Search

Detecting changes in spatial datasets is important for many fields such as early warning systems that monitor environmental conditions, epidemiology, crime monitoring, and automatic surveillance. The goal of the presented research is the development of methodologies for change analysis in spatial datasets. An approach to change analysis is presented that analyses how the interesting regions in one time frame differ from the interesting regions in the next time frame with respect to a given interestingness perspective. Users are allowed to define their own interestingness perspectives that are captured in form of a reward-based fitness function. A change analysis framework is proposed in which reward-based fitness functions are maximized by clustering algorithms to find interesting regions for each dataset. Next, the relationships between interesting regions are analyzed. Two approaches to analyze correspondence between interesting regions are proposed. The first approach directly compares cluster models and strongly relies on analyzing relationships between polygons that capture the scope of an interesting region. The second approach compares cluster extensions that are computed by re-clustering. Finally, a knowledge base of change predicates is provided that allows analyzing various aspects of change, and report generators that are associated with particular predicates are proposed to generate summaries of change. Moreover, in contrast to most existing research, the framework does not assume that the identity of objects in the analyzed environment is known. The proposed framework is demonstrated in a case study that analyzes changes in earthquake patterns.

Polygons can serve an important role in the analysis of geo-referenced data as they provide a natural representation for particular types of spatial objects and in that they can be used as models for spatial clusters. This paper claims that polygon analysis is particularly useful for mining related, spatial datasets. A novel methodology for clustering polygons that have been extracted from different spatial datasets is proposed which consists of a meta clustering module that clusters polygons and a summary generation module that creates a final clustering from a polygonal meta clustering based on user preferences. Moreover, a density-based polygon clustering algorithm is introduced. Our methodology is evaluated in a real-world case study involving ozone pollution in Texas; it was able to reveal interesting relationships between different ozone hotspots and interesting associations between ozone hotspots and other meteorological variables.

Analyzing change in spatial data is critical for many applications including developing early warning systems that monitor environmental conditions, epidemiology, crime monitoring, and automatic surveillance. In this paper, we present a framework for the detection and analysis of patterns of change; the framework analyzes change by comparing sets of polygons. A contour clustering algorithm is utilized to obtain polygon models from spatial datasets. A set of change predicates is introduced to analyze changes between different models which capture various types of changes, such as novel concepts, concept drift, and concept disappearance. We evaluate our framework in case studies that center on ozone pollution monitoring, and on diagnosing glaucoma from visual field analysis.

Domain experts are frequently interested to analyze multiple related spatial datasets. This capability is important for change analysis and contrast mining. In this paper, a novel clustering approach called correspondence clustering is introduced that clusters two or more spatial datasets by maximizing cluster interestingness and correspondence between clusters derived from different datasets. A representative-based correspondence clustering framework and clustering algorithms are introduced. In addition, the paper proposes a novel cluster similarity assessment measure that relies on re-clustering techniques and co-occurrence matrices. We conducted experiments in which two earthquake datasets had to be clustered by maximizing cluster interestingness and agreement between the spatial clusters obtained. The results show that correspondence clustering can reduce the variance inherent to representative-based clustering algorithms, which is important for reducing the likelihood of false positives in change analysis. Moreover, high agreements could be obtained by only slightly lowering cluster quality.

This paper addresses two main challenges for clustering which require extensive human effort: selecting appropriate parameters for an arbitrary clustering algorithm and identifying alternative clusters. We propose an architecture and a concrete system MR-CLEVER for multi-run clustering that integrates active learning with clustering algorithms. The key hypothesis of this work is that better clustering results can be obtained by combining clusters that originate from multiple runs of clustering algorithms. By defining states that represent parameter settings of a clustering algorithm, the proposed architecture actively learns a state utility function. The utility of a parameter setting is assessed based on clustering run-time, quality and novelty of the obtained clusters. Furthermore, the utility function plays an important role in guiding the clustering algorithm to seek novel solutions. Cluster novelty measures are introduced for this purpose. Finally, we also contribute a cluster summarization algorithm that assembles a final clustering as a combination of high-quality clusters originating from multiple runs. Merits of our proposed system are that it is generic and therefore can be used in conjunction with different clustering algorithms, and it reduces human effort for selecting the parameters, for comparing clustering results and for assembling clustering results. We evaluate the proposed system in conjunction with a representative based clustering algorithm namely CLEVER for a challenging data mining task involving an earthquake dataset. The obtained results demonstrate that, in comparison to the best single-run clustering, multi-run clustering discovers solutions of higher quality.

Detecting changes in spatial datasets is important for many fields. In this paper, we introduce a methodology for change analysis in spatial datasets that combines contouring algorithms with supervised density estimation techniques. The methodology allows users to define their own criteria for features of interest and to identify changes in those features between two datasets. Change analysis is performed by comparing interesting regions that have been derived using contour clustering. A novel clustering algorithm called DCONTOUR is introduced for this purpose that computes contour polygons that describe the boundary of a supervised density function at a given density threshold. Relationships between old and new data are analyzed relying on polygon operations. We evaluate our methodology in a case study that analyzes changes in earthquake patterns.