Index- supported Pattern Matching on Tuples of Time- dependent Values
Lately, the amount of mobility data recorded by GPS-enabled (and o ther) devices has increased drastically, entailing the necessity of efficient processing and analysis methods. In many cases, not only the geographic position, but also additional tim e-dependent informa- tion are traced and/or generated, according to the purpose of t he evaluation. For example, in the field of animal behavior research, besides the position of the m onitored animal, biol- ogists are interested in further data like the altitude or the temper ature at every measuring point. Other application domains comprise the names of streets, pla ces of interest, or trans- portation modes that can be recorded along with the geographic po sition of a person. In this paper, we present in detail a framework for analyzing dataset s with arbitrarily many time-dependent attributes. This can be considered as a major ext ension of our previous work, a comprehensive framework for pattern matching on symbo lic trajectories with index support. For an efficient processing of different data types, a var iable number of indexes of four different types that correspond to the data types of the attributes are applied. We demonstrate the expressiveness and efficiency of our approach b y querying a real dataset representing taxi trips in Rome and, particularly, with a broad serie s of experiments using trajectories generated by BerlinMOD combined with geological rast er data.
Nutzung und Vervielfältigung:
Alle Rechte vorbehalten