Ki-Joune Li

A spatial data mining method by Delaunay triangulation

It becomes an important task to discover significant pattern or characteristics which may implicitly exist in huge spatial dntabases, such as geographical or medical databases. In this paper, we present a spatial data mining method named SMiYN (Spatial data Mining by Triangulated Irregular Network), which is based on Delaunay Triangulation. Sh47ZN demonstrates important advantages over the previous works. First, it discovers even sophisticated pattern like nested doughnuts, and hierarchical structure of cluster distribution. Second, in order to execute SMTIN, we do not need to know a priori the nature of distribution, for example the number of clusters, which is indispensable to other methods. Third, experiments show that SMTIN requires less CPU processing time than other methods such as BIRCH and CLARANS. Finally it is not ordering sensitive and handles efticiently outliers.

Spatio-temporal similarity analysis between trajectories on road networks

In order to analyze the behavior of moving objects, a measure for determining the similarity of trajectories needs to be defined. Although research has been conducted that retrieved similar trajectories of moving objects in Euclidean space, very little research has been conducted on moving objects in the space defined by road networks. In terms of real applications, most moving objects are located in road network space rather than in Euclidean space. In this paper, we investigate the properties of similar trajectories in road network space. And we propose a method to retrieve similar trajectories based on this observation and similarity measure between trajectories on road network space. Experimental results show that this method provides not only a practical method for searching for similar trajectories but also a clustering method for trajectories.

Trajectory Analysis for Soccer Players

In order to make good strategies, soccer coaches analyze the archives of matches, which can be effectively considered as a set of trajectories. We can extract several useful information by analyzing the trajectories of moving objects, which consist of 22 players and a ball. Since each moving object interacts with others and produces a trajectory, its trajectory has a certain number of relationships with others, which are a basic type of information to make soccer strategies. In this paper, we propose a model to quantitatively express the performance of soccer players. This model is based on the relationships between trajectories of 22 players and a ball and allows to evaluate the performance of several players in quantitative way

Indoor Space: A New Notion of Space

With the rapid progress of GIS and ubiquitous computing technologies, the space that we are dealing with is no longer limited to outdoor space but being extended to indoor space. Indoor spaces has a number of differences from outdoor space, which make it difficult to realize GIS for indoor space. In order to extend the scope of GIS to indoor space, and provide integrated and seamless services, it is required to establish new theories, data models, database management systems for indoor spatial data, and applications for indoor space. In this paper, we investigate the research issues of indoor space and indoor GIS.

Fast indexing and updating method for moving objects on road networks

Fast updates in spatial databases become a crucial issue in several application domains, such as LBS (location-based service). In particular, moving objects with frequent updates on their positions require fast update mechanisms in spatiotemporal database systems. In this paper, we propose a new method for the indexing of the current positions of moving objects on road network. Our method significantly improves not only the update cost but also the performance of query processing. The only constraint of our method is that the moving objects should be on roads, but we believe that most applications dealing moving objects imply this constraint. We provide a performance analysis on our model and by this we suggest further improvements on performance of our method. By experimental studies, we show that our method takes about 40% less in disk accesses compared to TPR-tree for updating and at maximum 4 times faster than TPR-tree for processing timestamp range queries.

Spatial and spatiotemporal analysis of soccer

Soccer matches consist of spatial and spatiotemporal objects including a ball, players, and a field. Players and ball are also moving objects. This means that spatial and spatiotemporal analysis on ball and players could be useful in studying soccer tactics. From this view point, we extend the application areas of spatial information science to soccer tactic analysis. First we define a feature model to specify the basic units of analysis. Second, we study the morphological properties of each feature type. Third, a set of operations are defined for each feature type. Fourth, we perform an experiment with a real data set of the 2006 World Cup final match to validate the usefulness of our framework. Even though we focus on soccer match, we expect that this framework can be applied to similar sports such as handball and basket ball.

Integrating IndoorGML and CityGML for Indoor Space

Recent progress on indoor positioning and mobile devices allows to provide indoor spatial information services such as indoor LBS or indoor disaster management. In order to realize these services, indoor maps are a crucial and expensive component of the system. For this reason, the interoperability among services and sharing indoor map and spatial information is a fundamental requirement of the indoor spatial information system. Several geospatial standards have been and being developed to meet this requirements, among which CityGML LoD 4 (Level of Detail 4) and IndoorGML are the most relevant ones for indoor spatial information. However the objectives and scope of these standards are different although their integration may give a synergy effect. In this paper, we discuss the issues on the integration of IndoorGML and CityGML LoD 4 and propose two methods: automatic derivation of IndoorGML data from CityGML LoD 4 data set and external references from IndoorGML instance to an object in CityGML data. The derivation and reference of external objects are based on the mapping relationship between feature types in CityGML and IndoorGML to be investigated in this paper. A simple prototype will be also presented, which has been developed to validate our methods.

Searching for similar trajectories on road networks using spatio-temporal similarity

In order to search similar moving object trajectories, the previously used methods focused on Euclidean distance and considered only spatial similarity. Euclidean distance is not appropriate for road network space, where the distance is limited to the space adjacent to the roads. In this paper, we consider the properties of moving objects in road network space and define temporal similarity as well as spatio-temporal similarity between trajectories based on POI (Points of Interest) and TOI (Times of Interest) on road networks. Based on these definitions, we propose methods for searching for similar trajectories in road network space. Experimental results show the accuracy of our methods and the average search time in query processing.

Topological Consistency for Collapse Operation in Multi-scale Databases

When we derive multi-scale databases from a source spatial database, the geometries and topological relations, which are a kind of constraints defined explicitly or implicitly in the source database, are transformed. It means that the derived databases should be checked to see if or not the constraints are respected during a derivation process. In this paper, we focus on the topological consistency between the source and derived databases, which is one of the important constraints to respect. In particular, we deal with the method of assessment of topological consistency, when 2-dimensional objects are collapsed to 1-dimensional ones. We introduce 8 topological relations between 2-dimensional objects and 19 topological relations between 1-dimensional and 2-dimensional objects. Then we propose four different strategies to convert the 8 topological relations in the source database to the 19 topological relations in the target database. A case study shows that these strategies guarantee the topological consistency between multi-scale databases.

The Spatial Locality and a Spatial Indexing Method by Dynamic Clustering in Hypermap System

The rapid accessibility of information is an important requirement of the hypermap systems or the geographical informations systems. But the massive volumes of data oblige us to store it in secondary memory which greatly slows down the access time. Thus the indexing technique, which determines the way of secondary memory access, is an essential point in hypermap systems. There are several researches on spatial indexing techniques, but due to the lack of comparative studies which must be based on the analysis of the nature of the spatial data, it is difficult to say that one is better than another.