Exploiting local node cache in top%2Dk queries within wireless sensor networks.

Reverse k%2DNearest Neighbor monitoring on mobile objects.

Techniques for efficiently searching in spatial

Subspace similarity search using the ideas of ranking and top%2Dk retrieval.

Boosting spatial pruning on optimal pruning of MBRs.

Towards Archaeo%2Dinformatics Scientific Data Management for Archaeobiology.

Can Shared%2DNeighbor Distances Defeat the Curse of Dimensionality%3F

Can Shared%2DNeighbor Distances Defeat the Curse of Dimensionality%3F

Subspace Similarity Search Efficient k%2DNN Queries in Arbitrary Subspaces.

QUeST 2009 Workshop Report The 1st International Workshop on Querying and Mining Uncertain Spatio%2DTemporal Data %28Seattle

Metric spaces in data mining applications to clustering.
(ACM) Special Interest Group on Spatial Information (SIGSPATIAL). ACM SIGSPATIAL addresses issues related to the acquisition, management, and processing of spatially-related information with a focus on algorithmic, geometric, and visual considerations. The scope includes, but is not limited to, geographic information systems. Current ACM SIGSPATIAL officers are:


OSSOBOOK database and knowledgemanagement techniques for archaeozoology.

Periodic Pattern Analysis in Time Series Databases.
Abstract. Similarity search in time series data is used in diverse domains. The most prominent work has focused on similarity search considering either complete time series or certain subsequences of time series. Often, time series like temperature measurements consist of periodic patterns, i.e. patterns that repeatedly occur in defined periods over time. For example, the behavior of the temperature within one day is commonly correlated to that of the next day. Analysis of changes within the patterns and over consecutive patterns could be very valuable for many application domains, in particular finance, medicine, meteorology and ecology. In this paper, we present a framework that provides similarity search in time series databases regarding specific periodic patterns. In particular, an efficient threshold-based similarity search method is applied that is invariant against small distortions in time. Experiments on real-world data show that our novel similarity measure is more meaningful than established measures for many applications. 1
Techniques for Efficiently Searching in Spatial

Reverse k%2Dnearest neighbor search in dynamic and general metric databases.
In this paper, we propose an original solution for the general reverse k-nearest neighbor (RkNN) search problem. Compared to the limitations of existing methods for the RkNN search, our approach works on top of any hierarchically organized tree-like index structure and, thus, is applicable to any type of data as long as a metric distance function is defined on the data objects. We will exemplarily show how our approach works on top of the most prevalent index structures for Euclidean and metric data, the R-Tree and the M-Tree, respectively. Our solution is applicable for arbitrary values of k and can also be applied in dynamic environments where updates of the database frequently occur. Although being the most general solution for the RkNN problem, our solution outperforms existing methods in terms of query execution times because it exploits different strategies for pruning false drops and identifying true hits as soon as possible. 1.
Constrained reverse nearest neighbor search on mobile objects.

Incremental Reverse Nearest Neighbor Ranking.%2
Abstract. In this paper, we formalize the novel concept of incremental reverse nearest neighbor ranking and suggest an original solution for this problem. We propose an efficient approach for reporting the results incrementally without the need to restart the search from scratch. Our approach can be applied to a multidimensional feature database which is hierarchically organized by any R-tree like index structure. Our solution does not assume any preprocessing steps which makes it applicable for dynamic environments where updates of the database frequently occur. Experiments show that our approach reports the ranking results with much less page accesses than existing approaches designed for traditional reverse nearest neighbor search applied to the ranking problem. 1
Incremental Reverse Nearest Neighbor Ranking.%2
Abstract. In this paper, we formalize the novel concept of incremental reverse nearest neighbor ranking and suggest an original solution for this problem. We propose an efficient approach for reporting the results incrementally without the need to restart the search from scratch. Our approach can be applied to a multidimensional feature database which is hierarchically organized by any R-tree like index structure. Our solution does not assume any preprocessing steps which makes it applicable for dynamic environments where updates of the database frequently occur. Experiments show that our approach reports the ranking results with much less page accesses than existing approaches designed for traditional reverse nearest neighbor search applied to the ranking problem. 1
Outlier Detection in Axis%2DParallel Subspaces of High Dimensional Data.
Abstract. We propose an original outlier detection schema that detects outliers in varying subspaces of a high dimensional feature space. In particular, for each object in the data set, we explore the axis-parallel subspace spanned by its neighbors and determine how much the object deviates from the neighbors in this subspace. In our experiments, we show that our novel subspace outlier detection is superior to existing fulldimensional approaches and scales well to high dimensional databases. 1
Probabilistic Similarity Search for Uncertain Time Series.
Abstract. A probabilistic similarity query over uncertain data assigns to each uncertain database object o a probability indicating the likelihood that o meets the query predicate. In this paper, we formalize the notion of uncertain time series and introduce two novel and important types of probabilistic range queries over uncertain time series. Furthermore, we propose an original approximate representation of uncertain time series that can be used to efficiently support both new query types by upper and lower bounding the Euclidean distance. 1
Reverse k%2DNearest Neighbor Search Based on Aggregate Point Access Methods.
Abstract. We propose an original solution for the general reverse k-nearest neighbor (RkNN) search problem in Euclidean spaces. Compared to the limitations of existing methods for the RkNN search, our approach works on top of Multi-Resolution Aggregate (MRA) versions of any index structures for multi-dimensional feature spaces where each non-leaf node is additionally associated with aggregate information like the sum of all leaf-entries indexed by that node. Our solution outperforms the state-of-the-art RkNN algorithms in terms of query execution times because it exploits advanced strategies for pruning index entries. 1
Incremental Reverse Nearest Neighbor Ranking in Vector Spaces.
Abstract. In this paper, we formalize the novel concept of incremental reverse nearest neighbor ranking and suggest an original solution for this problem. We propose an efficient approach for reporting the results incrementally without the need to restart the search from scratch. Our approach can be applied to a multidimensional feature database which is hierarchically organized by any R-tree like index structure. Our solution does not assume any preprocessing steps which makes it applicable for dynamic environments where updates of the database frequently occur. Experiments show that our approach reports the ranking results with much less page accesses than existing approaches designed for traditional reverse nearest neighbor search applied to the ranking problem. 1
Multi%2DStep Query Processing.
For an increasing number of modern database applications, efficient support of similarity search becomes an important task. Along with the complexity of the objects such as images, molecules and mechanical parts, also the complexity of the similarity models increases more and more. Whereas algorithms that are directly based on indexes work well for simple medium-dimensional similarity distance functions, they do not meet the efficiency requirements of complex high-dimensional and adaptable distance functions. The use of a multi-step query processing strategy is recommended in these cases, and our investigations substantiate that the number of candidates which are produced in the filter step and exactly evaluated in the refinement step is a fundamental efficiency parameter. After revealing the strong performance shortcomings of the state-of-the-art algorithm for k-nearest neighbor search [Korn et al. 1996], we present a novel multi-step algorithm which is guaranteed to produce the minim...
Subspace Clustering Techniques.
Document clustering is a fundamental task of text mining, by which efficient organization, navigation, summarization and retrieval of documents can be achieved. The clustering of documents presents difficult challenges due to the sparsity and the high dimensionality of text data, and to the complex semantics of the natural language. Subspace clustering is an extension of traditional clustering that is designed to capture local feature relevance, and to group documents with respect to the features (or words) that matter the most. This paper presents a subspace clustering technique based on a Locally Adaptive Clustering (LAC) algorithm. To improve the subspace clustering of documents and the identification of keywords achieved by LAC, kernel methods and semantic distances are deployed. The basic idea is to define a local kernel for each cluster by which semantic distances between pairs of words are computed to derive the clustering and the local term weightings. The proposed approach, called Semantic LAC, is evaluated using benchmark datasets. Our experiments show that Semantic LAC is capable of improving the clustering quality. 1
Subspace and projected clustering experimental evaluation and analysis.



