knowledge in the field of web mining Ministerio de Educación y Ciencia DPI2007- 60128 Junta de Andalucía. Consejería de Innovación, Ciencia y Empresa P07-TIC-02621

Machine tool owners require knowledge of their machine’s capabilities, and the emphasis increases with\ud areas of high accuracy manufacturing. An aspect of a machine’s capability is its geometric accuracy. International\ud Standards and best-practice guides are available to aid understanding of the required measurements\ud and to advise on how to perform them. However, there is an absence of any intelligent method\ud capable of optimising the duration of a calibration plan, minimising machine down-time. In this work,\ud artificial intelligence in the form of automated planning is applied to the problem of machine tool\ud pseudo-static geometric error calibration. No prior knowledge of Artificial Intelligence (AI) planning is\ud required throughout this paper. The authors have written this paper for calibration engineers to see\ud the benefits that automated planning can provide. Two models are proposed; the first produces a sequential\ud calibration plan capable of finding the optimal calibration plan. The second model has the additional\ud possibility of planning for concurrent measurements, adding the possibility of further reducing machine\ud down-time. Both models take input regarding a machine’s configuration and available instrumentation.\ud The efficacy of both models is evaluated by performing a case study of a five-axis gantry machine,\ud whereby calibration plans are produced and compared against both an academic and industrial expert.\ud From this, the effectiveness of this novel method for producing optimal calibration plan is evaluated,\ud stimulating potential for future work.

Work related low back disorders (LBDs) due to manual lifting tasks (MLTs) have long been recognized as one of the main occupational disabling injury that affects the quality of life of the industrial working population in the U.S. There have been a number of intensive research efforts devoted to understanding the phenomena of LBDs and building classification models that could effectively distinguish between high risk and low risk MLTs that contribute to LBDs. As of today, however, such models and the occupational exposure limits of different risk factors causing LBDs as well as the guidelines preventing them have not yet been fully proposed. One of the first efforts to comprehend the nature and phenomenon of LBDs was undertaken by Marras et al. (1993). They created a seminal data set and used it to build logistic regression (LR) models to identify significant variables and classify MLTs into high risk and low risk with respect to LBDs. Since then a number of studies have used the same data set to build and test various classifiers to detect the likelihood of LBDs due to manual material handling jobs. This paper summarizes and critiques the previous studies. It also employs this data set to build and test seven classification models, two of which have not been applied in this context yet. The parameters of the models have been calibrated for the best performance, and the models were constructed and validated on the full set and the reduced set of features. Though the performances of our best models are better than those reported in National Institute for Occupational Health and Safety (NIOHS) Guides and two of our previous studies, they are generally less optimistic than those reported in several other studies; this paper proposes a systematic and more reliable approach to creating and validating classifiers to distinguish between low and high risk MLTs that contribute to LBDs.

Sensors are vital components for control and advanced health management techniques. However, sensors continue to be considered the weak link in many engineering applications since often they are less reliable than the system they are observing. This is in part due to the sensors' operating principles and their susceptibility to interference from the environment. Detecting and mitigating sensor failure modes are becoming increasingly important in more complex and safety-critical applications. This paper reports on different techniques for sensor fault detection, disambiguation, and mitigation. It presents an expert system that uses a combination of object-oriented modeling, rules, and semantic networks to deal with the most common sensor faults, such as bias, drift, scaling, and dropout, as well as system faults. The paper also describes a sensor correction module that is based on fault parameters extraction (for bias, drift, and scaling fault modes) as well as utilizing partial redundancy for dropout sensor fault modes). The knowledge-based system was derived from the results obtained in a previously deployed Neural Network (NN) application for fault detection and disambiguation. Results are illustrated on an electro-mechanical actuator application where the system faults are jam and spalling. In addition to the functions implemented in the previous work, system fault detection under sensor failure was also modeled. The paper includes a sensitivity analysis that compares the results previously obtained with the NN. It concludes with a discussion of similarities and differences between the two approaches and how the knowledge based system provides additional functionality compared to the NN implementation.

In order to determine how data mining techniques (DMT) and their applications have developed, during the past decade, this paper reviews data mining techniques and their applications and development, through a survey of literature and the classification of articles, from 2000 to 2011. Keyword indices and article abstracts were used to identify 216 articles concerning DMT applications, from 159 academic journals (retrieved from five online databases), this paper surveys and classifies DMT, with respect to the following three areas: knowledge types, analysis types, and architecture types, together with their applications in different research and practical domains. A discussion deals with the direction of any future developments in DMT methodologies and applications: (1) DMT is finding increasing applications in expertise orientation and the development of applications for DMT is a problem-oriented domain. (2) It is suggested that different social science methodologies, such as psychology, cognitive science and human behavior might implement DMT, as an alternative to the methodologies already on offer. (3) The ability to continually change and acquire new understanding is a driving force for the application of DMT and this will allow many new future applications.

Association rule mining and classification are important tasks in data mining. Using association rules has proved to be a good approach for classification. In this paper, we propose an accurate classifier based on class association rules (CARs), called CAR-IC, which introduces a new pruning strategy for mining CARs, which allows building specific rules with high confidence. Moreover, we propose and prove three propositions that support the use of a confidence threshold for computing rules that avoids ambiguity at the classification stage. This paper also presents a new way for ordering the set of CARs based on rule size and confidence. Finally, we define a new coverage strategy, which reduces the number of non-covered unseen-transactions during the classification stage. Results over several datasets show that CAR-IC beats the best classifiers based on CARs reported in the literature.

