Complex objects are often described by several attributes of arbitrary types. To calculate the similarity between complex objects, we propose the CoSim process, which provides a composed similarity function. CoSim applies existing similarity functions for each data type of the object. To calculate the overall similarity, we weight these type-specific similarities. We demonstrate the practical application of our approach by using a simple example of objects. The example involves numerical and categorical attributes. However, our presented idea is applicable to objects with attributes of arbitrary types.

Modern information systems consist of many distributed computer and database systems. The integration of such distributed data into a single data warehouse system is confronted with the well known problem of low data quality. In this paper we present an approach that facilitates a dynamic identification of spurious and error-prone data stored in a large data warehouse. The identification of data quality problems is based on data mining techniques, such as clustering, subspace clustering and classification. Furthermore, we present via a case study the applicability of our approach on real data. The experimental results show that our approach efficiently identifies data quality problems.

We introduce in this work an approach, which tackles the knowledge integration from distributed biomedical databases. Traditional data evaluation is performed only on local data. From a global context these evaluation results are partial and incomplete. This is due to the fact, that the knowledge residing on other existing databases is not considered. Thus, making the necessity of global knowledge integration inevitable. This is exactly what we propose in this work, by creating a globally integrated knowledge network on top of the existing distributed biomedical databases, i.e. we are not aiming at integrating the whole data, but integrating only the knowledge residing on these databases. To do this, we apply distributed data mining techniques, which make it possible to analyze the distributed data without integrating it into a singe data warehouse.

Analysing large amounts of biomedical data is the new challenge in the post-genomic era. One of the goals in gene research is the computation of the similarity between diseases based on the genes they are related to. Identifying biomedical relationships between diseases can lead to finding of new drugs and medicaments. The human disease network (Diseasome) illustrates the association between diseases based on genes these diseases share. A disadvantage of this network is the data itself, as Diseasome is based only on a single database (OMIM). There exist, however, a large number of other biomedical databases, and integrating them, in order to be able to profit from all their data, is an impossible task. Thus, we propose a different approach, namely, to focus only on the integration of the knowledge of all these databases. In our approach, we extend Diseasome by integrating the knowledge from other distributed databases, without needing to integrate the data itself. To compute the similarity between diseases we apply data mining techniques.

Clustering has become an increasingly important task in modern application domains. Mostly, the data are originally collected at different sites. In order to extract information from these data. they are merged at a central site and then clustered. Another approach is to cluster the data locally and extract suitable representatives from these clusters. Based on these representatives a global server tries to reconstruct the complete clustering. In this paper, we discuss the complex problem of finding a suitable quality measure for evaluating the quality of such a distributed clustering. We introduce a discrete and continuous quality criterion which we empirically compare to each other.

Clustering has become an increasingly important task in analysing huge amounts of data. Traditional applications require that all data has to be located at the site where it is scrutinized. Nowadays, large amounts of heterogeneous, complex data reside on different, independently working computers which are connected to each other via local or wide area networks. In this paper, we propose a scalable density-based distributed clustering algorithm which allows a user-defined trade-off between clustering quality and the number of transmitted objects from the different local sites to a global server site. Our approach consists of the following steps: First, we order all objects located at a local site according to a quality criterion reflecting their suitability to serve as local representatives. Then we send the best of these representatives to a server site where they are clustered with a slightly enhanced density-based clustering algorithm. This approach is very efficient, because the local detemination of suitable representatives can be carried out quickly and independently from each other. Furthermore, based on the scalable number of the most suitable local representatives, the global clustering can be done very effectively and efficiently. In our experimental evaluation, we will show that our new scalable density-based distributed clustering approach results in high quality clusterings with scalable transmission cost.

Similarity search in database systems is becoming an increasingly im- portant task in modern application domains such as multimedia, molecular biology, medical imaging, computer aided engineering, marketing and purchasing assist- ance as well as many others. In this paper, we show how visualizing the hierarchi- cal clustering structure of a database of objects can aid the user in his time consuming task to find similar objects. We present related work and explain its shortcomings which led to the development of our new methods. Based on reach- ability plots, we introduce approaches which automatically extract the significant clusters in a hierarchical cluster representation along with suitable cluster repre- sentatives. These techniques can be used as a basis for visual data mining. We im- plemented our algorithms resulting in an industrial prototype which we used for the experimental evaluation. This evaluation is based on real world test data sets and points out that our new approaches to automatic cluster recognition and extraction of cluster representatives create meaningful and useful results in comparatively short time.

Clustering has become an increasingly important task in modern application domains such as marketing and purchasing assistance, multimedia, molecular biology as well as many others. In many of these areas, the data are originally collected at different sites. In order to extract information out of these data, they are brought together and then clustered. In this paper, we propose a different approach. We cluster the data locally and extract suitable representatives out of these clusters. These representatives are sent to a global server site where we restore the complete clustering based on the local representatives. This approach is very efficient, because the local clustering can be carried out quickly and independently from each other. Furthermore, we have low transmission cost, as the number of transmitted representatives is much smaller than the cardinality of the complete data set. Based on this small number of representatives, the global clustering can be done very efficiently. For both the local and the global clustering, we use a density based clustering algorithm. The combination of both the local and the global clustering forms our new DBDC (Density Based Distributed Clustering) algorithm. In our experimental evaluation, we will show that we do not have to sacrifice the clustering quality in order to gain an efficiency advantage if we use distributed clustering.