Rough sets and current trends in computing : third international conference, RSCTC 2002, Malvern, PA, USA, October 14-16, 2002 : proceedings

This volume contains the papers selected for presentation at the Third Inter- tional Conference on Rough Sets and Current Trends in Computing (RSCTC 2002) held at Penn State Great Valley, Malvern, Pennsylvania, U.S.A., 14–16 October 2002. Rough set theoryand its applications constitute a branch of soft computing that has exhibited a signi?cant growth rate during recent years. RSCTC 2002 provided a forum for exchanging ideas among manyresearchers in the rough set communityand in various areas of soft computing and served as a stimulus for mutual understanding and cooperation. In recent years, there have been a number of advances in rough set theoryand applications. Hence, we have witnessed a growing number of international workshops on rough sets and their applications. In addition, it should be observed that one of the beauties of rough sets and the rough set philosophyis that it tends to complement and reinforce research in manytraditional research areas and applications. This is the main reason that manyinternational conferences are now including rough sets into the list of topics. Rough Sets and Current Trends in Computing Preface RSCTC 2002 Conference Committee Table of Contents In Pursuit of Patterns in Data Reasoning from Data - The Rough Set Way Introduction Basic Concepts Decision Rules Properties of Decision Rules Granularity of Data and Flow Graphs Decision Space Flow Diagrams, Another Approach Conclusions References Toward a Theory of Hierarchical Definability (THD) Modelling Biological Phenomena with Rough Sets Database Mining on Derived Attributes Introduction Main Idea Some Basics Two Views of Derived Attributes A Numerical Example Table of Derived Attributes Derived Attributes -- Features Extraction and Construction New Attributes and Attribute Transformations Derived Attributes and Extension Function Dependency Traditional Feature Extractions and Constructions Derived Attributes in the Canonical Model Patterns of Isomorphic Relations Isomorphisms of Attributes and Relations An Example Canonical Models and Interpretations Examining a Single Attribute Full Table Universal Model and Feature Completion The Relation Lattice -- Non Standard Rough Set Theory Completing the Relation Lattice Data Mining with Background Knowledge Conclusions Appendix -- Invisible Patterns Appendix -- Attribute-Oriented Generalization and Concept Hierarchies Appendix -- Functions and Maps References A Proposed Evolutionary, Self-Organizing Automaton for the Control of Dynamic Systems 1 Introduction 2 The Simple Boxes Signature Table Controller 2.1 Overall Performance Metric 2.2 The Importance of State Boundaries 2.3 Unique State Number Allocation 2.4 Run-Time Data 2.5 Statistical Database Update 2.6 Decision Strength Determination 3 Further Use of the Statistical Data 3.1 Winner-Takes-All Voting 3.2 Statistical Strength Aggregation 4 Self Organization Post Œprocessing: Initial Condition 5 Conclusions and Future Research References Rough Set Analysis of Preference-Ordered Data How Prior Knowledge Influences Knowledge Discovery? How Prior Knowledge about Preference Order in Data Influences the Granular Computing? Dominance-Based Rough Set Approach (DRSA) Granular Computing with Dominance Cones Induction of Decision Rules Extensions of DRSA Dealing with Preference-Ordered Data Rough Approximation of Preference Relations and Decision Rule Preference Model Missing Values of Attributes and Criteria Fuzzy Set Extension of DRSA DRSA for Decisions under Risk Hierarchical Structure of Attributes and Criteria Induction of Association Rules with Prior Knowledge of Type (i) and (iii) Conclusions References Fuzzy Sets, Multi-valued Mappings, and Rough Sets Investigating the Choice of l and u Values in the Extended Variable Precision Rough Sets Model 1 Introduction 2 Brief Description of Extended VPRS 3 Description of (l, u)-Quality Graph 4 (l, u) Weighted Lines 5 Identifying a Choice of l and u Values 6 Conclusion References A Quantitative Analysis of Preclusivity vs. Similarity Based Rough Approximations Preclusivity Approximation Spaces Quasi $BZ$ Distributive Lattice and Rough Approximations Tolerance Relations Results References Heyting Wajsberg Algebras as an Abstract Environment Linking Fuzzy and Rough Sets Real Unit Interval Fuzzy Sets HW Algebras Rough Approximations Modal Operators from BZ$^{(dM)}$ Distributive Lattices Rough Approximation Spaces in BZ Lattices Rough Approximation Spaces from Information Systems References Dominance-Based Rough Set Approach Using Possibility and Necessity Measures Introduction Rough Approximation by Dominance Relations Possibility and Necessity Measures Approximations by Means of Fuzzy Dominance Relations Conclusion References Generalized