Classification Inductive Rule Learning with Negated Features.
Abstract. This paper reports on an investigation to compare a number of strategies to include negated features within the process of Inductive Rule Learning (IRL). The emphasis is on generating the negation of features while rules are being “learnt”; rather than including (or deriving) the negation of all features as part of the input. Eight different strategies are considered based on the manipulation of three feature sub-spaces. Comparisons are also made with Associative Rule Learning (ARL) in the context of multi-class text classification. The results indicate that the option to include negated features within the IRL process produces more effective classifiers. Key words: Rule Learning, Negation, Multi-class Text Classification 1
Frequent Pattern Trend Analysis in Social Networks.
Abstract. This paper describes an approach to identifying and comparing frequent pattern trends in social networks. A frequent pattern trend is defined as a sequence of time-stamped occurrence (support) values for specific frequent patterns that exist in the data. The trends are generated according to epochs. Therefore, trend changes across a sequence epochs can be identified. In many cases, a great many trends are identified and difficult to interpret the result. With a combination of constraints, placed on the frequent patterns, and clustering and cluster analysis techniques, it is argued that analysis of the result is enhanced. Clustering technique uses a Self Organising Map approach to produce a sequence of maps, one per epoch. These maps can then be compared and the movement of trends identified. This Frequent Pattern Trend Mining framework has been evaluated using two non-standard types of social networks, the cattle movement network and the insurance quote network.
Finding Frequent Subgraphs in Longitudinal Social Network Data Using a Weighted Graph Mining Approach.


Best Clustering Configuration Metrics Towards Multiagent Based Clustering.


Region of Interest Based Image Categorization.

Multi%2DAgent Based Clustering Towards Generic Multi%2DAgent Data Mining.
Abstract. A framework for Multi Agent Data Mining (MADM) is described. The framework comprises a collection of agents cooperating to address given data mining tasks. The fundamental concept underpinning the framework is that it should support generic data mining. The vision is that of a system that grows in an organic manner. The central issue to facilitating this growth is the communication medium required to support agent interaction. This issue is partly addressed by the nature of the proposed architecture and partly through an extendable ontology; both are described. The advantages offered by the framework are illustrated in this paper by considering a clustering application. The motivation for the latter is that no “best ” clustering algorithm has been identified, and consequently an agent-based approach can be adopted to identify “best ” clusters. The application serves to demonstrates the full potential of MADM. Keywords: Multi Agent Data Mining, Agent-Based Clustering.
Frequent Sub%2Dgraph Mining on Edge Weighted Graphs.
Abstract. Frequent sub-graph mining entails two significant overheads. The first is concerned with candidate set generation. The second with isomorphism checking. These are also issues with respect to other forms of frequent pattern mining but are exacerbated in the context of frequent sub-graph mining. To reduced the search space, and address these twin overheads, a weighted approach to sub-graph mining is proposed. However, a significant issue in weighted sub-graph mining is that the antimonotone property, typically used to control candidate set generation, no longer holds. This paper examines a number of edge weighting schemes; and suggests three strategies for controlling candidate set generation. The three strategies have been incorporated into weighted variations of gSpan: ATW-gSpan, AW-gSpan and UBW-gSpan respectively. A complete evaluation of all three approaches is presented.
Multi%2DAgent Based Clustering Towards Generic Multi%2DAgent Data Mining.
Abstract. A framework for Multi Agent Data Mining (MADM) is described. The framework comprises a collection of agents cooperating to address given data mining tasks. The fundamental concept underpinning the framework is that it should support generic data mining. The vision is that of a system that grows in an organic manner. The central issue to facilitating this growth is the communication medium required to support agent interaction. This issue is partly addressed by the nature of the proposed architecture and partly through an extendable ontology; both are described. The advantages offered by the framework are illustrated in this paper by considering a clustering application. The motivation for the latter is that no “best ” clustering algorithm has been identified, and consequently an agent-based approach can be adopted to identify “best ” clusters. The application serves to demonstrates the full potential of MADM. Keywords: Multi Agent Data Mining, Agent-Based Clustering.
Hybrid DIAAF%2FRS Statistical Textual Feature Selection for Language%2DIndependent Text Classification.

Finding Temporal Patterns in Noisy Longitudinal Data A Study in Diabetic Retinopathy.

Trend Mining in Social Networks A Study Using a Large Cattle Movement Database.

Web%2DSite Boundary Detection.
Abstract. Defining the boundaries of a web-site, for (say) archiving or information retrieval purposes, is an important but complicated task. In this paper a web-page clustering approach to boundary detection is suggested. The principal issue is feature selection, hampered by the observation that there is no clear understanding of what a web-site is. This paper proposes a definition of a web-site, founded on the principle of user intention, directed at the boundary detection problem; and then reports on a sequence of experiments, using a number of clustering techniques, and a wide range of features and combinations of features to identify web-site boundaries. The preliminary results reported seem to indicate that, in general, a combination of features produces the most appropriate result. Keywords: Web-site definition, Web-page Clustering, Web Archiving. 1
Detecting Temporal Pattern and Cluster Changes in Social Networks A Study Focusing UK Cattle Movement Database.

