M. Sulaiman Khan

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.

Weighted Association Rule Mining from Binary and Fuzzy Data

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 Fuzzy Association Rules from Composite Items

This paper presents an approach for mining fuzzy Association Rules (ARs) relating the properties of composite items, i.e. items that each feature a number of values derived from a common schema. We partition the values associated to properties into fuzzy sets in order to apply fuzzy Association Rule Mining (ARM). This paper describes the process of deriving the fuzzy sets from the properties associated to composite items and a unique Composite Fuzzy Association Rule Mining (CFARM) algorithm founded on the certainty factor interestingness measure to extract fuzzy association rules. The paper demonstrates the potential of composite fuzzy property ARs, and that a more succinct set of property ARs can be produced using the proposed approach than that generated using a non-fuzzy method.

Towards healthy association rule mining (HARM): a fuzzy quantitative approach

Association Rule Mining (ARM) is a popular data mining technique that has been used to determine customer buying patterns. Although improving performance and efficiency of various ARM algorithms is important, determining Healthy Buying Patterns (HBP) from customer transactions and association rules is also important. This paper proposes a framework for mining fuzzy attributes to generate HBP and a method for analysing healthy buying patterns using ARM. Edible attributes are filtered from transactional input data by projections and are then converted to Required Daily Allowance (RDA) numeric values. Depending on a user query, primitive or hierarchical analysis of nutritional information is performed either from normal generated association rules or from a converted transactional database. Query and attribute representation can assume hierarchical or fuzzy values respectively. Our approach uses a general architecture for Healthy Association Rule Mining (HARM) and prototype support tool that implements the architecture. The paper concludes with experimental results and discussion on evaluating the proposed framework.

A Framework for Mining Fuzzy Association Rules from Composite Items

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.

Finding correlations between 3-D surfaces: a study in asymmetric incremental sheet forming

A mechanism for describing 3-D local geometries is presented which is suitable for input into a classifier generator. The objective is to predict the springback that will occur when Asymmetric Incremental Sheet Forming (AISF) is applied to sheet metal to produce a desired shape so that corrective measures can be applied. The springback is localised hence the desired before shape and the actual after shape are expressed using the concept of a Local Geometry Matrix (LGMs). The reported evaluation demonstrates that the LGM idea can be usefully employed to capture local geometries with respect to individual shapes.

On Clustering Attribute-oriented Induction

Conceptual clustering forms groups of related data items using some distance metrics. Inductive techniques like attribute-oriented induction AOI) generate meta-level descriptions of attribute values without explicitly stated distance metrics and overall goodness functions required for a clustering algorithm. The generalisation process in AOI, per attribute basis, groups attribute values using concise descriptions of a tree hierarchy for that attribute. A conceptual clustering approach is considered for attribute-oriented induction where goodness functions for maintaining intra-cluster tightness within clusters, inter-cluster dissimilarity between clusters and cluster quality evaluation are defined. Attributes are partitioned into natural common parent concept clusters, their tightness, dissimilarity and quality computed for determining a cluster to generalise within the chosen attribute. This principle minimises overgeneralisation and follows a natural clustering approach. Overall, AOI is presented as an agglomerative clustering algorithm, clusterAOI and comparative effectiveness with classical AOI analysed.

Query coordination for distributed data sharing in P2P networks

Organisations often store information about the same entity objects or features in different formats. Accessing and integrating this distributed information can be a difficult task because of schema differences and database platform issues. In this paper, a discussion on query coordination in schema conflicting databases of Peer-to-Peer (P2P) systems is presented. The coordination mechanism with a global query to resolve the issue is proposed. The coordination mechanism is done without peers knowing each others schemas by translating the global query into the local query according to the under lying database schema while a wrapper is used to deal with database platform issues. The paper simulates a real-life application and shows that schema resolution by a query coordination mechanism in P2P systems is effective and minimises most of the complexities encountered by schema integration systems.