Decision Algorithms, Rough Inference Rules, and Flow Graphs Introduction Decision Rules and Decision Algorithms Some Properties of Decision Algorithms Total Probability Theorems and Rough Inference Rules Decision Algorithms and Flow Graphs Conclusions References Generalized Rough Sets and Rule Extraction Introduction Traditional Rough Sets Definitions and Properties Rough Set Analysis of an Information Table Generalization of Rough Sets Distinction among Positive, Negative, and Boundary Elements Approximations by Means of Elementary Sets Rule Extraction Rule Extraction Based on Positive Regions Rule Extraction Based on Lower Approximations Comparison and Correspondence between Definitions and Rules A Simple Example References Towards a Mereological System for Direct Products and Relations Introduction Grids and Mereological Systems Mereological Systems for Direct Products and Relations Final Comment References On the Structure of Rough Approximations Introduction Preliminaries Generalizations of Approximations References Modification of Weights of Conflict Profile's Elements and Dependencies of Attributes in Consensus Model Introduction The Consensus Model Susceptibility to Consensus Modification of Profile' Elements Weights Dependencies of Attributes in Determining Consensus Conclusions References Reasoning about Information Granules Based on Rough Logic 1 Introduction 2 Rough Logic 2.1 Syntax 2.2 Axioms and Rules 2.3 Notation 2.4 Semantics 3 Granules and Granular Computing 4 Logical Reasoning Based on Granular Computing 5 Conclusion References A Rough Set Framework for Learning in a Directed Acyclic Graph 1 Introduction 2 Classi .cation in an Ontology 3 Extensions to Rough Set Theory 3.1 DAG-Decision Systems 3.2 Unknown Set Membership 3.3 Operators 4 A Learning Algorithm 5 Conclusions and Future Research References On Compressible Information Systems Introduction Basic Notions $h$-Compressible Information Systems $psi $-Compressible Information Systems Conclusion References Functional Dependencies in Relational Expressions Based on Or-Sets Introduction Relational databases based on or-sets Framework Restrictions and Supplementary Attribute Values Satisfaction Degrees with Rules Satisfaction Degrees with Rules Accompanied by a Factor Functional Dependencies Formulation of Functional Dependencies Inference Rules of Functional Dependencies Conclusions References On Asymptotic Properties of Rough-Set-Theoretic Approximations. Fractal Dimension, Exact Sets, and Rough Inclusion in Potentially Infinite Information Systems Introduction Rough Set Theory: A Nutshell Account Fractal Dimensions Metrics on Rough Sets Fractal Dimension $dim_{@mathcal {A}}$ ${@mathcal {A}}$--Exact Sets Rough Mereology and Dependencies on Infinitary Concepts References About Tolerance and Similarity Relations in Information Systems Introduction Information Systems Tolerance Relation Similarity of Systems Independency with Respect to Similarity References Rough Sets, Guarded Command Language, and Decision Rules Introduction Rough Sets and Guarded Command Language Generating Rough Data from an Explicit Rule Analyzing Rough Data and Expressing Rough Rules Discussion and Concluding Remarks References Collaborative Query Processing in DKS Controlled by Reducts Introduction Distributed Information Systems Query Answering Based on Reducts Conclusion References A New Method for Determining of Extensions and Restrictions of Information Systems Introduction Consistent Extensions and Restrictions Consistent Extensions of Information Systems Consistent Restrictions of Information Systems Strict Consistent Extensions and Restrictions Conclusions References A Logic Programming Framework for Rough Sets Introduction Preliminaries Rough Sets Logic Programs Definite Extended Logic Programs and Rough Relations A Query Language for Rough Relations Extending Rough Relations Conclusions References Attribute Core of Decision Table 1 Introduction 2 Problem in Calculating Attribute Core of a Decision Table 3 Information View of Rough Set 4 Calculating the Attribute Core of a Decision Table ALGORITHM for finding the attribute core of a decision table in the infor-mation view: 5 Conclusion References Signal Analysis Using Rough Integrals Introduction Indistinguishability and Set Approximation Rough Measure Discrete Rough Integral Conclusion References How Much Privacy? - A System to Safe Guard Personal Privacy while Releasing Databases Introduction Related Work System Architecture and Methods Measuring the Privacy Breach Total Cost Model Average Benefit Model External Generalization Conclusion References Rough Clustering: An Alternative to Find Meaningful Clusters by Using the Reducts from a Dataset Introduction and Motivation Proposed Approach Example of Application: The Animal Taxonomy Problem Conclusions and Future Work References Concept Learning