Detecting Temporal Pattern and Cluster Changes in Social Networks A Study Focusing UK Cattle Movement Database.

Rule Learning with Negation Issues Regarding Effectiveness.
ABSTRACT: An investigation of rule learning processes that allow the inclusion of negated features is described. The objective is to establish whether the use of negation in inductive rule learning systems is effective with respect to classification. This paper seeks to answer this question by considering two issues relevant to such systems; feature identification and rule refinement. Both synthetic and real datasets are used to illustrate solutions to the identified issues and to demonstrate that the use of negative features in inductive rule learning systems is indeed beneficial.
Text classification using graph mining%2Dbased feature extraction.
Abstract A graph-based approach to document classification is described in this paper. The graph representation offers the advantage that it allows for a much more expressive document encoding than the more standard bag of words/phrases approach, and consequently gives an improved classification accuracy. Document sets are represented as graph sets to which a weighted graph mining algorithm is applied to extract frequent subgraphs, which are then further processed to produce feature vectors (one per document) for classification. Weighted subgraph mining is used to ensure classification effectiveness and computational efficiency; only the most significant subgraphs are extracted. The approach is validated and evaluated using several popular classification algorithms together with a real world textual data set. The results demonstrate that the approach can outperform existing text classification algorithms on some dataset. When the size of dataset increased, further processing on extracted frequent features is essential.
A sliding windows based dual support framework for discovering emerging trends from temporal data.
(DSAT) algorithm. This is a novel technique for discovering Jumping Emerging Patterns (JEPs) from time series data using a sliding window technique. Our approach is particularly effective when performing trend analysis in order to explore the itemset variations over time. Our proposed framework is different from the previous work on JEP in that we do not rely on itemsets borders with a constrained search space. DSAT exploits previously mined time stamped data by using a sliding window concept, thus requiring less memory, minimum computational cost and very low dataset accesses. DSAT discovers all JEPs, as in “naïve” approaches, but utilises less memory and scales linearly with large datasets sets as demonstrated in the experimental section. 1
Corpus callosum MR image classification.
Abstract An approach to classifying Magnetic Resonance (MR) image data is described. The specific application is the classification of MRI scan data according to the nature of the corpus callosum, however the approach has more general applicability. A variation of the “spectral segmentation with multi-scale graph decomposition” mechanism is introduced. The result of the segmentation is stored in a quad-tree data structure to which a weighted variation (also developed by the authors) of the gSpan algorithm is applied to identify frequent sub-trees. As a result the images are expressed as a set frequent sub-trees. There may be a great many of these and thus a decision tree based feature reduction technique is applied before classification takes place. The results show that the proposed approach performs both efficiently and effectively, obtaining a classification accuracy of over 95 % in the case of the given application. 1

A Hybrid Statistical Data Pre%2Dprocessing Approach for Language%2DIndependent Text Classification.
Data pre-processing is an important topic in Text Classification (TC). It aims to convert the original textual data in a data-mining-ready structure, where the most significant text-features that serve to differentiate between textcategories are identified. Broadly speaking, textual data pre-processing techniques can be divided into three groups: (i) linguistic, (ii) statistical, and (iii) hybrid (i)  &amp; (ii). With regard to language-independent TC, our study relates to the statistical aspect only. The nature of textual data pre-processing includes:

Multi%2DParty Argument from Experience.
Abstract. A framework, PISA, for conducting dialogues to resolve disputes concerning the correct categorisation of particular cases, is described. Unlike previous systems to conduct such dialogues, which have typically involved only two agents, PISA allows any number of agents to take part, facilitating discussion of cases which permit many possible categorizations. A particular feature of the framework is that the agents argue directly from individual repositories of experiences rather than from a previously engineered knowledge base, as is the usual case, and so the knowledge engineering bottleneck is avoided. Argument from experience is enabled by real time data-mining conducted by individual agents to find reasons to support their viewpoints, and critique the arguments of other parties. Multiparty dialogues raise a number of significant issues, necessitating appropriate design choices. The paper describes how these issues were resolved in the protocol and implemented in PISA, and illustrates the operation of PISA using an example based of a dataset relating to nursery provision. Finally some experiments comparing PISA
Arguing from Experience to Classifying Noisy Data.
Abstract. A process, based on argumentation theory, is described for classifying very noisy data. More specifically a process founded on a concept called “arguing from experience ” is described where by several software agents “argue ” about the classification of a new example given individual “case bases ” containing previously classified examples. Two “arguing from experience ” protocols are described: PADUA which has been applied to binary classification problems and PISA which has been applied to multi-class problems. Evaluation of both PADUA and PISA indicates that they operate with equal effectiveness to other classification systems in the absence of noise. However, the systems out-perform comparable systems given very noisy data. Keywords: Classification, Argumentation, Noisy data 1.
A Generic and Extendible Multi%2DAgent Data Mining Framework.
Abstract. A generic and extendible Multi-Agent Data Mining (MADM)
Integrating Data Mining and Agent Based Modeling and Simulation.
Abstract. In this paper, we introduce an integration study which combines Data Mining (DM) and Agent Based Modeling and Simulation (ABMS). This study, as a new paradigm for DM/ABMS, is concerned with two approaches: (i) applying DM techniques in ABMS investigation, and inversely (ii) utilizing ABMS results in DM research. Detailed description of each approach is presented in this paper. A conclusion and the future work of this (integration) study are given at the end.
Application of Classification Association Rule Mining for Mammalian Mesenchymal Stem Cell Differentiation.
Abstract. In this paper, data mining is used to analyze the differentiation of mammalian Mesenchymal Stem Cells (MSCs). A database comprising the key parameters which, we believe, influence the destiny of mammalian MSCs has been constructed. This paper introduces Classification Association Rule Mining (CARM) as a data mining technique in the domain of tissue engineering and initiates a new promising research field. The experimental results show that the proposed approach performs well with respect to the accuracy of (classification) prediction. Moreover, it was found that some rules mined from the constructed MSC database are meaningful and useful.
Text Classification using Graph Mining%2Dbased Feature Extraction.
Abstract A graph-based approach to document classification is described in this paper. The graph representation offers the advantage that it allows for a much more expressive document encoding than the more standard bag of words/phrases approach, and consequently gives an improved classification accuracy. Document sets are represented as graph sets to which a weighted graph mining algorithm is applied to extract frequent subgraphs, which are then further processed to produce feature vectors (one per document) for classification. Weighted subgraph mining is used to ensure classification effectiveness and computational efficiency; only the most significant subgraphs are extracted. The approach is validated and evaluated using several popular classification algorithms together with a real world textual data set. The results demonstrate that the approach can outperform existing text classification algorithms on some dataset. When the size of dataset increased, further processing on extracted frequent features is essential.
Corpus Callosum MR Image Classification.
Abstract An approach to classifying Magnetic Resonance (MR) image data is described. The specific application is the classification of MRI scan data according to the nature of the corpus callosum, however the approach has more general applicability. A variation of the “spectral segmentation with multi-scale graph decomposition” mechanism is introduced. The result of the segmentation is stored in a quad-tree data structure to which a weighted variation (also developed by the authors) of the gSpan algorithm is applied to identify frequent sub-trees. As a result the images are expressed as a set frequent sub-trees. There may be a great many of these and thus a decision tree based feature reduction technique is applied before classification takes place. The results show that the proposed approach performs both efficiently and effectively, obtaining a classification accuracy of over 95 % in the case of the given application. 1
A Sliding Windows based Dual Support Framework for Discovering Emerging Trends from Temporal Data.
(DSAT) algorithm. This is a novel technique for discovering Jumping Emerging Patterns (JEPs) from time series data using a sliding window technique. Our approach is particularly effective when performing trend analysis in order to explore the itemset variations over time. Our proposed framework is different from the previous work on JEP in that we do not rely on itemsets borders with a constrained search space. DSAT exploits previously mined time stamped data by using a sliding window concept, thus requiring less memory, minimum computational cost and very low dataset accesses. DSAT discovers all JEPs, as in “naïve” approaches, but utilises less memory and scales linearly with large datasets sets as demonstrated in the experimental section. 1
PADUA a protocol for argumentation dialogue using association rules.

Analysis of Mesenchymal Stem Cell Differentiation in vitro Using Classification Association Rule Mining.

EMADS An extendible multi%2Dagent data miner.
In this paper we describe EMADS, an Extendible Multi-Agent Data mining System. The EMADS vision is that of a community of data mining agents, contributed by many individuals, interacting under decentralised control to address data mining requests. EMADS is seen both as an end user application and a research tool. This paper details the EMADS vision, the associated conceptual framework and the current implementation. Although EMADS may be applied to many data mining tasks; the study described here, for the sake of brevity, concentrates on agent based data classification. A full description of EMADS is presented.
Improved methods for extracting frequent itemsets from interim%2Dsupport trees.


Research and Development in Intelligent Systems XXIV
A multiset (also referred to as a bag) is a set (collection of elements where the order is of no importance), where the elements do not need to be unique. A vase with n blue and m red marbles is a multiset for example. We propose a new class of distance measures (metrics) designed for multisets, both of which are a recurrent theme in many data mining [2] applications. One particular instance of this class originated from