Highlights? ePortfolio can assist in diagnosing students' programming learning styles. ? Cat intelligence is a good viewpoint for building multi-objective optimization. ? The CatOpt algorithm proposed is effective for fuzzy knowledge fusion. ? The CatOpt algorithm proposed outperformed the compared heuristic optimizations. ? The heuristic knowledge fusion improved the validity and stability of diagnosis. Computer programming is a high-level thinking activity. In the educational area, using learning styles to understand how students learn is a significant issue. The electronic Portfolio (ePortfolio) is a popular educational management and assessment tool. Unfortunately, few researchers investigate programming learning style diagnosis. This paper addresses this gap in research: this study constructs an ePortfolio-based programming learning style diagnosis to detect students' styles. The fusion of multiple fuzzy-based diagnosis knowledge is the main contribution of this work. This paper built a heuristic optimization method to integrate multiple diagnosis knowledge bases. Performance evaluations and empirical studies were implemented to verify the proposed algorithm and fusion solution. Experimental results showed that the proposed heuristic optimization firms the validity and stability of a diagnostic system, and the ePortfolio-based programming learning style diagnosis is highly accepted by students. Furthermore, teachers agreed that the knowledge fusion mechanism and diagnosis system were usable.

Highlights? We propose a goal-regulation mechanism that applies a reinforcement learning approach to autonomous goal-formation. ? Individual goals are regulated by a neural network-based fuzzy inference system called goal-regulation network (GRN). ? The GRN is updated by a reinforcement signal from another neural network called goal-evaluation network (GEN). ? The GEN approximates the compatibility of goals with current environmental situation. ? The proposed mechanism is validated by a simulation study on a production planning problem. Up-to-date market dynamics has been forcing manufacturing systems to adapt quickly and continuously to the ever-changing environment. Self-evolution of manufacturing systems means a continuous process of adapting to the environment on the basis of autonomous goal-formation and goal-oriented dynamic organization. This paper proposes a goal-regulation mechanism that applies a reinforcement learning approach, which is a principal working mechanism for autonomous goal-formation. Individual goals are regulated by a neural network-based fuzzy inference system, namely, a goal-regulation network (GRN) updated by a reinforcement signal from another neural network called goal-evaluation network (GEN). The GEN approximates the compatibility of goals with current environmental situation. In this paper, a production planning problem is also examined by a simulation study in order to validate the proposed goal regulation mechanism.

Highlights? A hybrid grouping-encoding algorithm for clustering problems is presented. ? Details on the encoding, operators and parallelization are given. ? Results in synthetic and real clustering problems are provided. In this paper we present a novel grouping genetic algorithm for clustering problems. Though there have been different approaches that have analyzed the performance of several genetic and evolutionary algorithms in clustering, the grouping-based approach has not been, to our knowledge, tested in this problem yet. In this paper we fully describe the grouping genetic algorithm for clustering, starting with the proposed encoding, different modifications of crossover and mutation operators, and also the description of a local search and an island model included in the algorithm, to improve the algorithm's performance in the problem. We test the proposed grouping genetic algorithm in several experiments in synthetic and real data from public repositories, and compare its results with that of classical clustering approaches, such as K-means and DBSCAN algorithms, obtaining excellent results that confirm the goodness of the proposed grouping-based methodology.

Highlights? The paper considers user support for selecting optimal logistic regression models. ? A number of heuristics to arrive at optimal logistic regression models are reviewed and formalized. ? Two computational algorithms that incorporate these formalized heuristics are devised. ? An implementation of these algorithms in an interactive decision support system is presented. This study concerns itself with providing user support for a decision problem in logistic regression analysis: given a set of metric variables and one binary dependent variable, select the optimal subset of variables that can best predict this dependent variable. The problem requires an evaluation of competing models based on heuristic selection criteria such as goodness-of-fit and prediction accuracy. This paper documents the heuristics, formalizes the algorithms, and eventually presents an interactive decision support system to facilitate the selection of such an optimal model.This study adds to the sparsely studied domain of expert systems for social science researchers, and makes three contributions to the literature. First, the study formalizes a number of heuristics to arrive at optimal logistic regression models. Second, the study presents two computational algorithms that incorporate these formalized heuristics. Third, the paper documents an implementation of these algorithms through an interactive decision support system. The study concludes with a discussion on the risks of relying too heavily on the system and with future opportunities for research.

Highlights? Probably the first report of experiments attempting to automatically predict verbal intelligence. ? This work has shown that verbal intelligence may be recognized by computers through language cues. ? Some of the most popular text categorization methods have been successfully applied to this task. ? Class-based feature selection approach enables better performance and lower computational cost. ? Word lemmatisation, typically successful for TC tasks, has not proven to be helpful for this one. In this paper we investigate whether conventional text categorization methods may suffice to infer different verbal intelligence levels. This research goal relies on the hypothesis that the vocabulary that speakers make use of reflects their verbal intelligence levels. Automatic verbal intelligence estimation of users in a spoken language dialog system may be useful when defining an optimal dialog strategy by improving its adaptation capabilities. The work is based on a corpus containing descriptions (i.e. monologs) of a short film by test persons yielding different educational backgrounds and the verbal intelligence scores of the speakers. First, a one-way analysis of variance was performed to compare the monologs with the film transcription and to demonstrate that there are differences in the vocabulary used by the test persons yielding different verbal intelligence levels. Then, for the classification task, the monologs were represented as feature vectors using the classical TF-IDF weighting scheme. The Naive Bayes, k-nearest neighbors and Rocchio classifiers were tested. In this paper we describe and compare these classification approaches, define the optimal classification parameters and discuss the classification results obtained.

Highlights? Within IT, grids and data centres represent the hungriest consumers of energy. ? Energy saving involves two direct benefits: sustainability and cost reduction. ? We model seven energy policies for Grid computing environments. ? This paper proves proved that 40policy is applied. ? Simulation software is available and results for real scenarios can be obtained. Energy saving involves two direct benefits: sustainability and cost reduction. Within the field of Information Technology, clusters, grids and data centres represent the hungriest consumers of energy and therefore energy (saving) policies for these infrastructures should be applied in order to maximize their resources. It is proved in this paper that approximately 40% of energy can be saved in a data centre if an adequate policy is applied. Furthermore, a software tool is presented where simulations can be run and results for real scenarios can be obtained.

