Summarising Data by Clustering Items.

Making pattern mining useful.

Identifying the Components.
To understand a component and how it is used in a system, an engineer must understand the externally visible properties and relationships with other components. We applied a technique, Interface Identification, to a product&#039;s source code to identify all of the interfaces between a component and the system. This permitted us to quickly harvest a component from one product for reuse in a new one. The technique relies on source code and knowledge of what source files make up a component. A reverse engineering tool aids the process.
Low%2DEntropy Set Selection.

Identifying the components.
To understand a component and how it is used in a system, an engineer must understand the externally visible properties and relationships with other components. We applied a technique, Interface Identification, to a product&#039;s source code to identify all of the interfaces between a component and the system. This permitted us to quickly harvest a component from one product for reuse in a new one. The technique relies on source code and knowledge of what source files make up a component. A reverse engineering tool aids the process.
Filling in the Blanks %2D Krimp Minimisation for Missing Data.



Preserving Privacy through Data Generation.

Characterising the difference.
Characterising the differences between two databases is an often occurring problem in Data Mining. Detection of change over time is a prime example, comparing databases from two branches is another one. The key problem is to discover the patterns that describe the difference. Emerging patterns provide only a partial answer to this question. In previous work, we showed that the data distribution can be captured in a pattern-based model using compression [12]. Here, we extend this approach to define a generic dissimilarity measure on databases. Moreover, we show that this approach can identify those patterns that characterise the differences between two distributions. Experimental results show that our method provides a wellfounded way to independently measure database dissimilarity that allows for thorough inspection of the actual differences. This illustrates the use of our approach in real world data mining.
Compression Picks Item Sets That Matter.

Item Sets that Compress.
ABSTRACT: Association rule mining has been the main technique to identify meaningful patterns in market basket data (also known as synthetic transactional data). However, this approach has several limitations. Especially the large number of rules that have been generated from association creates difficulties in extracting and analyzing useful information. So, clustering association rules is introduced to overcome the problem. In this paper, we discuss three combined approaches to cluster association rules. First, we create hierarchical concepts and relationships of domain ontologies to merge and simplify items into more general concepts. This significantly reduces the total number of rules that are created. Adding to this, subspace clustering further reduces the resulting association rules to a more relevant problem space according to the level of details. Finally, a new ranking function is introduced to give more emphasis to the meaningful clustered groups via frequency, recency, and monetary value, which are the most important factors in the marketing domain. Our experiment shows that the total number of rules has decreased while the value of lift has increased. Also, the result of hierarchical subspace clustering shows that our approach groups data effectively to more simple, understandable and relevant problem space that is more suitable for further analysis.
