Databases and Information Systems Commons^™

Duol: A Double Updating Approach For Online Learning, Peilin Zhao, Steven C. H. Hoi, Rong Jin

Research Collection School Of Computing and Information Systems

In most online learning algorithms, the weights assigned to the misclassified examples (or support vectors) remain unchanged during the entire learning process. This is clearly insufficient since when a new misclassified example is added to the pool of support vectors, we generally expect it to affect the weights for the existing support vectors. In this paper, we propose a new online learning method, termed Double Updating Online Learning, or DUOL for short. Instead of only assigning a fixed weight to the misclassified example received in current trial, the proposed online learning algorithm also tries to update the weight for one …

Learning Bregman Distance Functions And Its Application For Semi-Supervised Clustering, Lei Wu, Rong Jin, Steven C. H. Hoi, Jianke Zhu, Nenghai Yu

Research Collection School Of Computing and Information Systems

Learning distance functions with side information plays a key role in many machine learning and data mining applications. Conventional approaches often assume a Mahalanobis distance function. These approaches are limited in two aspects: (i) they are computationally expensive (even infeasible) for high dimensional data because the size of the metric is in the square of dimensionality; (ii) they assume a fixed metric for the entire input space and therefore are unable to handle heterogeneous data. In this paper, we propose a novel scheme that learns nonlinear Bregman distance functions from side information using a nonparametric approach that is similar to …

To Trust Or Not To Trust? Predicting Online Trusts Using Trust Antecedent Framework, Viet-An Nguyen, Ee Peng Lim, Jing Jiang, Aixin Sun

Research Collection School Of Computing and Information Systems

This paper analyzes the trustor and trustee factors that lead to inter-personal trust using a well studied Trust Antecedent framework in management science. To apply these factors to trust ranking problem in online rating systems, we derive features that correspond to each factor and develop different trust ranking models. The advantage of this approach is that features relevant to trust can be systematically derived so as to achieve good prediction accuracy. Through a series of experiments on real data from Epinions, we show that even a simple model using the derived features yields good accuracy and outperforms MoleTrust, a trust …

Cyber Attacks: Does Physical Boundary Matter?, Qiu-Hong Wang, Seung-Hyun Kim

Research Collection School Of Computing and Information Systems

Information security issues are characterized with interdependence. Particularly, cyber criminals can easily cross national boundaries and exploit jurisdictional limitations between countries. Thus, whether cyber attacks are spatially autocorrelated is a strategic issue for government authorities and a tactic issue for insurance companies. Through an empirical study of cyber attacks across 62 countries during the period 2003-2007, we find little evidence on the spatial autocorrelation of cyber attacks at any week. However, after considering economic opportunity, IT infrastructure, international collaboration in enforcement and conventional crimes, we find strong evidence that cyber attacks were indeed spatially autocorrelated as they moved over time. …

Wake Up Or Fall Asleep: Value Implication Of Trusted Computing, Nan Hu, Jianhui Huang, Ling Liu, Yingjiu Li, Dan Ma

Research Collection School Of Computing and Information Systems

More than 10 years have passed since trusted computing (TC) technology was introduced to the market; however, there is still no consensus about its value. The increasing importance of user and enterprise security and the security promised by TC, coupled with the increasing tension between the proponents and the opponents of TC, make it timely to investigate the value relevance of TC in terms of both capital market and accounting performance. Based on both price and volume studies, we found that news releases related to the adoption of the TC technology had no information content. All investors, regardless of whether …

On Strategies For Imbalanced Text Classification Using Svm: A Comparative Study, Aixin Sun, Ee Peng Lim, Ying Liu

Research Collection School Of Computing and Information Systems

Many real-world text classification tasks involve imbalanced training examples. The strategies proposed to address the imbalanced classification (e.g., resampling, instance weighting), however, have not been systematically evaluated in the text domain. In this paper, we conduct a comparative study on the effectiveness of these strategies in the context of imbalanced text classification using Support Vector Machines (SVM) classifier. SVM is the interest in this study for its good classification accuracy reported in many text classification tasks. We propose a taxonomy to organize all proposed strategies following the training and the test phases in text classification tasks. Based on the taxonomy, …