Document%2DBase Extraction for Single%2DLabel Text Classification.
Abstract. Many text mining applications, especially when investigating Text Classification (TC), require experiments to be performed using common textcollections, such that results can be compared with alternative approaches. With regard to single-label TC, most text-collections (textual data-sources) in their original form have at least one of the following limitations: the overall volume of textual data is too large for ease of experimentation; there are many predefined classes; most of the classes consist of only a very few documents; some documents are labeled with a single class whereas others have multiple classes; and there are documents found with little or no actual text-content. In this paper, we propose a standard approach to automatically extract “qualified” document-bases from a given textual data-source that can be used more effectively and reliably in single-label TC experiments. The experimental results demonstrate that document-bases extracted based on our approach can be used effectively in single-label TC experiments.
A Weighted Utility Framework for Mining Association Rules.
Association rule mining (ARM) identifies frequent itemsets from databases and generates association rules by assuming that all items have the same significance and frequency of occurrence in a record i.e. their weight and utility is the same (weight=1 and utility=1) which is not always the case. However, items are actually different in many aspects in a number of real applications such as retail marketing, nutritional pattern mining etc. These differences between items may have a strong impact on decision making in many application unlike the use of standard ARM. Our framework, Weighted Utility ARM (WUARM), considers the varied significance and different frequency values of individual items as their weights and utilities. Thus, weighted utility mining focuses on identifying the itemsets with weighted utilities higher than the user specified weighted utility threshold. We conduct experiments on synthetic and real data sets using standard ARM, weighted ARM and Weighted Utility ARM (WUARM) and present analysis of the results. 1.
On Extraction of Nutritional Patterns %28NPS%29 Using Fuzzy Association Rule Mining.
Abstract. This paper proposes a framework for mining market basket data to generate Nutritional Patterns (NPs) and a method for analysing generated nutritional patterns using Fuzzy Association Rule Mining. Edible attributes are filtered from transactional input data by projections and are then converted to Required Daily Allowance (RDA) numeric values. The RDA database is then converted to a fuzzy database that contains expended normalized fuzzy attributes comprising of different fuzzy sets. Analysis of nutritional information is performed either from normal generated association rules or from a converted fuzzy transactional database. Our approach uses prototype support tool that extract Nutritional Patterns (NPs) and signifies the level of nutritional content in an association rule per item. The paper presents various performance tests and interestingness measures to demonstrate the effectiveness of the approach and concludes with experimental results and discussion on evaluating the proposed framework.
Weighted Association Rule Mining from Binary and Fuzzy Data.
Abstract. A novel approach is presented for mining weighted association rules (ARs) from binary and fuzzy data. We address the issue of invalidation of downward closure property (DCP) in weighted association rule mining where each item is assigned a weight according to its significance w.r.t some user defined criteria. Most works on weighted association rule mining so far struggle with invalid downward closure property and some assumptions are made to validate the property. We generalize the weighted association rule mining problem for databases with binary and quantitative attributes with weighted settings. Our methodology follows an Apriori approach [9] and avoids pre and post processing as opposed to most weighted association rule mining algorithms, thus eliminating the extra steps during rules generation. The paper concludes with experimental results and discussion on evaluating the proposed approach.
Mining Allocating Patterns in One%2DSum Weighted Items.
An Association Rule (AR) is a common knowledge model in data mining that describes an implicative cooccurring relationship between two disjoint sets of binary-valued transaction database attributes (items), expressed in the form of an “antecedent ⇒ consequent ” rule. A variant of the AR is the Weighted Association Rule (WAR). With regard to a marketing context, this paper introduces a new knowledge model in data mining ⎯ ALlocating Pattern (ALP). An ALP is a special form of WAR, where each rule item is associated with a weighting score between 0 and 1, and the sum of all rule item scores is 1. It can not only indicate the implicative co-occurring relationship between two (disjoint) sets of items in a weighted setting, but also inform the “allocating ” relationship among rule items. ALPs can be demonstrated to be applicable in marketing and possibly a surprising variety of other areas. We further propose an Apriori based algorithm to extract hidden and interesting ALPs from a “one-sum ” weighted transaction database. The experimental results show the effectiveness of the proposed algorithm. 1.
Agent Based Frequent Set Meta Mining Introducing EMADS.
mining System, to support the dynamic creation of communities of data mining agents; explore the capabilities of such agents and demonstrate (by experiment) their application to data mining on distributed data. Although, EMADS is not restricted to one data mining task, the study described here, for the sake of brevity, concentrates on agent based Association Rule Mining (ARM), in particular what we refer to as frequent set meta mining (or Meta ARM). A full description of our proposed Meta ARM model is presented where we describe the concept of Meta ARM and go on to describe and analyse a number of potential solutions in the context of EMADS. Experimental results are considered in terms of: the number of data sources, the number of records in the data sets and the number of attributes represented.
Mining Fuzzy Association Rules from Composite Items.
Abstract. A novel framework is described for mining fuzzy Association Rules (ARs) relating the properties of composite attributes, i.e. attributes or items that each feature a number of values derived from a common schema. To apply fuzzy Association Rule Mining (ARM) we partition the property values into fuzzy property sets. This paper describes: (i) the process of deriving the fuzzy sets (Composite Fuzzy ARM or CFARM) and (ii) a unique property ARM algorithm founded on the correlation factor interestingness measure. The paper includes a complete analysis, demonstrating: (i) the potential of fuzzy property ARs, and (ii) that a more succinct set of property ARs (than that generated using a non-fuzzy method) can be produced using the proposed approach.
Fuzzy Weighted Association Rule Mining with Weighted Support and Confidence Framework.
In this paper we extend the problem of mining weighted association rules. A classical model of boolean and fuzzy quantitative association rule mining is adopted to address the issue of invalidation of downward closure property (DCP) in weighted association rule mining where each item is assigned a weight according to its significance w.r.t some user defined criteria. Most works on DCP so far struggle with invalid downward closure property and some assumptions are made to validate the property. We generalize the problem of downward closure property and propose a fuzzy weighted support and confidence framework for boolean and quantitative items with weighted settings. The problem of invalidation of the DCP is solved using an improved model of weighted support and confidence framework for classical and fuzzy association rule mining. Our methodology follows an Apriori algorithm approach and avoids pre and post processing as opposed to most weighted ARM algorithms, thus eliminating the extra steps during rules generation. The paper concludes with experimental results and discussion on evaluating the proposed framework.
Argument Based Moderation of Benefit Assessment.