Bibliographic information networks, formed by online bibliographic databases, such as ACM Digital Library and IEEE/IET Electronic Library, contain abundant information about authors, papers, venues (journals/conferences), and have been widely studies in recent years. However, few studies examine the problem of ranking objects in these networks. In this paper, we study this problem and present a novel model, called PAV, for ranking heterogeneous objects, such as authors, papers, and venues. Based on PAV model, we transform the problem of ranking objects into the problem of estimating probability distribution. We propose an efficient algorithm to estimate probability parameters by use of the fact that the PAV model is a regular Markov chain. For evaluating PAV model, we apply it on one real dataset, which was crawled from ACM Digital Library. The experimental results show that the proposed model is effective.

A 2010 survey (Nielsen) showed that 22.7% of the time spent on the Internet is on a social network. Moreover, there is an increasing demand to access social networks by mobile phones, i.e., around 30% globally. Social networking has become a reality, and it generates an incredible amount of information that is sometimes difficult for users to process, especially from mobile phones. Several links, activities, and recommendations are proposed by networked friends every hour, which together are nearly impossible to manage. There is a need to filter and make accessible such information to users, which is the motivation behind developing a mobile recommender that exploits social network information. Thus, in this paper, we propose the design and the implementation of a SOcial Mobile Activity Recommender (SOMAR) that can integrate Facebook social network mobile data and sensor data to propose activities to the user (e.g., concert or computer science seminar). The recommendations are completely calculated in situ in the mobile device with an embedded data mining component. We analyze how to compute and update the social graph in case of changes in social relationships or user context. The paper also presents a case study to analyze the performance of the method.

Using a bibliometric approach, this paper analyzes research trends and forecasts of data mining from 1989 to 2009 by locating heading ''data mining'' in topic in the SSCI database. The bibliometric analytical technique was used to examine the topic in SSCI journals from 1989 to 2009, we found 1181 articles with data mining. This paper implemented and classified data mining articles using the following eight categories-publication year, citation, country/territory, document type, institute name, language, source title and subject area-for different distribution status in order to explore the differences and how data mining technologies have developed in this period and to analyze technology tendencies and forecasts of data mining under the above results. Also, the paper performs the K-S test to check whether the analysis follows Lotka's law. Besides, the analysis also reviews the historical literatures to come out technology diffusions of data mining. The paper provides a roadmap for future research, abstracts technology trends and forecasts, and facilitates knowledge accumulation so that data mining researchers can save some time since core knowledge will be concentrated in core categories. This implies that the phenomenon ''success breeds success'' is more common in higher quality publications.

Generally, stock trading expert systems (STES) called also ''mechanical trading systems'' are based on the technical analysis, i.e., on methods for evaluating securities by analyzing statistics generated by the market activity, such as past prices and volumes (number of transactions during a unit of a timeframe). In other words, such STES are based on the Level 1 information. Nevertheless, currently the Level 2 information is available for the most of traders and can be successfully used to develop trading strategies especially for the day trading when a significant amount of transactions are made during one trading session. The Level 2 tools show in-depth information on a particular stock. Traders can see not only the ''best'' bid (buying) and ask (selling) orders, but the whole spectrum of buy and sell orders at different volumes and different prices. In this paper, we propose some new technical analysis indices bases on the Level 2 and Level 1 information which are used to develop a stock trading expert system. For this purpose we adapt a new method for the rule-base evidential reasoning which was presented and used in our recent paper for building the stock trading expert system based the Level 1 information. The advantages of the proposed approach are demonstrated using the developed expert system optimized and tested on the real data from the Warsaw Stock Exchange.

This paper has two main purposes. The first one is the development of a rigorous rule-based mechanism for identifying the rounding bottoms (also known as saucers) pattern and resistant levels. The design of this model is based solely on principles of technical analysis, and thus making it a proper system for evaluating the efficacy of the aforementioned technical trading patterns. The second aim of this paper is measuring the predictive power of buy-signals generated by these technical patterns. Empirical results obtained from seven US tech stocks indicate that simple resistant levels outperform saucers patterns. Furthermore, positive statistical significant excess returns are being generated only in first sub-periods of examination. These returns decline or even vanish as the experiment proceeds to recent years. Our findings are aligned with the results reported by various former studies. The proposed identification mechanism can be used as a component of an expert system to assist academic community in evaluating trading strategies where technical patterns are embedded.

An adaptation phase is crucial for a good and reasonable case-based design (CBD) process, which is responsible for finding a solution to solve a new problem in the principle of k-nearest neighbors (k-NN). Statistical adaptation method is a classical method for feature-based case adaptation (FCA) because of its domain-independent and easily to be implemented, but with low adaptation accuracy. Therefore, this paper presents a new adaptation method for solution feature values of retrieved cases by introducing the adaptability value to improve the adaptation performance, called as adaptability-based FCA (AFCA). Unlike the classical statistical FCA method (SFCA) based on similarity or distance value, AFCA is performed in terms of the adaptability of old solution feature calculated by the adaptability measurement (AM) mechanism. A new AM method is studied as well in this paper, where the adaptability value for each solution feature is computed by utilizing the decision tree technique and similarity value, and the similarity is derived from the multi-algorithm-oriented hybrid SM strategy. Furthermore, to validate the feasibility and superiority of AFCA, the proposed method was applied to the power transformer design and was compared with the classical SFCAs. Empirical comparison results indicated that AFCA achieves the better adaptation performance under k-NN than other SFCAs on the basis of the adaptation accuracy.

Public health monitoring for solitary elderly people must be implemented in a particularly economic way because of their low-income status. The aim of this paper is to propose a model-based method combined with conventional reasoning methods such as multiple regression and the boosting strategy. The role of model-based reasoning is to generate secondary situational information from activity data gathered at home. Current health condition information is then provided as part of an activity-based smart management system for health monitoring. Only one activity sensor per house is considered. In this paper, we also discuss how the Korean government has actually applied this method in smart-care services for more than ten thousand solitary elderly people. The experiments are conducted based on the u-care system, which is composed of an activity sensor connected to a remote server using a wireless sensor network. The remote server includes multi-agents which analyze and diagnose current health conditions.