A Contrast Pattern Based Clustering Quality Index For Categorical Data, Qingbao Liu, Guozhu Dong

Kno.e.sis Publications

Since clustering is unsupervised and highly explorative, clustering validation (i.e. assessing the quality of clustering solutions) has been an important and long standing research problem. Existing validity measures have significant shortcomings. This paper proposes a novel contrast pattern based clustering quality index (CPCQ) for categorical data, by utilizing the quality and diversity of the contrast patterns (CPs) which contrast the clusters in clusterings. High quality CPs can characterize clusters and discriminate them against each other. Experiments show that the CPCQ index (1) can recognize that expert-determined classes are the best clusters for many datasets from the UCI repository; (2) does …

High-Dimensional Software Engineering Data And Feature Selection, Huanjing Wang, Taghi M. Khoshgoftaar, Kehan Gao

Computer Science Faculty Publications

Software metrics collected during project development play a critical role in software quality assurance. A software practitioner is very keen on learning which software metrics to focus on for software quality prediction. While a concise set of software metrics is often desired, a typical project collects a very large number of metrics. Minimal attention has been devoted to finding the minimum set of software metrics that have the same predictive capability as a larger set of metrics – we strive to answer that question in this paper. We present a comprehensive comparison between seven commonly-used filter-based feature ranking techniques (FRT) …

Ontology-Driven Provenance Management In Escience: An Application In Parasite Research, Satya S. Sahoo, D. Brent Weatherly, Raghava Mutharaju, Pramod Anantharam, Amit P. Sheth, Rick L. Tarleton

Kno.e.sis Publications

Provenance, from the French word “provenir”, describes the lineage or history of a data entity. Provenance is critical information in scientific applications to verify experiment process, validate data quality and associate trust values with scientific results. Current industrial scale eScience projects require an end-to-end provenance management infrastructure. This infrastructure needs to be underpinned by formal semantics to enable analysis of large scale provenance information by software applications. Further, effective analysis of provenance information requires well-defined query mechanisms to support complex queries over large datasets. This paper introduces an ontology-driven provenance management infrastructure for biology experiment data, as part …

An Anytime Algorithm For Computing Inconsistency Measurement, Yue Ma, Guilin Qi, Guohui Xiao, Pascal Hitzler, Zuoquan Lin

Computer Science and Engineering Faculty Publications

Measuring inconsistency degrees of inconsistent knowledge bases is an important problem as it provides context information for facilitating inconsistency handling. Many methods have been proposed to solve this problem and a main class of them is based on some kind of paraconsistent semantics. In this paper, we consider the computational aspects of inconsistency degrees of propositional knowledge bases under 4-valued semantics. We first analyze its computational complexity. As it turns out that computing the exact inconsistency degree is intractable, we then propose an anytime algorithm that provides tractable approximation of the inconsistency degree from above and below. We show that …

A Gis Hub At Pace University, Peggy Minis, Hsui-Lin Winkler

Cornerstone 2 Reports : Community Outreach and Empowerment Through Service Learning and Volunteerism

The Thinkfinity Grant is to use technology to develop a GIS Hub at Pace University. The Hub is intended to show the larger community the work done at Pace and to show that our students and faculty are using GIS to solve geographically-based problems for communities and organizations. It also is intended to serve as a site from which users can download data to make their own maps and as a place where the larger community can find examples of maps and have the ability to manipulate maps.

Vireo/Dvmm At Trecvid 2009: High-Level Feature Extraction, Automatic Video Search, And Content-Based Copy Detection, Chong-Wah Ngo, Yu-Gang Jiang, Xiao-Yong Wei, Wanlei Zhao, Yang Liu, Jun Wang, Shiai Zhu, Shih-Fu Chang

Research Collection School Of Computing and Information Systems

This paper presents overview and comparative analysis of our systems designed for 3 TRECVID 2009 tasks: high-level feature extraction, automatic search, and content-based copy detection.