Fuzzy Weighted Association Rule Mining with Weighted Support and Confidence Framework.
In this paper we extend the problem of mining weighted association rules. A classical model of boolean and fuzzy quantitative association rule mining is adopted to address the issue of invalidation of downward closure property (DCP) in weighted association rule mining where each item is assigned a weight according to its significance w.r.t some user defined criteria. Most works on DCP so far struggle with invalid downward closure property and some assumptions are made to validate the property. We generalize the problem of downward closure property and propose a fuzzy weighted support and confidence framework for boolean and quantitative items with weighted settings. The problem of invalidation of the DCP is solved using an improved model of weighted support and confidence framework for classical and fuzzy association rule mining. Our methodology follows an Apriori algorithm approach and avoids pre and post processing as opposed to most weighted ARM algorithms, thus eliminating the extra steps during rules generation. The paper concludes with experimental results and discussion on evaluating the proposed framework.
A Framework for Mining Fuzzy Association Rules from Composite Items.
Abstract. A novel framework is described for mining fuzzy Association Rules (ARs) relating the properties of composite attributes, i.e. attributes or items that each feature a number of values derived from a common schema. To apply fuzzy Association Rule Mining (ARM) we partition the property values into fuzzy property sets. This paper describes: (i) the process of deriving the fuzzy sets (Composite Fuzzy ARM or CFARM) and (ii) a unique property ARM algorithm founded on the correlation factor interestingness measure. The paper includes a complete analysis, demonstrating: (i) the potential of fuzzy property ARs, and (ii) that a more succinct set of property ARs (than that generated using a non-fuzzy method) can be produced using the proposed approach.
Mining Efficiently Significant Classification Association Rules.

Hybrid Rule Ordering in Classification Association Rule Mining.
Abstract. Classification Association Rule Mining (CARM) is an approach to classifier generation that builds an Association Rule Mining based classifier using Classification Association Rules (CARs). Regardless of which particular CARM algorithm is used, a similar set of CARs is always generated from data, and a classifier is usually presented as an ordered list of CARs, based on a selected rule ordering strategy. Hence to produce an accurate classifier, it is essential to develop a rational rule ordering mechanism. In the past decade, a number of rule ordering strategies have been introduced. Six major ones can be identified: Confidence Support &amp; size-of-Antecedent (CSA), size-of-
PADUA Protocol Strategies and Tactics.

A Novel Rule Weighting Approach in Classification Association Rule Mining.

A Novel Rule Ordering Approach in Classification Association Rule Mining.
Abstract. A Classification Association Rule (CAR), a common type of mined knowledge in Data Mining, describes an implicative co-occurring relationship between a set of binary-valued data-attributes (items) and a pre-defined class, expressed in the form of an “antecedent ⇒ consequent-class ” rule. Classification Association Rule Mining (CARM) is a recent Classification Rule Mining (CRM) approach that builds an Association Rule Mining (ARM) based classifier using CARs. Regardless of which particular methodology is used to build it, a classifier is usually presented as an ordered CAR list, based on an applied rule ordering strategy. Five existing rule ordering mechanisms can be identified: (1) Confidence-Support-size_of_Antecedent (CSA), (2) size_of_Antecedent-Confidence-Support (ACS), (3) Weighted Relative Accuracy (WRA), (4) Laplace Accuracy, and (5) χ 2 Testing. In this paper, we divide the above mechanisms into two groups: (i) pure “support-confidence ” framework like, and (ii) additive score assigning like. We consequently propose a hybrid rule ordering approach by combining one approach taken from (i) and another approach taken from (ii). The experimental results show that the proposed rule ordering approach performs well with respect to the accuracy of classification.
Statistical Identification of Key Phrases for Text Classification.
Abstract. Algorithms for text classification generally involve two stages, the first of which aims to identify textual elements (words and/or phrases) that may be relevant to the classification process. This stage often involves an analysis of the text that is both language-specific and possibly domain-specific, and may also be computationally costly. In this paper we examine a number of alternative keyword-generation methods and phrase-construction strategies that identify key words and phrases by simple, language-independent statistical properties. We present results that demonstrate that these methods can produce good classification accuracy, with the best results being obtained using a phrase-based approach. Keywords: Text Classification, Text Preprocessing. 1
Frequent Set Meta Mining Towards Multi%2DAgent Data Mining.