with Approximation: Rough Version Spaces Introduction Concept Learning Notations Version Spaces Approximation and Version Spaces Rough Consistency RVS Definition Properties Algorithms and Implementation RVS Approximation Combination Refining RVS Approximation RVS by Hand Conclusion References Variable Consistency Monotonic Decision Trees Introduction Variable-Consistency Dominance-Based Rough Set Approach (VC-DRSA) Variable Consistency Monotonic Decision Trees Architecture of Decision Trees A Procedure for Building Variable Consistency Decision Trees Conclusions References Importance and Interaction of Conditions in Decision Rules Introduction Evaluation of Decision Rules Induced from Examples Main Concepts Related to Analysis of Set Functions Analysis of Confidence of Decision Rules Using Set Indices Example of Using Set Indices Conclusions References Time Complexity of Rough Clustering: GAs versus K-Means Introduction Rough Set Genome and Its Evaluation Adaptation of K-Means to Rough Set Theory Time Requirements of the Two Rough Clustering Techniques Summary and Conclusions References Induction of Decision Rules and Classification in the Valued Tolerance Approach Introduction Basic Concepts of Valued Tolerance Approach Algorithms of Rule Induction Algorithm Inducing All Rules Algorithm Inducing Minimal Set of Rules Strategies for Classifying New Objects Experiments Conclusions References Time Series Model Mining with Similarity-Based Neuro-Fuzzy Networks and Genetic Algorithms: A Parallel Implementation Introduction Problem Formulation A Soft Computing Model Mining Strategy Parallel Implementation Example Conclusions References Closeness of Performance Map Information Granules: A Rough Set Approach Introduction Performance Maps Indistinguishability and Set Approximation Indistinguishability Relation Conclusion References Granular Computing on Binary Relations Introduction Granular Computing on Binary Relations The Induced Rough Computing Symmetric Binary Granulation The Axioms on the Conflict Interests Relations (CIR) A New View of Chinese Wall Security Policy Conclusions References Measures of Inclusion and Closeness of Information Granules: A Rough Set Approach Introduction Indistinguishability and Set Approximation Indistinguishability Relation Sample Set Approximation Rough Membership Set Function Rough Measures Conclusion References Rough Neurocomputing: A Survey of Basic Models of Neurocomputation 1 Introduction 2 Preprocessing with Rough Sets 3Granular Neural Network Architecture 4 Rough Neurons 4.1 Interval-BasedRough Neuron 4.2 Approximation Neurons 5 Hybrid Neural Networks 6 Concluding Remarks References Rough Neurocomputing Based on Hierarchical Classifiers Introduction Information Granule Systems and Parameterized Approximation Spaces Classifiers as Information Granules Approximation Spaces in Rough Neurocomputing Standards, Productions, and $AR$-Schemes Conclusion References Using Granular Objects in Multi-source Data Fusion Multi-source Data Fusion Formulation of Source Support Including Knowledge about Reasonableness Granular Objects as Fused Value References Induction of Classification Rules by Granular Computing Introduction A Granular Computing Model Information Tables Measures Associated with Granules Induction of Classification Rules by Searching Granules Consistent Classification Problems Construction of a Granule Network Conclusion References Acquisition Methods for Contextual Weak Independence Introduction Contextual Weak Independence Specification of CWIs by a Human Expert Detecting CWIs in a Conditional Probability Table Conclusion References A Method for Detecting Context-Specific Independence in Conditional Probability Tables Introduction Context-Specific Independence Specification of CSIs by a Human Expert Detecting CSIs in a Conditional Probability Table Conclusion References Properties of Weak Conditional Independence Introduction Object-Oriented Bayesian Networks Weak Conditional Independence Properties of Weak Conditional Independence Conclusion References A Proposal of Probability of Rough Event Based on Probability of Fuzzy Event Introduction Probability of Fuzzy Event Probability of Rough Event Conclusion References Approximate Bayesian Network Classifiers Introduction Probabilities in Information Systems Probabilistic Decision Reducts Entropy-Based Approximations Bayesian Networks BN-Based Classification Related Optimization Problems Experimental Results Conclusions References Accuracy and Coverage in Rough Set Rule Induction Introduction Accuracy and Coverage Definition of Accuracy and Coverage Fundamental Characteristics of Accuracy and Coverage Coverage as Likelihood Intuitive Interpretation Semi-formal Interpretation Statistical Dependence Tradeoff between Accuracy and Coverage MDL Principle Conclusion References Statistical Test for Rough Set Approximation