Approximate Clustering of Time Series Using Compact Model%2DBased Descriptions.
Abstract. Clustering time series is usually limited by the fact that the length of the time series has a significantly negative influence on the runtime. On the other hand, approximative clustering applied to existing compressed representations of time series (e.g. obtained through dimensionality reduction) usually suffers from low accuracy. We propose a method for the compression of time series based on mathematical models that explore dependencies between different time series. In particular, each time series is represented by a combination of a set of specific reference time series. The cost of this representation depend only on the number of reference time series rather than on the length of the time series. We show that using only a small number of reference time series yields a rather accurate representation while reducing the storage cost and runtime of clustering algorithms significantly. Our experiments illustrate that these representations can be used to produce an approximate clustering with high accuracy and considerably reduced runtime. 1
Analysis of Time Series Using Compact Model%2DBased Descriptions.
Abstract. Recently, we have proposed a novel method for the compression of time series based on mathematical models that explore dependencies between different time series. This representation models each time series by a combination of a set of specific reference time series. The cost of this representation depend only on the number of reference time series rather than on the length of the time series. In this demonstration, we present a Java toolkit which is able to perform several data mining tasks based on this novel time series representation. In particular, this framework allows the user to explore the properties of our novel approach in comparison to other state-of-the-art compression methods. The results are visually presented in a very concise way so that the user can easily identify important settings of the model-based time series representation. 1
Continuous proximity monitoring in road networks.
In this paper, we consider the following scenario: a set of mobile objects continuously track their positions in a road network and are able to communicate with a central server. The server which gets position updates from the moving objects has to detect the event that two objects reach or exceed a specified proximity distance. This way, the server is permanently aware of all pairs of objects that are within a certain distance range. Obviously, the communication costs between the objects and the server quickly become the bottleneck if a position update is sent to the server at each tracking time slot. We propose update strategies in order to reduce the communication overhead by defining special regions for each object. These regions are defined such that no position updates at the server are required as long as the objects do not leave their corresponding regions. We present efficient algorithms for updating these regions and detecting proximity/separation when objects leave their corresponding regions. Furthermore, we empirically evaluate the different strategies in terms of communication overhead, i.e. the number of required position updates.
Efficient Query Processing in Large Traffic Networks.
Abstract — We present an original graph embedding to speedup distance-range and k-nearest neighbor queries on static and/or dynamic objects located on a (weighted) graph. Our method is used to compute a lower and upper bounding filter distance which approximates the true shortest path distance significantly better than traditional filters. In addition, we discuss how the computation of the exact shortest path distance in the refinement step can be boosted by using the embedded graph. I.
T%2DTime Threshold%2DBased Data Mining on Time Series.
Abstract — Mining time series data is an important approach for the analysis in many application areas as diverse as biology, environmental research, medicine, or stock chart analysis. As nearly all data mining tasks on this kind of data depend on a distance function between two time series, a huge number of such functions has been developed during the last decades. The introduction of threshold-based distance functions presented a new concept of time series similarity and these functions were applied to data mining techniques on a wide spectrum of time series data. In this demonstration, we present the Java toolkit T-Time which is able to perform several data mining tasks for a complete range of threshold values in an interactive way. The results are visually presented in a very concise way so that the user can easily identify important threshold values. Combined with domain-specific knowledge, these pivotal values can yield novel insights beyond the means of the underlying data mining techniques the analysis is based on. I.
Similarity Search in Multimedia Time Series Data Using Amplitude%2DLevel Features.