Dynamic Rule Mining for Argumentation Based Systems.
Argumentation has proved to be a very influential reasoning mechanism particularly in the context of multi agent systems. In this paper we introduce PADUA (Protocol for Argumentation Dialogue Using Association Rules), a novel argumentation formalism that dynamically mines Association Rules (ARs) from the case background as a means to: (i) generate the arguments exchanged among dialogue participants, and (ii) represent each participant’s background domain knowledge, thus avoiding the traditional knowledge base representations. Dialogue participants mine ARs from their own case data and then use these rules as arguments and counter arguments. This paper fully describes the PADUA formalism and proposes a suite of dynamic ARM algorithms to provide support for the argumentation process.
The effect of threshold values on association rule based classification accuracy.
Classification Association Rule Mining (CARM) systems operate by applying an Association Rule Mining (ARM) method to obtain classification rules from a train-ing set of previously-classified data. The rules thus generated will be influenced by the choice of ARM parameters employed by the algorithm (typically support and confidence threshold values). In this paper we examine the effect that this choice has on the predictive accuracy of CARM methods. We show that the accuracy can almost always be improved by a suitable choice of parameters, and describe a hill-climbing method for finding the best parameter settings. We also demonstrate that the proposed hill-climbing method is most effective when coupled with a fast CARM algorithm such as the TFPC algorithm which is also described.
Tree%2Dbased partitioning of date for association rule mining.

The Knowledge Bazaar.
As a result of a general dissatisfaction with transmission models of learning and theories of cognition that preclude context, &amp;quot;constructivism, &amp;quot; or perhaps more accurately, a constructivist metatheory drawn from many sources, has come to dominate contemporary educational psychology. One of constructivism&#039;s principal implications, according to Sack, Soloway and Weingrad [9], is that &amp;quot;...activities of knowledge production (e.g., science) and reproduction (e.g., education) are about convincing, recruiting and enculturating others. in short a constructivist analysis of knowledge foregrounds rhetoric: the powers of persuasion and the difficulties of dispute &amp;quot; (p. 357). This rhetoricized conception of knowledge construction, we would argue, has not been adequately reflected in the work of educational technologists working from constructivist theory. Subtitled “A Rhetorical Approach to Constructivist Learning, ” Reality Check is a prototype of educational software that we are currently developing to embody many of the lessons we believe knowledge negotiation offers educators and educational technologists. Using the metaphor of education as “knowledge-negotiation” Reality Check prompts educators and students independently to construct their representation of a task by identifying relevant variables and giving reasons and evidence that permit fruitful interrogation. By setting up alternative representations, Reality Check explicitly invites the teacher and student to participate in a negotiation of knowledge in which evidence is exchanged and positions are clarified, whereby each party in this educational activity can better understand the other and, more importantly, make productive use of this understanding.
Partitioning strategies for distributed association rule mining.
In this paper a number of alternative strategies for distributed/parallel association rule mining are investigated. The methods examined make use of a data structure, the T-tree, introduced previously by the authors as a structure for organising sets of attributes for which support is being counted. We consider six different approaches, representing different ways of parallelising the basic Apriori-T algorithm that we use. The methods focus on different mechanisms for partitioning the data between processes, and for reducing the message-passing overhead. Both ‘horizontal ’ (data distribution) and ‘vertical ’ (candidate distribution) partitioning strategies are considered, including a vertical partitioning algorithm (DATA-VP) which we have developed to exploit the structure of the T-tree. We present experimental results examining the performance of the methods in implementations using JavaSpaces. We conclude that in a JavaSpaces environment, candidate distribution strategies offer better performance than those that distribute the original dataset, because of the lower messaging overhead, and the DATA-VP algorithm produced results that are especially encouraging.
Research and Development in Intelligent Systems XXI

Obtaining Best Parameter Values for Accurate Classification.
In this paper we examine the effect that the choice of support and confidence thresholds has on the accuracy of classifiers obtained by Classification Association Rule Mining. We show that accuracy can almost always be improved by a suitable choice of threshold values, and we describe a method for finding the best values. We present results that demonstrate this approach can obtain higher accuracy without the need for coverage analysis of the training data.
Threshold Tuning for Improved Classification Association Rule Mining.
Abstract. One application of Association Rule Mining (ARM) is to identify Classification Association Rules (CARs) that can be used to classify future instances from the same population as the data being mined. Most CARM methods first mine the data for candidate rules, then prune these using coverage analysis of the training data. In this paper we describe a CARM algorithm that avoids the need for coverage analysis, and a technique for tuning its threshold parameters to obtain more accurate classification. We present results to show this approach can achieve better accuracy than comparable alternatives at lower cost. 1
The Knowledge Bazaar.
As a result of a general dissatisfaction with transmission models of learning and theories of cognition that preclude context, &amp;quot;constructivism, &amp;quot; or perhaps more accurately, a constructivist metatheory drawn from many sources, has come to dominate contemporary educational psychology. One of constructivism&#039;s principal implications, according to Sack, Soloway and Weingrad [9], is that &amp;quot;...activities of knowledge production (e.g., science) and reproduction (e.g., education) are about convincing, recruiting and enculturating others. in short a constructivist analysis of knowledge foregrounds rhetoric: the powers of persuasion and the difficulties of dispute &amp;quot; (p. 357). This rhetoricized conception of knowledge construction, we would argue, has not been adequately reflected in the work of educational technologists working from constructivist theory. Subtitled “A Rhetorical Approach to Constructivist Learning, ” Reality Check is a prototype of educational software that we are currently developing to embody many of the lessons we believe knowledge negotiation offers educators and educational technologists. Using the metaphor of education as “knowledge-negotiation” Reality Check prompts educators and students independently to construct their representation of a task by identifying relevant variables and giving reasons and evidence that permit fruitful interrogation. By setting up alternative representations, Reality Check explicitly invites the teacher and student to participate in a negotiation of knowledge in which evidence is exchanged and positions are clarified, whereby each party in this educational activity can better understand the other and, more importantly, make productive use of this understanding.
A Tree Partitioning Method for Memory Management in Association Rule Mining.