Even though expressive ontology representation languages like OWL have been proposed in the last years, most available ontologies only focus on taxonomical and, in a few cases, non-taxonomical knowledge. A way to add expressivity to an ontology is to include axioms that describe the basic logical properties of the modeled relationships. However, due to the manual knowledge acquisition bottleneck, axioms are usually missing in handcrafted ontologies. Moreover, automatic or semi-automatic ontology learning approaches aimed to acquire axioms from textual resources to enrich ontologies are scarce. This paper presents a novel methodology to learn axioms associated to non-taxonomic relationships modeled in an ontology in an automatic and unsupervised way, using the Web as corpus of textual resources. The proposed method is based on the use of specially tailored linguistic patterns, the exploitation of Web search engines as massive information retrieval tools, the application of shallow natural language processing and the assessment of semantic relatedness by means of Web-scale co-occurrence statistics. The paper describes the learning algorithm and presents promising results on the assessment of the following axioms: functional, inverse, symmetrical, transitive, reflexive and inverse functional properties.

The automatic generation of individualized plans in specific domains is an open problem that combines aspects related to automated planning, machine learning or recommendation systems technology. In this paper, we focus on a specific instance of that task; that of generating e-learning courses adapted to students' profiles, within the automated planning paradigm. One of the open problems in this type of automated planning application relates to what is known as oversubscription: given a set of goals, each one with a utility, obtain a plan that achieves some (or all) the goals, maximizing the utility, as well as minimizing the cost of achieving those goals. In the generation of e-learning designs there is only one goal: generating a course design for a given student. However, in order to achieve the goal, the course design can include many different kinds of activities, each one with a utility (that depend on the student profile) and cost. Furthermore, these activities are usually grouped into clusters, so that at least one of the activities in each cluster is needed, though many more can be used. Finally, there is also an overall cost threshold (usually in terms of student time). In this paper, we present our work on building an individualized e-learning design. We pose each course design as a variation of the oversubscription problem, that we call the clustered-oversubscription problem, and we use linear programming for assisting a planner to generate the design that better adapts to the student.

Safety assessment of thermal power plants (TPPs) is one of the important means to guarantee the safety of production in thermal power production enterprises. Due to various technical limitations, existing assessment approaches, such as analytic hierarchy process (AHP), Monte Carlo methods, artificial neural network (ANN), etc., are unable to meet the requirements of the complex security assessment of TPPs. Currently, most of the security assessments of TPP are completed by the means of experts' evaluations. Accordingly, the assessment conclusions are greatly affected by the subjectivity of the experts. Essentially, the evaluation of power plant systems relies to a large extent on the knowledge and length of experience of the experts. Therefore in this domain case-based reasoning (CBR) is introduced for the security assessment of TPPs since this methodology models expertise through experience management. Taking the management system of TPPs as breakthrough point, this paper presents a case-based approach for the Safety assessment decision support of TPPs (SATPP). First, this paper reviews commonly used approaches for TPPs security assessment and the current general evaluation process of TPPs security assessment. Then a framework for the Management System Safety Assessment of Thermal Power Plants (MSSATPP) is constructed and an intelligent decision support system for MSSATPP (IDSS-MSSATPP) is functionally designed. IDSS-MSSATPP involves several key technologies and methods such as knowledge representation and case matching. A novel case matching method named Improved Gray CBR (IGCBR) has been developed in which a statistical approach (logistic regression) and Gray System theory are integrated. Instead of applying Gray System theory directly, it has been improved to integrate it better into CBR. In addition this paper describes an experimental prototype system of IDSS-MSSATPP (CBRsys-TPP) in which IGCBR is integrated. The experimental results based on a MSSATPP data set show that CBRsys-TPP has high accuracy and systematically good performance. Further comparative studies with several other common classification approaches also show its competitive power in terms of accuracy and the synergistic effects of the integrated components.

Predicting the accurate prognosis of breast cancer from high throughput microarray data is often a challenging task. Although many statistical methods and machine learning techniques were applied to diagnose the prognosis outcome of breast cancer, they are suffered from the low prediction accuracy (usually lower than 70%). In this paper, we propose a better method (genetic algorithm-support vector machine, we called GASVM) to significant improve the prediction accuracy of breast cancer from gene expression profiles. To further improve the classification performance, we also apply GASVM model using combined clinical and microarray data. In this paper, we evaluate the performance of the GASVM model based on data provided by 97 breast cancer patients. Four kinds of gene selection methods are used: all genes (All), 70 correlation-selected genes (C70), 15 medical literature-selected genes (R15), and 50 T-test-selected genes (T50). With optimized parameter values identified from GASVM model, the average predictive accuracy of our model approaches 95% for T50 and 90% for C70 or R15 in all four kernel functions using integrated clinical and microarray data. Our model produces results more accurately than the average 70% predictive accuracy of other machine learning methods. The results indicate that the GASVM model has the potential to better assist physicians in the prognosis of breast cancer through the use of both clinical and microarray data.

One-Versus-All (OVA) classification is a classifier construction method where a k-class prediction task is decomposed into k 2-class sub-problems. One base model is constructed for each sub-problem and the base models are then combined into one model. Aggregate model implementation is the process of constructing several base models which are then combined into a single model for prediction. In essence, OVA classification is a method of aggregate modeling. This paper reports studies that were conducted to establish whether OVA classification can provide predictive performance gains when large volumes of data are available for modeling as is commonly the case in data mining. It is demonstrated in this paper that firstly, OVA modeling can be used to increase the amount of training data while at the same time using base model training sets whose size is much smaller than the total amount of available training data. Secondly, OVA models created from large datasets provide a higher level of predictive performance compared to single k-class models. Thirdly, the use of boosted OVA base models can provide higher predictive performance compared to un-boosted OVA base models. Fourthly, when the combination algorithm for base model predictions is able to resolve tied predictions, the resulting aggregate models provide a higher level of predictive performance.

