Mining Spatio-Temporal Datasets: Relevance, Challenges and Current Research Directions

Spatio-temporal data usually records the states over time of an object, an event or a position in space. Spatio-temporal data can be found in several application fields, such as traffic management, environment monitoring, weather forecast, etc. In the past, huge effort was devoted to spatial data representation and manipulation with particular focus on its visualisation. More recently, the interest of many users has shifted from static views of geospatial phenomena, which capture its “spatiality” only, to more advanced means of discovering dynamic relationships among the patterns and events contained in the data as well as understanding the changes occurring in spatial data over time. Spatio-temporal datasets present several characteristics that distinguish them from other datasets. Usually, they carry distance and/or topological information, organised as multidimensional spatial and temporal indexing structures. The access to these structures is done through special methods, which generally require spatial and temporal knowledge representation, geometric and temporal computation, as well as spatial and temporal reasoning. Until recently, the research in spatial and temporal data handling has been mostly done separately. The research in the spatial domain has focussed on supporting the modelling and querying along spatial dimensions of objects/patterns in the datasets. On the other hand, the research in the temporal domain has focussed on extending the knowledge about the current state of the system governed by the temporal data. However, spatial and temporal aspects of the same data should be studied in conjunction as they are often closely related and models that integrate the two can be beneficial to many important applications. Indeed the amount of available spatio-temporal datasets is growing at exponential speed and it is becoming impossible for humans to effectively analyse and process. Suitable techniques that incorporate human expertise are required. Data mining techniques have been identified as effective in several application domains. In this chapter we discuss the application of data mining techniques to effectively analyse very large spatio-temporal datasets. Spatio-temporal data mining is an emerging field that encompasses techniques for discovering useful spatial and temporal relationships or patterns that are not explicitly stored in spatio-temporal datasets. Usually these techniques have to deal with complex objects with spatial, temporal and other attributes. Both spatial and temporal dimensions add substantial complexity to the data mining process. Following the above mentioned O pe n A cc es s D at ab as e w w w .in te ch w eb .o rg