An Evaluation of Approaches to Classification Rule Selection.

Tree Structures for Mining Association Rules.

Data Structure for Association Rule Mining T%2DTrees and P%2DTrees.
the P-tree, together with associated algorithms, are described. The authors demonstrate that the structures and algorithms offer significant advantages in terms of storage and execution time.
T%2DTrees%2C Vertical Partitioning and Distributed Association Rule Mining.

Setting attribute weights for k%2DNN based binary classification via quadratic programming.
Abstract. The k-Nearest Neighbour (k-NN) method is a typical lazy learning paradigm for solving classification problems. Although this method was originally proposed as a non-parameterised method, attribute weight setting has been commonly adopted to deal with irrelevant attributes. In this paper, we propose a new attribute weight setting method for k-NN based classifiers using quadratic programming, which is particularly suitable for binary classification problems. Our method formalises the attribute weight setting problem as a quadratic programming problem and exploits commercial software to calculate attribute weights. To evaluate our method, we carried out a series of experiments on six established data sets. Experiments show that our method is quite practical for various problems and can achieve a stable increase in accuracy over the standard k-NN method as well as a competitive performance. Another merit of the method is that it can use small training sets.
An Attribute Weight Setting Method for k%2DNN Based Binary Classification using Quadratic Programming.
Abstract. In this paper, we propose a new attribute weight setting method for k-NN based classifiers using quadratic programming, which is particular suitable for binary classification problems. Our method formalises the attribute weight setting problem as a quadratic programming problem and exploits commercial software to calculate attribute weights. Experiments show that our method is quite practical for various problems and can achieve a competitive performance. Another merit of the method is that it can use small training sets. 1.
An Experimental Study of Increasing Diversity for Case%2DBased Diagnosis.
Abstract. Increasing diversity for case-based reasoning (CBR) is an issue that has recently drawn the attention of researchers in the CBR field. Several diversification techniques have been proposed and discussed in the literature. However, whether and to what extent those techniques can bring about benefits to end-users remains in question. In this paper, we report an experiment in applying a diversification technique to a case-based diagnosis tool in a product maintenance domain. The results of this offer some evidence in support of diversification techniques.
Finding Association Rules with Some Very Frequent Attributes.

Formalising optimal feature weight setting in case based diagnosis as linear programming problems.
 
An Architecture For Web%2DBased Post%2DSales Service an a Flexible Manufacturing Environment.

Computing Association Rules Using Partial Totals.
The problem of extracting all association rules from within a binary database is well-known.  Existing methods may involve multiple passes of the database, and cope badly with densely-packed  database records because of the combinatorial explosion in the number of sets of attributes which  must be considered. We present a new class of methods which use a single pass of the database to  perform a partial computation of the totals required, and to store these partial totals in a structure  which facilitates computation of the final totals. We describe generic algorithms for computing these  totals for which storage and performance characteristics are examined. The method opens the way  to a number of strategies for practical implementations, for which heuristics are discussed.  KEYWORDS: Association Rules, Partial Support, Set Enumeration  1  1 INTRODUCTION  A well-established approach to Knowledge Discovery in Databases (KDD) involves the identification of association rules [1] within a...
Verification
We show that every language in NP has a probablistic verifier that checks membership proofs for it using logarithmic number of random bits and by examining a constant number of bits in the proof. If a string is in the language, then there exists a proof such that the verifier accepts with probability 1 (i.e., for every choice of its random string). For strings not in the language, the verifier rejects every provided “proof &#034; with probability at least 1/2. Our result builds upon and improves a recent result of Arora and Safra [6] whose verifiers examine a nonconstant number of bits in the proof (though this number is a very slowly growing function of the input length). As a consequence we prove that no MAX SNP-hard problem has a polynomial time approximation scheme, unless NP=P. The class MAX SNP was defined by Papadimitriou and Yannakakis [82] and hard problems for this class include vertex cover, maximum satisfiability, maximum cut, metric TSP, Steiner trees and shortest superstring. We also improve upon the clique hardness results of Feige, Goldwasser, Lovász, Safra and Szegedy [42], and Arora and Safra [6] and shows that there exists a positive ɛ such that approximating the maximum clique size in an N-vertex graph to within a factor of N ɛ is NP-hard. 1
An Experiment in Discovering Association Rules in the Legal Domain.
In this paper we explore the applicability of an algorithm designed for finding association rules in large databases to the discovery of relevant associations from a large case base. 1.
Tesseral spatio%2Dtemporal reasoning for multi%2Ddimensional data.
A generally applicable approach to N-dimensional spatial reasoning is described. The approach is founded on a unique representation based on ideas concerning &#034;tesseral&#034; addressing. This offers many computational advantages including minimal data storage, computationally efficient translation of data, and simple data comparison, regardless of the number of dimensions under consideration. The representation allows spatial attributes associated with objects to be expressed simply and concisely in terms of sets of addresses which can then be related using standard set operations expressed as constraints. The approach has been incorporated into a spatial reasoning system --- the SPARTA (SPAtial Reasoning using Tesseral Addressing) system --- which has been successfully used in conjunction with a significant number of spatial application domains. Keywords: Spatio-Temporal Reasoning, Tesseral Addressing, N-Dimensional information processing. 1 INTRODUCTION  A versatile and generally applicabl...
Validation and Verification of Knowledge Based Systems %2D Theory