Based on Fisher's Exact Test Introduction From Information Systems to Contingency Tables Accuracy and Coverage Contingency Tables Fisher's Exact Test Log-Likelihood Ratio Test for Rough Set Approximation How to Make a New Table for Comparison Example Discussion: Negative Values in Cells Conclusion References Triangulation of Bayesian Networks: A Relational Database Perspective Introduction Triangulation in Bayesian Networks Construction of Acyclic Database Schemes Triangulation from a Relational Database Perspective Conclusion References A New Version of Rough Set Exploration System Introduction Basic Notions Contents of RSES v. 2.0 Input/Output Formats The Algorithms The RSES GUI New Features Case Study Generation of New Attributes Experimental Results Perspective References Local Attribute Value Grouping for Lazy Rule Induction 1 Introduction 2 Preliminaries 2.1 Minimal Rule Induction 2.2 Discretisation and Value Partitioning 2.3 Lazy Rule Induction 3 Lazy Rule Induction with Attribute alue Grouping 3.1 Metrics for Attribute Value Grouping and Example 3.2 Relation to Discretisation 4 Conclusions and Further Research References Incomplete Data Decomposition for Classification Introduction Preliminaries Method Description Decomposition Criteria Empirical Evaluation Conclusions References Extension of Relational Management Systems with Data Mining Capabilities Introduction Knowledge System Definition of Operator Conclusions References Reducing Number of Decision Rules by Joining Introduction System of Representatives Rule Induction Splitting the Set of Decision Rules into Groups Joining Rules Algorithm Parameters and Complexity Results of Experiments Tested Data System of Representatives The Discussion of the Results Conclusions References Scalable Classification Method Based on Rough Sets Introduction Preliminaries Classification Problem Rough Sets and Classification Problems Lazy Learning Lazy Learning for Rough Sets Methods Example Concluding Remarks References Parallel Data Mining Experimentation Using Flexible Configurations Introduction Distributed Data Mining Systems and Algorithms Flexible Optimization Environment MOIRAE Architecture MOIRAE Components MOIRAE Architecture Model MOIRAE Control Model Association Rules: Distributed Generalized Calculation Algorithm Family Implementation Experimental Results Conclusions References An Optimization of Apriori Algorithm through the Usage of Parallel I/O and Hints Introduction Association Rules Calculation Using Parallel I/O Problem Scenario I/O Model Apriori Algorithm: A Case of Hints Optimization Apriori Algorithm Evaluation Conclusions and Future Work References Patterns in Information Maps Introduction Preliminaries Information Maps Exemplary Problems Conclusions References Discernibility Matrix Approach to Exception Analysis Introduction Text and Text Measurement Discernibility Matrix and Core Exception Analysis Conclusion References Gastric Cancer Data Mining with Ordered Information Introduction Ordered Information Tables Mining Ordering Rules in Gastric Cancer Data Granular Computing with Ordered Information Forming Granules of Condition Attributes in Post-processing Forming Granules of Symbolic Data in Pre-processing Conclusions References A Granular Approach for Analyzing the Degree of Affability of a Web Site Introduction Methodology to Calculate Web Pages Utility Example Conclusions References Comparison of Classification Methods for Customer Attrition Analysis 1 Introduction 2 Business Problem 2.1 Data Preprocessing Goals Time Series "Unrolling" and Target Field Definition 2.2 Data Premodeling 3 Model Development Process 4 Conclusion References User Profile Model: A View from Artificial Intelligence Introduction Formalization of User Profiles Decision Model Summary References Mining the Client's Life Cycle Behaviour in the Web Introduction Related Work CRM in E-commerce Reasons Why E-commerce Has Failed to Establish Effective CRM Data Sources for E-CRM A Client's Life Cycle Model Based on a Neural Network Data Description The Client Model Exploiting the Life Cycle Model Conclusions References PagePrompter: An Intelligent Web Agent Created Using Data Mining Techniques Introduction Background and Related Research The PagePrompter System Knowledge Discovery from Web Log and Web Structures Association Rules Web Page Clustering Based on Website Structure Conclusion References VPRSM Approach to WEB Searching Introduction Document and Query Representation Rough Set-Based Document Ranking Rough Relations Approximate Document Ranking VPRSM-Based Document Ranking Basic Definitions Adapting VPRSM to Document Indexing Final Remarks References Rough Set Approach to the Survival Analysis Introduction Rough Set Framework Medical Data Survival Analysis The Kaplan-Meier Product-Limit Estimate The Cox's Proportional Hazard Model Prognostic Rules