Ensemble And Individual Noise Reduction Method For Induction-Motor Signature Analysis, Zhaoxia Wang, C.S. Chang, Tw Chua, W.W Tan

Research Collection School Of Computing and Information Systems

Unlike a fixed-frequency power supply, the voltagesupplying an inverter-fed motor is heavily corrupted by noises,which are produced from high-frequency switching leading tonoisy stator currents. To extract useful information from statorcurrentmeasurements, a theoretically sound and robust denoisingmethod is required. The effective filtering of these noisesis difficult with certain frequency-domain techniques, such asFourier transform or Wavelet analysis, because some noises havefrequencies overlapping with those of the actual signals, andsome have high noise-to-frequency ratios. In order to analyze thestatistical signatures of different types of signals, a certainnumber is required of the individual signals to be de-noisedwithout sacrificing the individual characteristic and quantity ofthe …

Online Fault Detection Of Induction Motors Using Independent Component Analysis And Fuzzy Neural Network, Zhaoxia Wang, C. S. Chang, X. German, W.W. Tan

Research Collection School Of Computing and Information Systems

This paper proposes the use of independent component analysis and fuzzy neural network for online fault detection of induction motors. The most dominating components of the stator currents measured from laboratory motors are directly identified by an improved method of independent component analysis, which are then used to obtain signatures of the stator current with different faults. The signatures are used to train a fuzzy neural network for detecting induction-motor problems such as broken rotor bars and bearing fault. Using signals collected from laboratory motors, the robustness of the proposed method for online fault detection is demonstrated for various motor …

What Makes Categories Difficult To Classify?, Aixin Sun, Ee Peng Lim, Ying Liu

Research Collection School Of Computing and Information Systems

In this paper, we try to predict which category will be less accurately classified compared with other categories in a classification task that involves multiple categories. The categories with poor predicted performance will be identified before any classifiers are trained and additional steps can be taken to address the predicted poor accuracies of these categories. Inspired by the work on query performance prediction in ad-hoc retrieval, we propose to predict classification performance using two measures, namely, category size and category coherence. Our experiments on 20-Newsgroup and Reuters-21578 datasets show that the Spearman rank correlation coefficient between the predicted rank of …

Trust Relationship Prediction Using Online Product Review Data, Nan Ma, Ee Peng Lim, Viet-An Nguyen, Aixin Sun

Research Collection School Of Computing and Information Systems

Trust between users is an important piece of knowledge that can be exploited in search and recommendation.Given that user-supplied trust relationships are usually very sparse, we study the prediction of trust relationships using user interaction features in an online user generated review application context. We show that trust relationship prediction can achieve better accuracy when one adopts personalized and cluster-based classification methods. The former trains one classifier for each user using user-specific training data. The cluster-based method first constructs user clusters before training one classifier for each user cluster. Our proposed methods have been evaluated in a series of experiments …

Trust-Oriented Composite Services Selection And Discovery, Lei Li, Yan Wang, Ee Peng Lim

Research Collection School Of Computing and Information Systems

In Service-Oriented Computing (SOC) environments, service clients interact with service providers for consuming services. From the viewpoint of service clients, the trust level of a service or a service provider is a critical issue to consider in service selection and discovery, particularly when a client is looking for a service from a large set of services or service providers. However, a service may invoke other services offered by different providers forming composite services. The complex invocations in composite services greatly increase the complexity of trust-oriented service selection and discovery. In this paper, we propose novel approaches for composite service representation, …

Udel/Smu At Trec 2009 Entity Track, Wei Zheng, Swapna Gottipati, Jing Jiang, Hui Fang

Research Collection School Of Computing and Information Systems

We report our methods and experiment results from the collaborative participation of the InfoLab group from University of Delaware and the school of Information Systems from Singapore Management University in the TREC 2009 Entity track. Our general goal is to study how we may apply language modeling approaches and natural language processing techniques to the task. Specically, we proposed to find supporting information based on segment retrieval, to extract entities using Stanford NER tagger, and to rank entities based on a previously proposed probabilistic framework for expert finding.

