Another extension of Nonlinear Integral, Upper and Lower Nonlinear Integrals, which is a pair of extreme nonlinear integrals to contain all types of Nonlinear Integrals in the same scheme, is also proposed. It can give a set of upper and lower bounds which include all types of Nonlinear Integrals. We tried to find a solution with the smallest distance between the upper and lower bounds and the smallest error which is a NP hard problem. So we use the multi-objective optimization method to find a set of results for the regression model based on the Upper and Lower Nonlinear Integrals. We can just select one or more optimal solution(s) for a specific problem from the set of results. A weather predictor based on this model has been constructed to predict the next days temperature changing trend and range. / Finally, a NI based data mining framework has been established for identifying the chance of developing liver cancer based on the Hepatitis B Virus DNA sequence data. We have shown that the framework obtains the best diagnosing performance amongst many existing classifiers. / Nonlinear Integral (NI) is a useful integration tool. It has been applied to many areas including classification and regression. The classical method relies on a large number of training data, which lead to large time and space complexity. Moreover, the classical Nonlinear Integral has many limitations. For dealing with different situation, we propose Double Nonlinear Integrals and Nonlinear Integrals with Polynomial Kernel to deal with the problems transversely and longitudinally. / The classical Nonlinear Integrals implement projection along a line with respect to the features. But in many cases the linear projection cannot achieve good performance for classification or regression due to the limitation of the integrand. The linear function used for the integrand is just a special type of polynomial functions with respect to the features. We propose Nonlinear Integral with Polynomial Kernel (NIPK) in which a polynomial function is used as the integrand of Nonlinear Integral. It enables the projection to be along different types of curves on the virtual space, so that the virtual values gotten by the Nonlinear Integrals with Polynomial Kernel can be better regularized and easier to deal with. Experiments show that there is evident improvement of performance for NIPK compared to classical NI. / When the data to be classified have special distribution in the data space, the projection may overlap and the classification accuracy will be lowered. For example, when one group of the data is surrounded by the data of another group, or the number of classes for the data is large. To handle this kind of problems; we propose a new classification model based on the Double Nonlinear Integrals. Double Nonlinear Integral means projecting to a 2-Dimensional space by using the Nonlinear Integral twice in succession and classifying the virtual values in the 2-D space corresponding to the original data. Double Nonlinear Integrals can lessen loss of information due to the intersection of different classes on real axis. Accuracy will also be increased accordingly. / Wang, Jinfeng. / Advisers: Kwong Sak Leung; Kin Hong Lee. / Source: Dissertation Abstracts International, Volume: 72-01, Section: B, page: . / Thesis (Ph.D.)--Chinese University of Hong Kong, 2010. / Includes bibliographical references (leaves 139-151). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese.