Cross-Decision Rules Conclusions References The Identification of Low-Paying Workplaces: An Analysis Using the Variable Precision Rough Sets Model 1 Introduction 2 2 Discussion of Problem and Data 3 Criteria in ‚Leave n OutTM VPRS Analysis 4 Results of VPRS Analysis 5 Comparison with Other Methods 6 Conclusion References A Search for the Best Data Mining Method to Predict Melanoma Introduction Melanoma Data Discretization, Rule Induction, Validation Experiments Conclusions References Towards the Classification of Musical Works: A Rough Set Approach Introduction Attributes of Selected Music Works Rough Set Methods Music Classification Methods Decision Tree Classification Method Non-discretization Rule-Based Classification Method Musical Composition Prediction Algorithm with Discretization Experimental Results Concluding Remarks References Segmentation of Medical Images Based on Approximations in Rough Set Theory Introduction Preliminary Rough Representation of a ROI Single Knowledge Multiple Knowledge Implementation of Rough ROI Representation on Medical Images Conclusions References Adaptive Robust Estimation for Filtering Motion Vectors Introduction Extraction of Motion Vectors Filtering of Motion Vectors Experimental Results and Discussions References Rough Set Feature Selection and Diagnostic Rule Generation for Industrial Applications Introduction Feature Selection and Rule Generation for Industrial Diagnostic Systems Feature Pre-processing and Pre-selection Rough Set Feature Selection and Rule Generation Industrial Chiller Application Conclusions References $lambda $-Connected Approximations for Rough Sets Introduction Basic Concept of Rough Sets and $lambda $-Connectedness $lambda $-Connected Representation for General Rough Sets $lambda $-Connected Approximations Normal $lambda $-Connected Sets and Rough Sets Rough Sets in the Real World Maximum Connectivity Spanning Tree and Rough Sets References Adaptive Classifier Construction: An Approach to Handwritten Digit Recognition Introduction Structural OCR Basics Relational Graph Model for Handwritten Digits Base Skeleton Graph Construction Graph Similarity Measures Adaptive Construction of Distance Functions Dissimilarity Measures Results of Experiments Conclusion References The Application of Support Diagnose in Mitochondrial Encephalomyopathies Introduction Aim of the Work Selection of an Information Method for Support Diagnose in MEM The Project of the Application of Support Decision Making Machine Learning Classification under Uncertainty The Selection of the Set of Attributes Quality of System Classification Conclusion References Obstacle Classification by a Line-Crawling Robot: A Rough Neurocomputing Approach Introduction Basic Features of Line-Crawling Robot LCR Navigation Problem LCR Neural Classification System Rough Inclusion Discrete Rough Integral Convex Sets Rough Neurocomputing by Robot Basic Architecture of RNN Experimental Results Concluding Remarks References Rough Neural Network for Software Change Prediction Introduction Rough Set Preliminaries Rough Inclusion Discrete Rough Integral Convex Sets Rough Neural Network Software Change Prediction Preprocessing VME Metric Data Neural Prediction Experiments Conclusion References Handling Spatial Uncertainty in Binary Images: A Rough Set Based Approach Introduction Rough Sets on Image Spaces A Framework for Using Rough Sets in Handling Spatial Uncertainty Deriving a Matching Template A Simple Application Conclusions References Evolutionary Algorithms and Rough Sets-Based Hybrid Approach to Classificatory Decomposition of Cortical Evoked Potentials Introduction Bayesian Motivated Model Evolutionary Algorithm for Proposed Sparse Coding Rough Sets-Based Selection of Classification-Relevant Components from a Potentially Overcomplete Set of Basis Functions Experiments and Results Data Analysis Conclusions References Rough Mereological Localization and Navigation Introduction Le{{accent 19 s}}niewski's Systems -- Ontology and Mereology Ontology Mereology Rough Mereology Spatial Description Based on Rough Mereology From Linguistic to Quantity Spatial Description by Rough Mereology Rough Mereology in Mobile Robot Localization and Navigation -- Examples System Description Tasks of Localization and Navigation References Author Index This book constitutes the refereed proceedings of the Third International Conference on Rough Sets and Current Trends in Computing, RSCTC 2002, held in Malvern, PA, USA in October 2002. The 76 revised regular papers and short communications presented together with 2 keynotes and 5 plenary papers were carefully reviewed and selected from more than 100 submissions. The book offers topical sections on foundation and methods; granular and neural computing; probabilistic reasoning; data mining, machine learning and pattern recognition; Web mining; and applications