Robust Clustering in Arbitrarily Oriented Subspaces.
In this paper, we propose an efficient and effective method to find arbitrarily oriented subspace clusters by mapping the data space to a parameter space defining the set of possible arbitrarily oriented subspaces. The objective of a clustering algorithm based on this principle is to find those among all the possible subspaces, that accommodate many database objects. In contrast to existing approaches, our method can find subspace clusters of different dimensionality even if they are sparse or are intersected by other clusters within a noisy environment. A broad experimental evaluation demonstrates the robustness, efficiency and effectivity of our method. 1
Hierarchical Graph Embedding for Efficient Query Processing in Very Large Traffic Networks.

A General Framework for Increasing the Robustness of PCA%2DBased Correlation Clustering Algorithms.
Abstract. Most correlation clustering algorithms rely on principal component analysis (PCA) as a correlation analysis tool. The correlation of each cluster is learned by applying PCA to a set of sample points. Since PCA is rather sensitive to outliers, if a small fraction of these points does not correspond to the correct correlation of the cluster, the algorithms are usually misled or even fail to detect the correct results. In this paper, we evaluate the influence of outliers on PCA and propose a general framework for increasing the robustness of PCA in order to determine the correct correlation of each cluster. We further show how our framework can be applied to PCA-based correlation clustering algorithms. A thorough experimental evaluation shows the benefit of our framework on several synthetic and real-world data sets. 1
A General Framework for Increasing the Robustness of PCA%2DBased Correlation Clustering Algorithms.
Abstract. Most correlation clustering algorithms rely on principal component analysis (PCA) as a correlation analysis tool. The correlation of each cluster is learned by applying PCA to a set of sample points. Since PCA is rather sensitive to outliers, if a small fraction of these points does not correspond to the correct correlation of the cluster, the algorithms are usually misled or even fail to detect the correct results. In this paper, we evaluate the influence of outliers on PCA and propose a general framework for increasing the robustness of PCA in order to determine the correct correlation of each cluster. We further show how our framework can be applied to PCA-based correlation clustering algorithms. A thorough experimental evaluation shows the benefit of our framework on several synthetic and real-world data sets. 1
Detecting clusters in moderate%2Dto%2Dhigh dimensional data subspace clustering
As a prolific research area in data mining, subspace clustering and related problems induced a vast amount of proposed solutions. However, many publications compare a new proposition – if at all – with one or two competitors or even with a so called “naïve ” ad hoc solution but fail to clarify the exact problem definition. As a consequence, even if two solutions are thoroughly compared experimentally, it will often remain unclear whether both solutions tackle the same problem or, if they do, whether they agree in certain tacit assumptions and how such assumptions may influence the outcome of an algorithm. In this tutorial, we try to clarify (i) the different problem definitions related to subspace clustering in general, (ii) the specific difficulties encountered in this field of research, (iii) the varying assumptions, heuristics, and intuitions forming the basis of different approaches, and (iv) how several prominent solutions essentially tackle different problems. 1.
Global Correlation Clustering Based on the Hough Transform.

Efficient Reverse k%2DNearest Neighbor Estimation.
Abstract: The reverse k-nearest neighbor (RkNN) problem, i.e. finding all objects in a data set the k-nearest neighbors of which include a specified query object, has received increasing attention recently. Many industrial and scientific applications call for solutions of the RkNN problem in arbitrary metric spaces where the data objects are not Euclidean and only a metric distance function is given for specifying object similarity. Usually, these applications need a solution for the generalized problem where the value of k is not known in advance and may change from query to query. In addition, many applications require a fast approximate answer of RkNN-queries. For these scenarios, it is important to generate a fast answer with high recall. In this paper, we propose the first approach for efficient approximative RkNN search in arbitrary metric spaces where the value of k is specified at query time. Our approach uses the advantages of existing metric index structures but proposes to use an approximation of the nearest-neighbor-distances in order to prune the search space. We show that our method scales significantly better than existing non-approximative approaches while producing an approximation of the true query result with a high recall. 1
Detection and Visualization of Subspace Cluster Hierarchies.
Abstract. Subspace clustering (also called projected clustering) addresses the problem that different sets of attributes may be relevant for different clusters in high dimensional feature spaces. In this paper, we propose the algorithm DiSH (Detecting Subspace cluster Hierarchies) that improves in the following points over existing approaches: First, DiSH can detect clusters in subspaces of significantly different dimensionality. Second, DiSH uncovers complex hierarchies of nested subspace clusters, i.e. clusters in lower-dimensional subspaces that are embedded within higher-dimensional subspace clusters. These hierarchies do not only consist of single inclusions, but may also exhibit multiple inclusions and thus, can only be modeled using graphs rather than trees. Third, DiSH is able to detect clusters of different size, shape, and density. Furthermore, we propose to visualize the complex hierarchies by means of an appropriate visualization model, the so-called subspace clustering graph, such that the relationships between the subspace clusters can be explored at a glance. Several comparative experiments show the performance and the effectivity of DiSH. 1
Interval%2DFocused Similarity Search in Time Series Databases.