A Survey Of The Semantic Specification Of Sensors, Michael Compton, Cory Andrew Henson, Laurent Lefort, Holger Neuhaus, Amit P. Sheth

Kno.e.sis Publications

Semantic sensor networks use declarative descriptions of sensors promote reuse and integration, and to help solve the difficulties of installing, querying and maintaining complex, heterogeneous sensor networks. This paper reviews the state of the art for the semantic specification of sensors, one of the fundamental technologies in the semantic sensor network vision. Twelve sensor ontologies are reviewed and analysed for the range and expressive power of their concepts. The reasoning and search technology developed in conjunction with these ontologies is also reviewed, as is technology for annotating OGC standards with links to ontologies. Sensor concepts that cannot be expressed accurately …

Provenir Ontology: Towards A Framework For Escience Provenance Management, Satya S. Sahoo, Amit P. Sheth

Kno.e.sis Publications

Provenance metadata describes the 'lineage' or history of an entity and necessary information to verify the quality of data, validate experiment protocols, and associate trust value with scientific results. eScience projects generate data and the associated provenance metadata in a distributed environment (such as myGrid) and on a very large scale that often precludes manual analysis. Given this scenario, provenance information should be, (a) interoperable across projects, research groups, and application domains, and (b) support analysis over large datasets using reasoning to discover implicit information. In this paper, we introduce an ontology-driven framework for eScience provenance management underpinned by an …

Suggestions For Owl 3, Pascal Hitzler

Computer Science and Engineering Faculty Publications

With OWL 2 about to be completed, it is the right time to start discussions on possible future modifications of OWL. We present here a number of suggestions in order to discuss them with the OWL user community. They encompass expressive extensions on polynomial OWL 2 profiles, a suggestion for an OWL Rules language, and expressive extensions for OWL DL.

Ibm Altocumulus: A Cross-Cloud Middleware And Platform, E. Michael Maximilien, Ajith Harshana Ranabahu, Roy Engehausen, Laura Anderson

Kno.e.sis Publications

Cloud computing has become the new face of computing and promises to offer virtually unlimited, cheap, readily available, "utility type" computing resources. Many vendors have entered this market with different offerings ranging from infrastructure-as-a-service such as Amazon, to fully functional platform services such as Google App Engine. However, as a result of this heterogeneity, deploying applications to a cloud and managing them needs to be done using vendor specific methods. This "lock in" is seen as a major hurdle in adopting cloud technologies to the enterprise. IBM Altocumulus, the cloud middleware platform from IBM Almaden Services Research, aims to solve …

Paraconsistent Reasoning For Owl 2, Yue Ma, Pascal Hitzler

Computer Science and Engineering Faculty Publications

A four-valued description logic has been proposed to reason with description logic based inconsistent knowledge bases. This approach has a distinct advantage that it can be implemented by invoking classical reasoners to keep the same complexity as under the classical semantics. However, this approach has so far only been studied for the basid description logic ALC. In this paper, we further study how to extend the four-valued semantics to the more expressive description logic SROIQ which underlies the forthcoming revision of the Web Ontology Language, OWL 2, and also investigate how it fares when adapated to tractable description logics including …

A Preferential Tableaux Calculus For Circumscriptive Alco, Stephan Grimm, Pascal Hitzler

Computer Science and Engineering Faculty Publications

Nonmonotonic extensions of description logics (DLs) allow for default and local closed-world reasoning and are an acknowledged desired feature for applications, e.g. in the Semantic Web. A recent approach to such an extension is based on McCarthy's circumscription, which rests on the principle of minimising the extension of selected predicates to close off dedicated parts of a domain model. While decidability and complexity results have been established in the literature, no practical algorithmisation for circumscriptive DLs has been proposed so far. In this paper, we present a tableaux calculus that can be used as a decision procedure for concept satisfiability …