Elementary and junior high school buildings in Taiwan are designed to serve not only as places of education but also as temporary shelters in the aftermath of major earthquakes. Effective evaluation of the seismic resistance of school buildings is a critical issue that deserves further investigation. The National Center for Research on Earthquake Engineering (in Taiwan) currently employs performance-target ground acceleration (A"P) as the index to evaluate school structure compliance with seismic resistance requirements. However, computational processes are complicated, time consuming, and require the input of many experts. To address this problem, this paper developed an evolutionary support vector machine inference system (ESIS) that integrated two AI techniques, namely, the support vector machine (SVM) and fast messy genetic algorithm (fmGA). Based on training results, the developed system can predict the A"P of a school building in a significantly shorter time base, thus increasing evaluation efficiency significantly. The validity of ESIS was tested using the 10-Fold Cross-Validation method. Another aim of this paper is to retain and apply expert knowledge and relevant experience to the solution of similar problems in the future.

In this paper text segmentation in generic displays is proposed through learning the best binarization values for a commercial optical character recognition (OCR) system. The commercial OCR is briefly introduced as well as the parameters that affect the binarization for improving the classification scores. The purpose of this work is to provide the capability to automatically evaluate standard textual display information, so that tasks that involve visual user verification can be performed without human intervention. The problem to be solved is to recognize text characters that appear on the display, as well as the color of the characters' foreground and background. The paper introduces how the thresholds are learnt through: (a) selecting lightness or hue component of a color input cell, (b) enhancing the bitmaps' quality, and (c) calculating the segmentation threshold range for this cell. Then, starting from the threshold ranges learnt at each display cell, the best threshold for each cell is gotten. The input and output data sets for testing the algorithms proposed are described, as well as the analysis of the results obtained.

Highlights? This paper proposed a novel Bayesian method based cognitive trust model. ? A trust dynamic level scheduling algorithm named Cloud-DLS is proposed. ? A benchmark is structured for evaluation of the proposed method. ? Experiments and case studies are carried out to evaluate our proposed method. Clouds are rapidly becoming an important platform for scientific applications. In the Cloud environment with uncountable numeric nodes, resource is inevitably unreliable, which has a great effect on task execution and scheduling. In this paper, inspired by Bayesian cognitive model and referring to the trust relationship models of sociology, we first propose a novel Bayesian method based cognitive trust model, and then we proposed a trust dynamic level scheduling algorithm named Cloud-DLS by integrating the existing DLS algorithm. Moreover, a benchmark is structured to span a range of Cloud computing characteristics for evaluation of the proposed method. Theoretical analysis and simulations prove that the Cloud-DLS algorithm can efficiently meet the requirement of Cloud computing workloads in trust, sacrificing fewer time costs, and assuring the execution of tasks in a security way.

The generation of leak along the pipeline carrying crude oils and liquid fuels results enormous financial loss to the industry and also affects the public health. Hence, the leak detection and localization problem has always been a major concern for the companies. In spite of the various techniques developed, accuracy and time involved in the prediction is still a matter of concern. In this paper, a novel leak detection scheme based on rough set theory and support vector machine (SVM) is proposed to overcome the problem of false leak detection. In this approach, 'rough set theory' is explored to reduce the length of experimental data as well as generate rules. It is embedded to enhance the decision making process. Further, SVM classifier is employed to inspect the cases that could not be detected by applied rules. For the computational training of SVM, this paper uses swarm intelligence technique: artificial bee colony (ABC) algorithm, which imitates intelligent food searching behavior of honey bees. The results of proposed leak detection scheme with ABC are compared with those obtained by using particle swarm optimization (PSO) and one of its variants, so-called enhanced particle swarm optimization (EPSO). The experimental results advocate the use of propounded method for detecting leaks with maximum accuracy.

In this paper, we set out to compare several techniques that can be used in the analysis of imbalanced credit scoring data sets. In a credit scoring context, imbalanced data sets frequently occur as the number of defaulting loans in a portfolio is usually much lower than the number of observations that do not default. As well as using traditional classification techniques such as logistic regression, neural networks and decision trees, this paper will also explore the suitability of gradient boosting, least square support vector machines and random forests for loan default prediction.Five real-world credit scoring data sets are used to build classifiers and test their performance. In our experiments, we progressively increase class imbalance in each of these data sets by randomly under-sampling the minority class of defaulters, so as to identify to what extent the predictive power of the respective techniques is adversely affected. The performance criterion chosen to measure this effect is the area under the receiver operating characteristic curve (AUC); Friedman’s statistic and Nemenyi post hoc tests are used to test for significance of AUC differences between techniques.The results from this empirical study indicate that the random forest and gradient boosting classifiers perform very well in a credit scoring context and are able to cope comparatively well with pronounced class imbalances in these data sets. We also found that, when faced with a large class imbalance, the C4.5 decision tree algorithm, quadratic discriminant analysis and k-nearest neighbours perform significantly worse than the best performing classifiers.

Highlights? We propose a multi-level-structured representation to express more semantic information of a document. ? A multi-level matching method incorporate with EMD distance solved by linear programming is introduced. ? A hybrid similarity including the global and local information is used to enhance the retrieval accuracy. ? Experimental results corroborate that our proposed method works well for lengthy documents. ? Our proposed two-step retrieval system can serve as a general and computationally efficient solution to DR. This paper presents a multi-level matching method for document retrieval (DR) using a hybrid document similarity. Documents are represented by multi-level structure including document level and paragraph level. This multi-level-structured representation is designed to model underlying semantics in a more flexible and accurate way that the conventional flat term histograms find it hard to cope with. The matching between documents is then transformed into an optimization problem with Earth Mover's Distance (EMD). A hybrid similarity is used to synthesize the global and local semantics in documents to improve the retrieval accuracy. In this paper, we have performed extensive experimental study and verification. The results suggest that the proposed method works well for lengthy documents with evident spatial distributions of terms.