Proximity queries in large traffic networks.
In this paper, we present an original network graph embedding to speed-up distance-range and k-nearest neighbor queries in (weighted) graphs. Our approach implements the paradigm of filter-refinement query processing and can be used for proximity queries on both static as well as dynamic objects. In particular, we present how our embedding can be used to compute a lower and upper bounding filter distance which approximates the true shortest path distance significantly better than traditional filters, e.g. the Euclidean distance. These distance approximations can be used within a filter step to prune true drops and true hits as well as in the refinement step in order to guide an informed A * search. Our experimental evaluation on several real-world data sets demonstrates a significant performance boosting of our proposed concepts over existing work.
Robust%2C Complete%2C and Efficient Correlation Clustering.
Correlation clustering aims at the detection of data points that appear as hyperplanes in the data space and, thus, exhibit common correlations between different subsets of features. Recently proposed methods for correlation clustering usually suffer from several severe drawbacks including poor robustness against noise or parameter settings, incomplete results (i.e. missed clusters), poor usability due to complex input parameters, and poor scalability. In this paper, we propose the novel correlation clustering algorithm COPAC (COrrelation PArtition Clustering) that aims at improved robustness, completeness, usability, and efficiency. Our experimental evaluation empirically shows that COPAC is superior over existing state-of-the-art correlation clustering methods in terms of runtime, accuracy, and completeness of the results. 1
On Exploring Complex Relationships of Correlation Clusters.
In high dimensional data, clusters often only exist in arbitrarily oriented subspaces of the feature space. In addition, these so-called correlation clusters may have complex relationships between each other. For example, a correlation cluster in a 1-D subspace (forming a line) may be enclosed within one or even several correlation clusters in 2-D superspaces (forming planes). In general, such relationships can be seen as a complex hierarchy that allows multiple inclusions, i.e. clusters may be embedded in several super-clusters rather than only in one. Obviously, uncovering the hierarchical relationships between the detected correlation clusters is an important information gain. Since existing approaches cannot detect such complex hierarchical relationships among correlation clusters, we propose the algorithm ERiC to tackle this problem and to visualize the result by means of a graph-based representation. In our experimental evaluation, we show that ERiC finds more information than state-of-the-art correlation clustering methods and outperforms existing competitors in terms of efficiency. 1.
Generalizing the Optimality of Multi%2Dstep k %2DNearest Neighbor Query Processing.
Abstract. Similarity search algorithms that directly rely on index structures and require a lot of distance computations are usually not applicable to databases containing complex objects and defining costly distance functions on spatial, temporal and multimedia data. Rather, the use of an adequate multi-step query processing strategy is crucial for the performance of a similarity search routine that deals with complex distance functions. Reducing the number of candidates returned from the filter step which then have to be exactly evaluated in the refinement step is fundamental for the efficiency of the query process. The state-of-the-art multi-step k-nearest neighbor (kNN) search algorithms are designed to use only a lower bounding distance estimation for candidate pruning. However, in many applications, also an upper bounding distance approximation is available that can additionally be used for reducing the number of candidates. In this paper, we generalize the traditional concept of R-optimality and introduce the notion of RI-optimality depending on the distance information I available in the filter step. We propose a new multi-step kNN search algorithm that utilizes lower- and upper bounding distance information (Ilu) in the filter step. Furthermore, we show that, in contrast to existing approaches, our proposed solution is RI lu-optimal. In an experimental evaluation, we demonstrate the significant performance gain over existing methods. 1

Future trends in data mining.
and Knowledge Discovery (DMKD’03), held annually in conjunction with ACM SIGMOD conference. The workshop aims to bring together data-mining researchers and experienced practitioners with the goal of discussing the next generation of data-mining tools. Rather than following a ”miniconference” format focusing on the presentation of polished research results, the DMKD workshop aims to foster an informal atmosphere, where researchers and practitioners can freely interact through short presentations and open discussions on their ongoing work as well as forward-looking research visions/experiences for future data-mining applications. In addition to research on novel data-mining algorithms and experiences with innovative mining systems and applications, of particular interest in this year’s DMKD workshop is the broad theme of ”Future Trends in Data Mining. ” Topics of interest include: • Privacy &amp; Security issues in Mining: Privacy preserving data mining techniques are invaluable in cases where one may not look at the detailed data, but one is allowed to infer high level information. This also has relevance for the use of mining for national security applications. • Mining Data Streams: In many emerging applications data arrives and needs to be processed on a continuous basis, i.e., there is need for mining without the benefit of several passes over a static, persistent snapshot. • Data Mining in Bioinformatics and Biological Database Systems: High-performance data mining tools will play a crucial role in the analysis of the ever-growing databases of bio-sequences/structures.

Semi%2DSupervised Threshold Queries on Pharmacogenomics Time Sequences.
The analysis of time series data is of capital importance for pharmacogenomics since the experimental evaluations are usually based on observations of time dependent reactions or behaviors of organisms. Thus, data mining in time series databases is an important instrument towards understanding the effects of drugs on individuals. However, the complex nature of time series poses a big challenge for effective and efficient data mining. In this paper, we focus on the detection of temporal dependencies between different time series: we introduce the novel analysis concept of threshold queries and its semi-supervised extension which supports the parameter setting by applying training datasets. Basically, threshold queries report those time series exceeding an user-defined query threshold at certain time frames. For semi-supervised threshold queries the corresponding threshold is automatically adjusted to the characteristics of the data set, the training dataset, respectively. In order to support threshold queries efficiently, we present a new efficient access method which uses the fact that only partial information of the time series is required at query time. In an extensive experimental evaluation we demonstrate the performance of our solution and show that semi-supervised threshold queries applied to gene expression data are very worthwhile. 1.
Approximate reverse k%2Dnearest neighbor queries in general metric spaces.
In this paper, we propose an approach for efficient approximative RkNN search in arbitrary metric spaces where the value of k is specified at query time. Our method uses an approximation of the nearest-neighbor-distances in order to prune the search space. In several experiments, our solution scales significantly better than existing non-approximative approaches while producing an approximation of the true query result with a high recall.
Similarity Search on Time Series Based on Threshold Queries.
Abstract. Similarity search in time series data is required in many application fields. The most prominent work has focused on similarity search considering either complete time series or similarity according to subsequences of time series. For many domains like financial analysis, medicine, environmental meteorology, or environmental observation, the detection of temporal dependencies between different time series is very important. In contrast to traditional approaches which consider the course of the time series for the purpose of matching, coarse trend information about the time series could be sufficient to solve the above mentioned problem. In particular, temporal dependencies in time series can be detected by determining the points of time at which the time series exceeds a specific threshold. In this paper, we introduce the novel concept of threshold queries in time series databases which report those time series exceeding a user-defined query threshold at similar time frames compared to the query time series. We present a new efficient access method which uses the fact that only partial information of the time series is required at query time. The performance of our solution is demonstrated by an extensive experimental evaluation on real world and artificial time series data. 1
TQuEST Threshold Query Execution for Large Sets of Time Series.

