Classification of Multi-Dimensional Trajectories Using Textual Approximation

Classification and searching time series in the multidimensional space is a big challenge. There has been much research work on this field for many years. TAX (Textual ApproXimation) is a method for searching time series data such as stock or electrocardiogram data. The main idea behind TAX is extract a set of temporal-terms from time series data to approximate using document retrieval methods. The main problem of applying TAX to multidimensional data, such as moving object trajectory data is that how we extract temporal-terms from multidimensional data. In this paper, we propose a novel method to obtain temporal-terms by decomposing original multidimensional data into smaller segmentations, deploying a clustering method to group those segmentations into groups, and assigning a temporal-term for each group. Our research focuses on two-dimensional moving object, representing trajectories and on classification of a large set of the moving object trajectories. The experimental results confirm that our proposed method is effective in multi-multidimensional data classification.