Highlights? The proposed self-adjusting e-course generation process supports personalized learning environments. The process adopts evolutionary algorithms to compose personalized e-courses for individual learners. The feedback information from learners is necessary for adjusting the difficulty of the e-learning material. The experimented results indicate that the past learners' experiences can benefit the next learner. This paper proposes a self-adjusting e-course generation process, which support to provide a truly personalized learning environment. The proposed process is divided into four steps: (1) determining learning concept structure, (2) adjusting the difficulty of the e-learning material, (3) analyzing a learner's ability and learning goals, and (4) composing personalized e-courses. Meanwhile, this paper applies the collaborative voting approach to determine the difficulty of the e-learning material, and the maximum likelihood estimation (MLE) to analyze a learner's ability and her/his learning goals. Since evolutionary algorithms (EAs) have been developed to find close optimal solutions, this paper adopts them to compose personalized e-courses that meet individual learners' demands. Once a learner learns one or more learning concepts covered in a personalized e-course, the feedback information from the learner must be returned to: (I) self-adjust the difficulty of the e-learning material for step 2, and (II) update the learners' ability and learning goals for step 3. Furthermore, to find appropriate EAs for personalized e-course composition, this paper devises some experiments to compare two widely applied EAs, Genetic algorithms (GA) and Particle Swarm Optimization (PSO). When the number of e-learning materials is less than 300, the experimented results indicate that the executing effectiveness of PSO is better than that of GA. Besides, to validate the practicability of the proposed process, an e-course authoring tool based on the proposed process is developed to generate personalized e-courses. The generated personalized e-courses have been provided to 103 actual learners who participate in an "Introduction to Computer" curriculum. The investigation results indicate that the proposed process adapts to learners by utilizing the feedback from many learners. In other words, learning experiences of one organization/class can benefit to another organization/class's learners in the same curriculum.

Technology intelligence systems are vital components for planning of technology development and formulation of technology strategies. Although such systems provide computation supports for technology analysis, much effort and intervention of experts, who may be expensive or unavailable, is required in gathering processes of information for analysis. As a remedy, this paper proposes TrendPerceptor, a system that uses a property–function based approach. The proposed system assists experts (1) to identify trends in invention concepts from patents, and (2) to perform evolution trend analysis of patents for technology forecasting. For this purpose, a module of the system uses grammatical analysis of textual information to automatically extract properties and functions, which show innovation directions in a given technology. Using the identified properties and functions, a module for invention concept analysis based on network analysis and a module for evolution trend analysis based on TRIZ (Russian acronym of the Theory of Inventive Problem Solving) trends are suggested. This paper describes the architecture of a system composed of these three modules, and illustrates two case studies using the system.

The purpose of this paper is to develop a customer-oriented organisational diagnostic model, 'PARA' model, based on data mining of customer-complaint databases. The proposed 'PARA' model, which is designed to diagnose and correct service failures, takes its name from the initial letters of the four analytical stages of the model: (i) 'primary diagnosis'; (ii) 'advanced diagnosis'; (iii) 'review'; and (iv) 'action'. In the primary-diagnosis stage, the customer-complaint database is comprehensively analysed to identify themes and categories of complaints. In the advanced-diagnosis stage, a data-mining technique is employed to investigate the relationship between the categories of customer complaints and the deficiencies of the service system. In the review stage, the identified weaknesses of the service system are reviewed and awareness of these weaknesses is enhanced among the organisation's employees. In the action stage, a strategy of action plans for improvement is developed. An empirical case study is conducted to demonstrate the practical efficacy of the 'PARA' model. The paper concludes by summarising the advantages of the proposed model and the implications for future research.

One of the essential goals of test designing is to select items with the most discriminative power. In the past, most research has assumed there is no dependent relationship among test items, so that test papers are often produced by selecting items with individual discriminations. However, in actuality, test items may relate to other items and the overall discrimination of a test paper cannot be simply aggregated. Therefore, this study proposes a two-step framework to design test papers by choosing discriminative item combinations from the item bank. The proposed approach (the process) first analyzing entails the archival tests to discover substitute items, as well as recognize discriminative test itemsets by using data mining technology. Then, test items can be recommended to compose a discriminative test paper. Finally, a real life case is used to test the proposed method. The test data is provided by the Chinese Enterprise Planning Association (CERP) in Taiwan. The experimental results indicate that: (1) the two-step method can complete the test design task efficiently; (2) the newly composed test paper presents highly discriminative; and (3) the discrimination power of our test paper is very close to the theoretic maximum value based on Item Response Theory.

Mobile communication has become the major choice of personal communication today. With the rapid proliferation of wireless network technologies, mobile users can communicate through more and more different wireless networks. In such a heterogeneous wireless network environment, traditional wisdom is to ensure mobile users always best connected (ABC) to one of their nearby available networks. However, the ABC strategy may be sub-optimal as the cost and quality of multimedia communication depends not only on the access network of the caller (source) but also on that of the callee (destination). Thus, intelligently selecting a best network connection, or a source and destination access network pair, for users is necessary in next generation multimedia communication, in which various media types, such as voice, video and data, are related and jointly contribute to the communication quality and cost. This paper utilized Analytic Network Process (ANP) to model the problem of call setup strategy for multimedia communication. ANP is powerful in representing complex networked decision process. In the paper, we illustrated, by an example, how ANP sets up calls that fulfill the requirement of users.

The objective of this paper is to construct a lightweight Intrusion Detection System (IDS) aimed at detecting anomalies in networks. The crucial part of building lightweight IDS depends on preprocessing of network data, identifying important features and in the design of efficient learning algorithm that classify normal and anomalous patterns. Therefore in this work, the design of IDS is investigated from these three perspectives. The goals of this paper are (i) removing redundant instances that causes the learning algorithm to be unbiased (ii) identifying suitable subset of features by employing a wrapper based feature selection algorithm (iii) realizing proposed IDS with neurotree to achieve better detection accuracy. The lightweight IDS has been developed by using a wrapper based feature selection algorithm that maximizes the specificity and sensitivity of the IDS as well as by employing a neural ensemble decision tree iterative procedure to evolve optimal features. An extensive experimental evaluation of the proposed approach with a family of six decision tree classifiers namely Decision Stump, C4.5, Naive Baye's Tree, Random Forest, Random Tree and Representative Tree model to perform the detection of anomalous network pattern has been introduced.