Similarity Search on Time Series Based on Threshold Queries.
Abstract. Similarity search in time series data is required in many application fields. The most prominent work has focused on similarity search considering either complete time series or similarity according to subsequences of time series. For many domains like financial analysis, medicine, environmental meteorology, or environmental observation, the detection of temporal dependencies between different time series is very important. In contrast to traditional approaches which consider the course of the time series for the purpose of matching, coarse trend information about the time series could be sufficient to solve the above mentioned problem. In particular, temporal dependencies in time series can be detected by determining the points of time at which the time series exceeds a specific threshold. In this paper, we introduce the novel concept of threshold queries in time series databases which report those time series exceeding a user-defined query threshold at similar time frames compared to the query time series. We present a new efficient access method which uses the fact that only partial information of the time series is required at query time. The performance of our solution is demonstrated by an extensive experimental evaluation on real world and artificial time series data. 1
Threshold Similarity Queries in Large Time Series Databases.
Similarity search in time series data is an active area of research. In this paper, we introduce the novel concept of threshold-similarity queries in time series databases which report those time series exceeding a user-defined query threshold at similar time frames compared to the query time series. In addition, we present a new data structure to support threshold similarity queries efficiently. The performance of our solution is demonstrated by an extensive experimental evaluation.

Deriving quantitative models for correlation clusters.
Correlation clustering aims at grouping the data set into correlation clusters such that the objects in the same cluster exhibit a certain density and are all associated to a common arbitrarily oriented hyperplane of arbitrary dimensionality. Several algorithms for this task have been proposed recently. However, all algorithms only compute the partitioning of the data into clusters. This is only a first step in the pipeline of advanced data analysis and system modelling. The second (post-clustering) step of deriving a quantitative model for each correlation cluster has not been addressed so far. In this paper, we describe an original approach to handle this second step. We introduce a general method that can extract quantitative information on the linear dependencies within a correlation clustering. Our concepts are independent of the clustering model and can thus be applied as a post-processing step to any correlation clustering algorithm. Furthermore, we show how these quantitative models can be used to predict the probability distribution that an object is created by these models. Our broad experimental evaluation demonstrates the beneficial impact of our method on several applications of significant practical importance.