Verification
We show that every language in NP has a probablistic verifier that checks membership proofs for it using logarithmic number of random bits and by examining a constant number of bits in the proof. If a string is in the language, then there exists a proof such that the verifier accepts with probability 1 (i.e., for every choice of its random string). For strings not in the language, the verifier rejects every provided “proof &#034; with probability at least 1/2. Our result builds upon and improves a recent result of Arora and Safra [6] whose verifiers examine a nonconstant number of bits in the proof (though this number is a very slowly growing function of the input length). As a consequence we prove that no MAX SNP-hard problem has a polynomial time approximation scheme, unless NP=P. The class MAX SNP was defined by Papadimitriou and Yannakakis [82] and hard problems for this class include vertex cover, maximum satisfiability, maximum cut, metric TSP, Steiner trees and shortest superstring. We also improve upon the clique hardness results of Feige, Goldwasser, Lovász, Safra and Szegedy [42], and Arora and Safra [6] and shows that there exists a positive ɛ such that approximating the maximum clique size in an N-vertex graph to within a factor of N ɛ is NP-hard. 1
Region Description and Comparative Analysis using a Tesseral Representation.

Rulebase Checking Using a Spatial Representation.
. In this paper we describe an approach to rule base V&amp;V founded on a N-dimensional tesseral representation of space. The advantages offered by the representation are that it is simple, concise and lends itself to computationally effective manipulation of objects (defined in the representation) using standard integer addition and subtraction. The encapsulation of a rulebase in this tesseral representation - as a multi-dimensional &#034;arena&#034; - then provides an appropriate mechanism for the application of spatial algorithms to address rulebase verification. The algorithms make full use of the advantages associated with the representation, and consequently offer a significant reduction in time complexity over more traditional approaches to rulebase verification. In particular this paper demonstrates the potential of tesserally encoded rulebases the full impact of which, in the author&#039;s opinion, has yet to be explored. 1 Introduction  In this paper an approach to rulebase verification and val...
Report on the 1st International Workshop on Validation

Rulebase Checking Using a Spatial Representation.
. In this paper we describe an approach to rule base V&amp;V founded on a N-dimensional tesseral representation of space. The advantages offered by the representation are that it is simple, concise and lends itself to computationally effective manipulation of objects (defined in the representation) using standard integer addition and subtraction. The encapsulation of a rulebase in this tesseral representation - as a multi-dimensional &#034;arena&#034; - then provides an appropriate mechanism for the application of spatial algorithms to address rulebase verification. The algorithms make full use of the advantages associated with the representation, and consequently offer a significant reduction in time complexity over more traditional approaches to rulebase verification. In particular this paper demonstrates the potential of tesserally encoded rulebases the full impact of which, in the author&#039;s opinion, has yet to be explored. 1 Introduction  In this paper an approach to rulebase verification and val...

KD in FM Knowledge Discovery in Facilities Management Databases.




Spatial Reasoning Using the Quad Tesseral Representation.

A Tesseral Approach to n%2DDimensional Spatial Reasoning.

The Verification and Validation of Rulebases Using a Binary Encoded Incidence Matrix Technique.

An Ontology for Linear Spatial Reasoning.
. An ontology for spatial reasoning based on a tesseral representation of space is presented. The principal advantage offered is that the representation has the effect of linearising multi-dimensional space while still supporting translation through the space in any direction and through any number of dimensions. Consequently, all multi-dimensional spatial reasoning can be implemented using one dimensional (temporal) reasoning techniques. As a result, many of the concerns associated with conventional multi-dimensional spatial reasoning systems, based on more traditional representations, no longer apply. KEYWORDS: Multi-dimensional spatial reasoning, Ontology 1 INTRODUCTION  An ontology is presented for multi-dimensional spatial reasoning based on an addressing system that has the effect of linearising space (regardless of the number of dimensions under consideration). As a result all multi-dimensional reasoning can be implemented in terms of a single dimension and hence many of the con...
Temporal reasoning using tesseral addressing towards an intelligent environmental impact assessment system.

Spatial Reasoning for GIS Using a Tesseral Data Representation

Developing Distributed Database Applications Using TSL

Advanced binary encoded matrix representation for rule base verification.

Representing Visual Conditions in a Legal knowledge Based System.

Two Aspects of the Validation and Verification of Knowledge%2DBased Systems.

Building Knowledge Based Systems for Maintainability.

Electronic Chart Representation and Interaction.

The maintenance of legal knowledge based systems.
One of the main problems in the eld of Arti cial Intelligence and Law, perhaps the main problem [McCarty, 1990], is the construction of adequate representations of legal
A Graphical Interactive Tool for KBS Maintenance.

Two Aspects of the Validation and Verification of Knowledge Based Systems.