This paper introduces a new open source, knowledge-based framework for automatic interpretation of remote sensing images, called InterIMAGE. This framework exhibits a flexible modular architecture, in which image processing operators can be associated to both root and leaf nodes of a semantic network, which accounts for a differential strategy in comparison to other object-based image analysis platforms currently available. The architecture, main features as well as an overview on the interpretation strategy implemented in InterIMAGE are presented. The paper also reports an experiment on the classification of landforms. Different geomorphometric and textural attributes obtained from ASTER/Terra images were combined with fuzzy logic to drive the interpretation semantic network. Object-based statistical agreement indices, estimated from a comparison between the classified scene and a reference map, were used to assess the classification accuracy. The InterIMAGE interpretation strategy yielded a classification result with strong agreement and proved to be effective for the extraction of landforms.

The Wii, which is a home video game console released by Nintendo, has rapidly become the world's most popular console since its release in November 2006. A distinguishing feature of the Wii is its innovative motion-sensitive controller, the Wii Remote, which could be used as a hand-held pointing device and that has brought many people to gaming for the first time. The popularity of Wii and the innovative design of Wii Remote provide us the motivation to develop a novel air painter system which includes basic painting functions, free sketching, and Chinese, English and Japanese recognition. Vibration and noise effect often influence the recognition of the system, so several fault tolerance mechanisms are proposed in this paper to reduce these effects. Moreover, a unifying character and free sketch recognition scheme, which is based on the stroke code sequences, is proposed to process the recognition. One of the advantages of this approach is that the same mechanism could be applied to the whole system and it could incorporate the fault tolerance mechanisms proposed in this paper.

The paper reports an interface designed for three subsystems-the knowledge acquisition subsystem with dynamic disease knowledge base, intelligent disease diagnosis subsystem with object-oriented intelligent-inference model and intelligent tutor for crop disease with audio-visual graphical user web-interface. The paper describes the design features; functionality and development of intelligent multimedia web interface. The results of its evaluation are also presented. A novel approach of rule promotion based on fuzzy logic is used in the system for drawing intelligent inferences for crop disease management. A text-to-speech (TTS) converter is used for providing capability of text-to-talking user interface. It provides the highly-effective interactive user interface on web for live interactions. It helps significantly and gives solutions of plant pathological problems in short spell. The application software is built using the Microsoft .NET framework, ASP.NET web-application framework provided in Visual Studio.NET. The Microsoft Speech SDK (5.1) is used to develop text to voice multimedia interface. The dynamic knowledge base is implemented using SQL server.

Stock index forecasting is one of the major activities of financial firms and private investors in making investment decisions. Although many techniques have been developed for predicting stock index, building an efficient stock index forecasting model is still an attractive issue since even the smallest improvement in prediction accuracy can have a positive impact on investments. In this paper, an efficient cerebellar model articulation controller neural network (CAMC NN) is proposed for stock index forecasting. The traditional CAMC NN scheme has been successfully used in robot control due to its advantages of fast learning, reasonable generalization capability and robust noise resistance. But, few studies have been reported in using a CMAC NN scheme for forecasting problems. To improve the forecasting performance, this paper presents an efficient CMAC NN scheme. The proposed CMAC NN scheme employs a high quantization resolution and a large generalization size to reduce generalization error, and uses an efficient and fast hash coding to accelerate many-to-few mappings. The forecasting results and robustness evaluation of the proposed CMAC NN scheme were compared with those of a support vector regression (SVR) and a back-propagation neural network (BPNN). Experimental results from Nikkei 225 and Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) closing indexes show that the performance of the proposed CMAC NN scheme was superior to the SVR and BPNN models.

This paper adopts data envelopment analysis (DEA), a robust and reliable evaluation method widely applied in various fields to explore the key indicators contributing to the learning performance of English freshmen writing courses in a university of Taiwan from the academic year 2004 to 2006. The results of DEA model applied in learning performance change our original viewpoint and reveal that some decision-making units (DMUs) with higher actual values of inputs and outputs have lower efficiency because the relative efficiency of each DMU is measured by their distance to the efficiency frontier. DMUs may refer to different facet reference sets according to their actual values located in lower or higher ranges. In the managerial strategy of educational field, the paper can encourage inefficient DMUs to always compare themselves with efficient DMUs in their range and make improvement little by little. The results of DEA model can also give clear indicators and the percentage of which input and output items to improve. The paper also demonstrates that the benchmarking characteristics of the DEA model can automatically segment all the DMUs into different levels based on the indicators fed into the performance evaluation mechanism. The efficient DMUs on the frontier curve can be considered as the boundaries of the classification which are systematically defined by the DEA model according to the statistic distribution.

In this paper, two data mining algorithms are applied to build a churn prediction model using credit card data collected from a real Chinese bank. The contribution of four variable categories: customer information, card information, risk information, and transaction activity information are examined. The paper analyzes a process of dealing with variables when data is obtained from a database instead of a survey. Instead of considering the all 135 variables into the model directly, it selects the certain variables from the perspective of not only correlation but also economic sense. In addition to the accuracy of analytic results, the paper designs a misclassification cost measurement by taking the two types error and the economic sense into account, which is more suitable to evaluate the credit card churn prediction model. The algorithms used in this study include logistic regression and decision tree which are proven mature and powerful classification algorithms. The test result shows that regression performs a little better than decision tree.

Building hazard assessment prior to earthquake occurrence exposes interesting problems especially in earthquake prone areas. Such an assessment provides an early warning system for building owners as well as the local and central administrators about the possible hazards that may occur in the next scenario earthquake event, and hence pre- and post-earthquake preparedness can be arranged according to a systematic program. For such an achievement, it is necessary to have efficient models for the prediction of hazard scale of each building within the study area. Although there are subjective intensity index methods for such evaluations, the objective of this paper is to propose a useful tool through fuzzy logic (FL) to classify the buildings that would be vulnerable to earthquake hazard. The FL is a soft computing intelligent reasoning methodology, which is rapid, simple and easily applicable with logical and rational association between the building-hazard categories and the most effective factors. In this paper, among the most important factors are the story number (building height), story height ratio, cantilever extension ratio, moment of inertia (stiffness), number of frames, column and shear wall area percentages. Their relationships with the five hazard categories are presented through a supervised hazard center classification method. These five categories are ''none'', ''slight'', ''moderate'', ''extensive'', and ''complete'' hazard classes. A new supervised FL classification methodology is proposed similar to the classical fuzzy c-means procedure for the allocation of hazard categories to individual buildings. The application of the methodology is presented for Zeytinburnu quarter of Istanbul City, Turkey. It is observed that out of 747 inventoried buildings 7.6%, 50.0%, 14.6%, 20.1%, and 7.7% are subject to expected earthquake with ''none'', ''slight'', ''moderate'', ''extensive'', and ''complete'' hazard classes, respectively.