DeLi%2DClu Boosting Robustness

Finding Hierarchies of Subspace Clusters.
Abstract. Many clustering algorithms are not applicable to high-dimensional feature spaces, because the clusters often exist only in specific subspaces of the original feature space. Those clusters are also called subspace clusters. In this paper, we propose the algorithm HiSC (Hierarchical Subspace Clustering) that can detect hierarchies of nested subspace clusters, i.e. the relationships of lowerdimensional subspace clusters that are embedded within higher-dimensional subspace clusters. Several comparative experiments using synthetic and real data sets show the performance and the effectivity of HiSC. 1
Efficient reverse k%2Dnearest neighbor search in arbitrary metric spaces.
The reverse k-nearest neighbor (RkNN) problem, i.e. finding all objects in a data set the k-nearest neighbors of which include a specified query object, is a generalization of the reverse 1-nearest neighbor problem which has received increasing attention recently. Many industrial and scientific applications call for solutions of the RkNN problem in arbitrary metric spaces where the data objects are not Euclidean and only a metric distance function is given for specifying object similarity. Usually, these applications need a solution for the generalized problem where the value of k is not known in advance and may change from query to query. However, existing approaches, except one, are designed for the specific R1NN problem. In addition — to the best of our knowledge — all previously proposed methods, especially the one for generalized RkNN search, are only applicable to Euclidean vector data but not for general metric objects. In this paper, we propose the first approach for efficient RkNN search in arbitrary metric spaces where the value of k is specified at query time. Our approach uses the advantages of existing metric index structures but proposes to use conservative and progressive distance approximations in order to filter out true drops and true hits. In particular, we approximate the k-nearest neighbor distance for each data object by upper and lower bounds using two functions of only two parameters each. Thus, our method does not generate any considerable storage overhead. We show in a broad experimental evaluation on real-world data the scalability and the usability of our novel approach. 1.
Mining Hierarchies of Correlation Clusters.
The detection of correlations between different features in high dimensional data sets is a very important data mining task. These correlations can be arbitrarily complex: One or more features might be correlated with several other features, and both noise features as well as the actual dependencies may be different for different clusters. Therefore, each cluster contains points that are located on a common hyperplane of arbitrary dimensionality in the data space and thus generates a separate, arbitrarily oriented subspace of the original data space. The few recently proposed algorithms designed to uncover these correlation clusters have several disadvantages. In particular, these methods cannot detect correlation clusters of different dimensionality which are nested into each other. The complete hierarchical structure of correlation clusters of varying dimensionality can only be detected by a hierarchical clustering approach. Therefore, we propose the algorithm HiCO (Hierarchical Correlation Ordering), the first hierarchical approach to correlation clustering. The algorithm determines the cluster hierarchy, and visualizes it using correlation diagrams. Several comparative experiments using synthetic and real data sets show the performance and the effectivity of HiCO. 1.
Efficient Query Processing in Arbitrary Subspaces Using Vector Approximations.
In this paper, we introduce the partial vector approximation file, an extension of the well known vector approximation file that is constructed to efficiently answer partial similarity queries in any possible subspace which is not known beforehand. The idea of the partial VA-File is to divide the VA-File into a separate file for each dimension and only load the dimensions that are necessary to answer the query. Thus, the partial VA-File is constructed to improve the query performance for systems that have to cope with a wide variety of previously unknown query subspaces. We propose novel algorithms for partial kNN and ε-range queries based on the new partial VA-File. In our experiments, we demonstrate that our proposed partial VA-File with the novel algorithms improves the average query performance in comparison to the original VA-File when answering partial similarity queries. 1.
Time Series Analysis Using the Concept of Adaptable Threshold Similarity.
The issue of data mining in time series databases is of utmost importance for many practical applications and has attracted a lot of research in the past years. In this paper, we focus on the recently proposed concept of threshold similarity which compares the time series based on the time frames within which they exceed a user-defined amplitude threshold τ. We propose a novel approach for cluster analysis of time series based on adaptable threshold similarity. The most important issue in threshold similarity is the choice of the threshold τ. Thus, the threshold τ is automatically adapted to the characteristics of a small training dataset using the concept of support vector machines. Thus, the optimal τ is learned from a small training set in order to yield an accurate clustering of the entire time series database. In our experimental evaluation we demonstrate that our cluster analysis using adaptable threshold similarity can be successfully applied to many scientific real-world data mining applications. 1.
Online Hierarchical Clustering in a Data Warehouse Environment.
Many important industrial applications rely on data mining methods to uncover patterns and trends in large data warehouse environments. Since a data warehouse is typically updated periodically in a batch mode, the mined patterns have to be updated as well. This requires not only accuracy from data mining methods but also fast availability of up-to-date knowledge, particularly in the presence of a heavy update load. To cope with this problem, we propose the use of online data mining algorithms which permanently store the discovered knowledge in suitable data structures and enable an efficient adaptation of these structures after insertions and deletions on the raw data. In this paper, we demonstrate how hierarchical clustering methods can be reformulated as online algorithms based on the hierarchical clustering method OPTICS, using a density estimator for data grouping. We also discuss how this algorithmic schema can be specialized for efficient online single-link clustering. A broad experimental evaluation demonstrates that the efficiency is superior with significant speed-up factors even for large bulk insertions and deletions. 1
A Generic Framework for Efficient Subspace Clustering of High%2DDimensional Data.
Subspace clustering has been investigated extensively since traditional clustering algorithms often fail to detect meaningful clusters in high-dimensional data spaces. Many recently proposed subspace clustering methods suffer from two severe problems: First, the algorithms typically scale exponentially with the data dimensionality and/or the subspace dimensionality of the clusters. Second, for performance reasons, many algorithms use a global density threshold for clustering, which is quite questionable since clusters in subspaces of significantly different dimensionality will most likely exhibt significantly varying densities. In this paper, we propose a generic framework to overcome these limitations. Our framework is based on an efficient filterrefinement architecture that scales at most quadratic w.r.t. the data dimensionality and the dimensionality of the subspace clusters. It can be applied to any clustering notions including notions that are based on a local density threshold. A broad experimental evaluation on synthetic and real-world data empirically shows that our method achieves a significant gain of runtime and quality in comparison to state-of-the-art subspace clustering algorithms. 1
Effective and Efficient Distributed Model%2DBased Clustering.

Selectivity Estimation of High Dimensional Window Queries via Clustering.