A Best Practice Model For Cloud Middleware Systems, Ajith Harshana Ranabahu, E. Michael Maximilien

Kno.e.sis Publications

Cloud computing is the latest trend in computing where the intention is to facilitate cheap, utility type computing resources in a service-oriented manner. However, the cloud landscape is still maturing and there are heterogeneities between the clouds, ranging from the application development paradigms to their service interfaces,and scaling approaches. These differences hinder the adoption of cloud by major enterprises. We believe that a cloud middleware can solve most of these issues to allow cross-cloud inter-operation. Our proposed system is Altocumulus, a cloud middleware that homogenizes the clouds. In order to provide the best use of the cloud resources and make …

Context And Domain Knowledge Enhanced Entity Spotting In Informal Text, Daniel Gruhl, Meena Nagarajan, Jan Pieper, Christine Robson, Amit P. Sheth

Kno.e.sis Publications

This paper explores the application of restricted relationship graphs (RDF) and statistical NLP techniques to improve named entity annotation in challenging Informal English domains. We validate our approach using on-line forums discussing popular music. Named entity annotation is particularly difficult in this domain because it is characterized by a large number of ambiguous entities, such as the Madonna album “Music” or Lilly Allen’s pop hit “Smile”.

We evaluate improvements in annotation accuracy that can be obtained by restricting the set of possible entities using real-world constraints. We find that constrained domain entity extraction raises the annotation accuracy significantly, making an …

Distribution-Based Concept Selection For Concept-Based Video Retrieval, Juan Cao, Hongfang Jing, Chong-Wah Ngo, Yongdong Zhang

Research Collection School Of Computing and Information Systems

Query-to-concept mapping plays one of the keys to concept-based video retrieval. Conventional approaches try to find concepts that are likely to co-occur in the relevant shots from the lexical or statistical aspects. However, the high probability of co-occurrence alone cannot ensure its effectiveness to distinguish the relevant shots from the irrelevant ones. In this paper, we propose distribution-based concept selection (DBCS) for query-to-concept mapping by analyzing concept score distributions of within and between relevant and irrelevant sets. In view of the imbalance between relevant and irrelevant examples, two variants of DBCS are proposed respectively by considering the two-sided and onesided …

Scalable Detection Of Partial Near-Duplicate Videos By Visual-Temporal Consistency, Hung-Khoon Tan, Chong-Wah Ngo, Richang Hong, Tat-Seng Chua

Research Collection School Of Computing and Information Systems

Following the exponential growth of social media, there now exist huge repositories of videos online. Among the huge volumes of videos, there exist large numbers of near-duplicate videos. Most existing techniques either focus on the fast retrieval of full copies or near-duplicates, or consider localization in a heuristic manner. This paper considers the scalable detection and localization of partial near-duplicate videos by jointly considering visual similarity and temporal consistency. Temporal constraints are embedded into a network structure as directed edges. Through the structure, partial alignment is novelly converted into a network flow problem where highly efficient solutions exist. To precisely …

Domain Adaptive Semantic Diffusion For Large Scale Context-Based Video Annotation, Yu-Gang Jiang, Jun Wang, Shih-Fu Chang, Chong-Wah Ngo

Research Collection School Of Computing and Information Systems

Learning to cope with domain change has been known as a challenging problem in many real-world applications. This paper proposes a novel and efficient approach, named domain adaptive semantic diffusion (DASD), to exploit semantic context while considering the domain-shift-of-context for large scale video concept annotation. Starting with a large set of concept detectors, the proposed DASD refines the initial annotation results using graph diffusion technique, which preserves the consistency and smoothness of the annotation over a semantic graph. Different from the existing graph learning methods which capture relations among data samples, the semantic graph treats concepts as nodes and the …

First Acm Sigmm International Workshop On Social Media (Wsm'09), Suzanne Boll, Steven C. H. Hoi, Jiebo Luo, Rong Jin, Dong Xu, Irwin King

Research Collection School Of Computing and Information Systems

No abstract provided.