Wong, Yuen Chau. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2008. / Includes bibliographical references (leaves 80-83). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Overview --- p.1 / Chapter 1.2 --- Motivation --- p.3 / Chapter 1.3 --- Objective --- p.5 / Chapter 1.4 --- Our contribution --- p.5 / Chapter 1.5 --- Organization of the Thesis --- p.6 / Chapter 2 --- Related Work --- p.7 / Chapter 2.1 --- Existing Sentiment Classification Approach --- p.7 / Chapter 2.2 --- Existing Sentiment Analysis Approach --- p.9 / Chapter 2.3 --- Our Approach --- p.11 / Chapter 3 --- Extracting Product Feature Sentences using Supervised Learning Algorithms --- p.12 / Chapter 3.1 --- Overview --- p.12 / Chapter 3.2 --- Association Rules Mining --- p.13 / Chapter 3.2.1 --- Apriori Algorithm --- p.13 / Chapter 3.2.2 --- Class Association Rules Mining --- p.14 / Chapter 3.3 --- Naive Bayesian Classifier --- p.14 / Chapter 3.3.1 --- Basic Idea --- p.14 / Chapter 3.3.2 --- Feature Selection Techniques --- p.15 / Chapter 3.4 --- Experiment --- p.17 / Chapter 3.4.1 --- Data Sets --- p.18 / Chapter 3.4.2 --- Experimental Setup and Evaluation Measures --- p.19 / Chapter 3.4.1 --- Class Association Rules Mining --- p.20 / Chapter 3.4.2 --- Naive Bayesian Classifier --- p.22 / Chapter 3.4.3 --- Effect on Data Size --- p.25 / Chapter 3.5 --- Discussion --- p.27 / Chapter 4 --- Extracting Product Feature Sentences Using Unsupervised Learning Algorithms --- p.28 / Chapter 4.1 --- Overview --- p.28 / Chapter 4.2 --- Unsupervised Learning Algorithms --- p.29 / Chapter 4.2.1 --- K-means Algorithm --- p.29 / Chapter 4.2.2 --- Density-Based Scan --- p.29 / Chapter 4.2.3 --- Hierarchical Clustering --- p.30 / Chapter 4.3 --- Distance Function --- p.32 / Chapter 4.3.1 --- Euclidean Distance --- p.32 / Chapter 4.3.2 --- Jaccard Distance --- p.32 / Chapter 4.4 --- Experiment --- p.33 / Chapter 4.4.1 --- Cluster Labeling --- p.33 / Chapter 4.4.2 --- K-means Algorithm --- p.34 / Chapter 4.4.3 --- Density-Based Scan --- p.35 / Chapter 4.4.4 --- Hierarchical Clustering --- p.36 / Chapter 4.5 --- Discussion --- p.37 / Chapter 5 --- Extracting Product Feature Sentences Using Concept Clustering --- p.39 / Chapter 5.1 --- Overview --- p.39 / Chapter 5.2 --- Distance Function --- p.40 / Chapter 5.2.1 --- Association Weight --- p.40 / Chapter 5.2.2 --- Chi Square --- p.41 / Chapter 5.2.3 --- Mutual Information --- p.41 / Chapter 5.3 --- Experiment --- p.41 / Chapter 5.3.1 --- Effect on Distance Functions --- p.42 / Chapter 5.3.2 --- Extraction of Product Features Clusters --- p.43 / Chapter 5.3.3 --- Labeling of Sentences --- p.45 / Chapter 5.4 --- Discussion --- p.48 / Chapter 6 --- Extracting Product Feature Sentences Using Concept Clustering and Proposed Unsupervised Learning Algorithm --- p.49 / Chapter 6.1 --- Overview --- p.49 / Chapter 6.2 --- Problem Statement --- p.50 / Chapter 6.3 --- Proposed Algorithm - Scalable Thresholds Clustering --- p.50 / Chapter 6.4 --- Properties of the Proposed Unsupervised Learning Algorithm --- p.54 / Chapter 6.4.1 --- Relationship between threshold functions & shape of clusters --- p.54 / Chapter 6.4.2 --- Expansion process --- p.56 / Chapter 6.4.3 --- Impact of Different Threshold Functions --- p.58 / Chapter 6.5 --- Experiment --- p.61 / Chapter 6.5.1 --- Comparative Studies for Clusters Formation and Sentences Labeling with Digital Camera Dataset --- p.62 / Chapter 6.5.2 --- Experiments with New Datasets --- p.67 / Chapter 6.6 --- Discussion --- p.74 / Chapter 7 --- Conclusion and Future Work --- p.76 / Chapter 7.1 --- Compare with Existing Work --- p.76 / Chapter 7.2 --- Contribution & Implication of this Work --- p.78 / Chapter 7.3 --- Future Work & Improvement --- p.79 / REFFERENCE --- p.80 / Chapter A --- Concept Clustering for DC data with DB Scan (Terms in Concept Clusters) --- p.84 / Chapter B --- Concept Clustering for DC data with Single-linkage Hierarchical Clustering (Terms in Concept Clusters) --- p.87 / Chapter C --- Concept Clusters for Digital Camera data (Comparative Studies) --- p.91 / Chapter D --- Concept Clusters for Personal Computer data (Comparative Studies) --- p.98 / Chapter E --- Concept Clusters for Mobile data (Comparative Studies) --- p.103 / Chapter F --- Concept Clusters for MP3 data (Comparative Studies) --- p.109

As computing technology continues to advance, it has become increasingly difficult to find businesses that do not rely, at least in part, upon the collection and analysis of data for the purpose of project management and process improvement. The cost of software tends to increase over time due to its complexity and the cost of employing humans to develop, maintain, and evolve it. To help control the costs, organizations often seek to improve the process by which software systems are developed and evolved. Improvements can be realized by discovering previously unknown or hidden relationships between the artifacts generated as a result of developing a software system. The objective of the work described in this thesis is to provide a visualization tool that helps managers and engineers better plan for future projects by discovering new knowledge gained by synthesizing and visualizing data mined from software repository records from previous projects. / by James J. Mulcahy. / Thesis (M.S.C.S.)--Florida Atlantic University, 2011. / Includes bibliography. / Electronic reproduction. Boca Raton, Fla., 2011. Mode of access: World Wide Web.

University of Macau / Faculty of Science and Technology / Department of Computer and Information Science

Indiana University-Purdue University Indianapolis (IUPUI) / The contribution of this thesis focuses on addressing the challenges of improving and integrating the UniFrame Discovery Service (URDS) and Multi-level Matching (MLM) concepts. The objective was to find enhancements for both URDS and MLM and address the need of a comprehensive discovery service which goes beyond simple attribute based matching. It presents a detailed discussion on developing an enhanced version of URDS with MLM (proURDS). After implementing proURDS, the thesis includes details of experiments with different deployments of URDS components and different configurations of MLM. The experiments and analysis were carried out using proURDS produced MLM contracts. The proURDS referred to a public dataset called QWS dataset. This dataset includes actual information of software components (i.e., web services), which were harvested from the Internet. The proURDS implements the different matching operations as independent operators at each level of matching (i.e., General, Syntactic, Semantic, Synchronization, and QoS). Finally, a case study was carried out with the deployed proURDS. The case study addresses real world component discovery requirements from the earth science domain. It uses the contracts collected from public portals which provide geographical and weather related data.