Online privacy has consistently been a major concern for customers that has grown commensurately with the growth of e-services. Service providers have responded by making their privacy policies clearer for customers; however, most providers use legacy systems that are unable to actually change and adapt to user's concerns, which can lead to fewer customers using the system. Hence, the purpose of this paper is to propose a context-aware privacy policy negotiation service. To do so, we adopt a Galois lattice theory to generate policy concepts embedded in e-services. Based on the Galois lattice, we develop a process for generating privacy policy rules. To show the feasibility of the ideas proposed in this paper, we perform a simulation test with two different online auction sites as an illustrative case in terms of two metrics: the number of rules generated and success throughput. Desirable features in applying the Galois lattice approach to context-aware privacy policy negotiation service are discussed.

Identification of transcription factor binding sites is a key task to understand gene regulation mechanism to discover gene networks and functions. Clustering approach is proved to be useful when finding such patterns residing in promoter regions of co-regulated genes. Four clustering algorithms, Self-Organizing Map, K-Means, Fuzzy C-Means and Expectation-Maximization are studied in this paper to discover motifs in datasets extracted from Saccharomyces cerevisiae, Escherichia coli, Droshophila melanogaster and Homo sapiens DNA sequences. Required modifications to clustering algorithms in order to adapt them to motif finding task are presented through the paper. Then, their motif-finding performances are discussed carefully and evaluated against a popular motif-finding method, MEME. © 2011 Elsevier Ltd. All rights reserved.

Kreng and Tan [Kreng, V. B. & Tan, S. J. (2010). The optimal replenishment decisions under two levels of trade credit policy depending on the order quantity. Expert Systems with Applications 37, 5514-5522] explore the optimal replenishment decisions under two levels of trade credit policy depending on the order quantity. Their inventory model is new and interesting. However, this paper indicates that their solution procedures can be improved further. So, the main purpose of this paper is to incorporate and simplify them for Kreng and Tan (2010).

The first part of the paper introduces a novel tool for modeling and simulation of discrete event system. This tool called GPenSIM is a Petri net based simulator and offers significant benefits to model builders such as flexibility to include diverse libraries, ease of extending the models, and ease of programming. The second part of the paper presents a case study on modeling and optimization of airport traffic management; this study is to explore air traffic management capability of Evenes airport in Norway. The case study shows that with GPenSIM, modeling and simulation problems of large industrial discrete event systems can be done. Future research directions are discussed as well.

This paper proposes a method for scene categorization by integrating region contextual information into the popular Bag-of-Visual-Words approach. The Bag-of-Visual-Words approach describes an image as a bag of discrete visual words, where the frequency distributions of these words are used for image categorization. However, the traditional visual words suffer from the problem when faced these patches with similar appearances but distinct semantic concepts. The drawback stems from the independently construction each visual word. This paper introduces Region-Conditional Random Fields model to learn each visual word depending on the rest of the visual words in the same region. Comparison with the traditional Conditional Random Fields model, there are two areas of novelty. First, the initial label of each patch is automatically defined based on its visual feature rather than manually labeling with semantic labels. Furthermore, the novel potential function is built under the region contextual constraint. The experimental results on the three well-known datasets show that Region Contextual Visual Words indeed improves categorization performance compared to traditional visual words.

Prudence analysis (PA) is a relatively new, practical and highly innovative approach to solving the problem of brittleness in knowledge based system (KBS) development. PA is essentially an online validation approach where as each situation or case is presented to the KBS for inferencing the result is simultaneously validated. Therefore, instead of the system simply providing a conclusion, it also provides a warning when the validation fails. Previous studies have shown that a modification to multiple classification ripple-down rules (MCRDR) referred to as rated MCRDR (RM) has been able to achieve strong and flexible results in simulated domains with artificial data sets. This paper presents a study into the effectiveness of RM in an eHealth document monitoring and classification domain using human expertise. Additionally, this paper also investigates what affect PA has when the KBS developer relied entirely on the warnings for maintenance. Results indicate that the system is surprisingly robust even when warning accuracy is allowed to drop quite low. This study of a previously little touched area provides a strong indication of the potential for future knowledge based system development.

Quite often, in order to derive meaningful insights, accounting researchers have to analyze large bodies of text. Usually, this is done manually by several human coders, which makes the process time consuming, expensive, and often neither replicable nor accurate. In an attempt to mitigate these problems, we perform a feasibility study investigating the applicability of computer-aided content analysis techniques onto the domain of accounting research. Krippendorff (1980) defines an algorithms reliability as its stability, reproducibility and accuracy. Since in computer-aided text classification, which is inherently objective and repeatable, the first two requirements, stability and reproducibility, are not an issue, this paper focuses exclusively on the third requirement, the algorithms accuracy. It is important to note that, although inaccurate classification results are completely worthless, it is surprising to see how few research papers actually mention the accuracy of the used classification methodology. After a survey of the available techniques, we perform an in depth analysis of the most promising one, LPU (Learning from Positive and Unlabelled), which turns out to have an F-value and accuracy of about 90%, which means that, given a random text, it has a 90% probability of classifying it correctly. Highlights ► We examine text classification algorithms in an accounting setting. ► The LPU-algorithm is the most appropriate one for our data. ► We develop a four stage classification process. ► LPU classifies 90% of the documents accurately into positive, negative or unlabelled.