Accurate and Efficient Similarity Search on 3D Objects Using Point Sampling
Abstract. With fast evolving resources for 3D objects such as the Protein Data Bank (PDB) or the World Wide Web, new techniques, so-called similarity models to efficiently and effectively search for these 3D objects become indispensible. Invariances w.r.t. specific geometric transformations such as scaling, translation, and rotation are important features of similarity models. In this paper, we focus on rotation invariance. We first propose a new method of representing objects more accurately in the context of rotation invariance than the well-known voxelization technique.In addition, we extend existing feature-based similarity models by proposing a new spherical partitioning of the data objects based on proportionality and redundancy 1, and generalizing an existing method for feature extraction. A broad experimental evaluation compares our method with existing methods in terms of accuracy and efficiency. In particular, we experimentally confirm that our point sampling method is better suited to represent 3D objects in the context of rotation invariance than voxelized representations. In addition, we empirically show that our new similarity model significantly outperfoms competitive rotation invariant models in terms of accuracy as well as efficiency. 1
BOSS Browsing OPTICS%2DPlots for Similarity Search.
Similarity search in database systems is becoming an increasingly important task in modern application domains such as multimedia, molecular biology, medical imaging, computer aided engineering, marketing and purchasing assistance as well as many others. In this paper, we show how visualizing the hierarchical 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 reachability plots, we introduce approaches which automatically extract the significant clusters in a hierarchical cluster representation along with suitable cluster representatives. These techniques can be used as a basis for visual data mining. We implemented 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.
Subspace Selection for Clustering High%2DDimensional Data.
In high-dimensional feature spaces traditional clustering algorithms tend to break down in terms of efficiency and quality. Nevertheless, the data sets often contain clusters which are hidden in various subspaces of the original feature space. In this paper, we present a feature selection technique called SURFING (SUbspaces Relevant For clusterING) that finds all subspaces interesting for clustering and sorts them by relevance. The sorting is based on a quality criterion for the interestingness of a subspace using the k-nearest neighbor distances of the objects. As our method is more or less parameterless, it addresses the unsupervised notion of the data mining task ”clustering ” in a best possible way. A broad evaluation based on synthetic and real-world data sets demonstrates that SURFING is suitable to find all relevant subspaces in high dimensional, sparse data sets and produces better results than comparative methods. 1.
Density Connected Clustering with Local Subspace Preferences.
Many clustering algorithms tend to break down in high-dimensional feature spaces, because the clusters often exist only in specific subspaces (attribute subsets) of the original feature space. Therefore, the task of projected clustering (or subspace clustering) has been defined recently. As a novel solution to tackle this problem, we propose the concept of local subspace preferences, which captures the main directions of high point density. Using this concept we adopt density-based clustering to cope with high-dimensional data. In particular, we achieve the following advantages over existing approaches: Our proposed method has a determinate result, does not depend on the order of processing, is robust against noise, performs only one single scan over the database, and is linear in the number of dimensions. A broad experimental evaluation shows that our approach yields results of significantly better quality than recent work on clustering high-dimensional data. 1.
Visually Mining through Cluster Hierarchies.
Similarity search in database systems is becoming an increasingly important task in modern application domains such as multimedia, molecular biology, medical imaging, computer aided engineering, marketing and purchasing assistance as well as many others. In this paper, we show how visualizing the hierarchical 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 reachability plots, we introduce approaches which automatically extract the significant clusters in a hierarchical cluster representation along with suitable cluster representatives. These techniques can be used as a basis for visual data mining. We implemented 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.
Density%2DConnected Subspace Clustering for High%2DDimensional Data.
Several application domains such as molecular biology and geography produce a tremendous amount of data which can no longer be managed without the help of efficient and effective data mining methods. One of the primary data mining tasks is clustering. However, traditional clustering algorithms often fail to detect meaningful clusters because most real-world data sets are characterized by a high dimensional, inherently sparse data space. Nevertheless, the data sets often contain interesting clusters which are hidden in various subspaces of the original feature space. Therefore, the concept of subspace clustering has recently been addressed, which aims at automatically identifying subspaces of the feature space in which clusters exist. In this paper, we introduce SUBCLU (density-connected Subspace Clustering), an effective and efficient approach to the subspace clustering problem. Using the concept of density-connectivity underlying the algorithm DBSCAN [EKSX96], SUBCLU is based on a formal clustering notion. In contrast to existing grid-based approaches, SUBCLU is able to detect arbitrarily shaped and positioned clusters in subspaces. The monotonicity of density-connectivity is used to efficiently prune subspaces in the process of generating all clusters in a bottom up way. While not examining any unnecessary subspaces, SUBCLU delivers for each subspace the same clusters DBSCAN would have found, when applied to this subspace separately.
Using Support Vector Machines for Classifying Large Sets of Multi%2DRepresented Objects.
Databases are a key technology for molecular biology which is a very data intensive discipline. Since molecular biological databases are rather heterogeneous, unification and data integration is mandatory to make use of the huge amount of available information. Currently, the most promising approach for integration is the use of ontologies. Since mapping biological entities into ontologies is usually achieved manually or semiautomatically, a system for automatic classification of biological entities into ontologies saves time and effort. Therefore, we present a support vector machine based approach that automatically classifies biological entities into a given ontology. To solve this difficult task, our method copes with the following aspects. Biological entities might belong to more than one class or may be placed in classes on varying abstraction levels. An object may be described by several representations. Thus, the classifier has to be enabled to draw information from all of them, but must consider the possibility that some objects are described incompletely. Therefore, our method introduces the technique of objectadjusted weighting which regulates the impact of each representation dynamically for each object. To significantly improve the time performance of the classifier we exploit the inheritance relations of the given ontology. Our experimental evaluation on protein data and several parts of an established molecular biological ontology shows that our prototype offers impressive accuracy and is efficient enough to cope with the large number of classes encountered in real world problems. ∗ Supported by the German Ministery for Education, Science,
Visually Mining through Cluster Hierarchies.
Similarity search in database systems is becoming an increasingly important task in modern application domains such as multimedia, molecular biology, medical imaging, computer aided engineering, marketing and purchasing assistance as well as many others. In this paper, we show how visualizing the hierarchical 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 reachability plots, we introduce approaches which automatically extract the significant clusters in a hierarchical cluster representation along with suitable cluster representatives. These techniques can be used as a basis for visual data mining. We implemented 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.
Immer gr%F6%DFere und komplexere Datenmengen Herausforderungen f%FCr Clustering%2DAlgorithmen.

Bioinformatics Databases State of the Art and Research Perspectives.
Bioinformatics or computational biology, i.e. the application of mathematical and computer science methods to solving problems in molecular biology that require large scale data, computation, and analysis, is a research area currently receiving a considerable attention. Databases play an essential role in molecular biology and consequently in bioinformatics. molecular biology data are often relatively cheap to produce, leading to a proliferation of databases: the number of bioinformatics databases accessible worldwide probably lies between 500 and 1.000. Not only molecular biology data, but also molecular biology literature and literature references are stored in databases. Bioinformatics databases are often very large (e.g. the sequence database GenBank contains more than 4 × 10 6 nucleotide sequences) and in general grows rapidly (e.g. about 8000 abstracts are added every month to the literature database PubMed). Bioinformatics databases are heterogeneous in their data, in their data modeling paradigms, in their management systems, and in the data analysis tools they supports. Furthermore, bioinformatics databases are often implemented, queried, updated, and managed using methods rarely applied for other databases. This presentation aims at introducing in current bioinformatics databases, stressing their aspects departing from conventional databases. A more detailed survey can be found in [1] upon which
Effective Similarity Search on Voxelized CAD Object.
Similarity search in database systems is becoming an increasingly important task in modern application domains such as multimedia, molecular biology, medical imaging and many others. Especially for CAD applications, suitable similarity models and a clear representation of the results can help to reduce the cost of developing and producing new parts by maximizing the reuse of existing parts. In this paper, we adapt two known similarity models to voxelized 3-D CAD data and introduce a new model based on eigenvectors. The experimental evaluation of our three similarity models is based on two real-world test datasets. Furthermore, we introduce hierarchical clustering as a new and effective way to analyse and compare similarity models. We show that both our similarity model as well as our evaluation procedure are suitable for industrial use.
Ranking Interesting Subspaces for Clustering High Dimensional Data.
Application domains such as life sciences, e.g. molecular biology  produce a tremendous amount of data which can no longer be managed  without the help of e#cient and e#ective data mining methods. One  of the primary data mining tasks is clustering. However, traditional clustering  algorithms often fail to detect meaningful clusters because of the  high dimensional, inherently sparse feature space of most real-world data  sets. Nevertheless, the data sets often contain clusters hidden in various  subspaces of the original feature space. We present a pre-processing step  for traditional clustering algorithms, which detects all interesting subspaces  of high-dimensional data containing clusters. For this purpose, we  define a quality criterion for the interestingness of a subspace and propose  an e#cient algorithm called RIS (Ranking I nteresting Subspaces)  to examine all such subspaces. A broad evaluation based on synthetic  and real-world data sets empirically shows that RIS is suitable to find  all relevant subspaces in large, high dimensional, sparse data and to rank  them accordingly.
Incremental OPTICS Efficient Computation of Updates in a Hierarchical Cluster Ordering.
Data warehouses are a challenging field of application for data mining  tasks such as clustering. Usually, updates are collected and applied to the data  warehouse periodically in a batch mode. As a consequence, all mined patterns  discovered in the data warehouse (e.g. clustering structures) have to be updated  as well. In this paper, we present a method for incrementally updating the clustering  structure computed by the hierarchical clustering algorithm OPTICS. We  determine the parts of the cluster ordering that are affected by update operations  and develop efficient algorithms that incrementally update an existing cluster ordering.
Ranking Interesting Subspaces for Clustering High Dimensional Data.
Application domains such as life sciences, e.g. molecular biology  produce a tremendous amount of data which can no longer be managed  without the help of e#cient and e#ective data mining methods. One  of the primary data mining tasks is clustering. However, traditional clustering  algorithms often fail to detect meaningful clusters because of the  high dimensional, inherently sparse feature space of most real-world data  sets. Nevertheless, the data sets often contain clusters hidden in various  subspaces of the original feature space. We present a pre-processing step  for traditional clustering algorithms, which detects all interesting subspaces  of high-dimensional data containing clusters. For this purpose, we  define a quality criterion for the interestingness of a subspace and propose  an e#cient algorithm called RIS (Ranking I nteresting Subspaces)  to examine all such subspaces. A broad evaluation based on synthetic  and real-world data sets empirically shows that RIS is suitable to find  all relevant subspaces in large, high dimensional, sparse data and to rank  them accordingly.
Using Sets of Feature Vectors for Similarity Search on Voxelized CAD Objects.
In modern application domains such as multimedia, molecular biology and medical imaging, similarity search in database  systems is becoming an increasingly important task. Especially for CAD applications, suitable similarity models can help to reduce the cost of developing and producing new parts by maximizing the reuse of existing parts. Most of the existing similarity models are based on feature vectors. In this paper, we shortly review three models which pursue this paradigm. Based on the most promising of these three models, we explain how sets of feature vectors can be used for more e#ective and still e#cient similarity search. We first introduce an intuitive distance measure on sets of feature vectors together with an algorithm for its e#cient computation. Furthermore, we present a method for accelerating the processing of similarity queries on vector set data. The experimental evaluation is based on two real world test data sets and points out that our new similarity approach yields more meaningful results in comparatively short time.
A Computational Biology Database Digest Data
Computational Biology or Bioinformatics has been defined as the application of mathematical and Computer Science methods to solving problems in Molecular Biology that require large scale data, computation, and analysis [26]. As expected, Molecular Biology databases play an essential role in Computational Biology research and development. This paper introduces into current Molecular Biology databases, stressing data modeling, data acquisition, data retrieval, and the integration of Molecular Biology data from different sources. This paper is primarily intended for an audience of computer scientists with a limited background in Biology. 1
Datenbanken in der Bioinformatik.

Data Bubbles Quality Preserving Performance Boosting for Hierarchical Clustering.
In this paper, we investigate how to scale hierarchical clustering methods (such as OPTICS) to extremely large databases by utilizing data compression methods (such as BIRCH or random sampling). We propose a three step procedure: 1) compress the data into suitable representative objects; 2) apply the hierarchical clustering algorithm only to these objects; 3) recover the clustering structure for the whole data set, based on the result for the compressed data. The key issue in this approach is to design compressed data items such that not only a hierarchical clustering algorithm can be applied, but also that they contain enough information to infer the clustering structure of the original data set in the third step. This is crucial because the results of hierarchical clustering algorithms, when applied naively to a random sample or to the clustering features (CFs) generated by BIRCH, deteriorate rapidly for higher compression rates. This is due to three key problems, which we identify. To solve these problems, we propose an efficient post-processing step and the concept of a Data Bubble as a special kind of compressed data item. Applying OPTICS to these Data Bubbles allows us to recover a very accurate approximation of the clustering structure of a large data set even for very high compression rates. A comprehensive performance and quality evaluation shows that we only trade very little quality of the clustering result for a great increase in performance.
