Citation Context

...random sampling is perhaps the most simple and intuitive algorithm for data reduction, aiming to capture the most significant features of the append-only stream. Reservoir algorithms such as those in =-=[12]-=- provide a uniform sample and make one pass over the data, taking online decisions on whether to accept or reject an item. However, as we further explain in Section 3, applying such a technique over s...

Citation Context

...r future research. 2 Related Work Several summarization techniques have been proposed for traditional databases, but recent work has focused particularly on data stream synopses [1], such as sampling =-=[2]-=-, histograms [6], wavelets [5], and sketches [4]. Of these, random sampling is perhaps the most simple and intuitive algorithm for data reduction, aiming to capture the most significant features of th...

Citation Context

...can be supported, this method is clearly geared towards current status without retaining the positional footprints of multiple distinct objects over long time periods. Other query-oriented techniques =-=[11, 8]-=- hash incoming points to efficient structures, discarding data irrelevant to the specific query, while our techniques act at a lower level, maintaining the original form of the data. 3 Problem Formula...

Citation Context

...h. 2 Related Work Several summarization techniques have been proposed for traditional databases, but recent work has focused particularly on data stream synopses [1], such as sampling [2], histograms =-=[6]-=-, wavelets [5], and sketches [4]. Of these, random sampling is perhaps the most simple and intuitive algorithm for data reduction, aiming to capture the most significant features of the append-only st...

Citation Context

...ork Several summarization techniques have been proposed for traditional databases, but recent work has focused particularly on data stream synopses [1], such as sampling [2], histograms [6], wavelets =-=[5]-=-, and sketches [4]. Of these, random sampling is perhaps the most simple and intuitive algorithm for data reduction, aiming to capture the most significant features of the append-only stream. Reservoi...

Citation Context

...can be supported, this method is clearly geared towards current status without retaining the positional footprints of multiple distinct objects over long time periods. Other query-oriented techniques =-=[11, 8]-=- hash incoming points to efficient structures, discarding data irrelevant to the specific query, while our techniques act at a lower level, maintaining the original form of the data. 3 Problem Formula...

Citation Context

...e of the techniques carefully maintain good samples along trajectories, their time complexity is O(N) per position update, where N is the current size of a trajectory. A similar approach wasstaken in =-=[3]-=-, offering analytical results also based on line simplification, but this time with respect to spatial query answering (e.g., spatial join, nearest neighbor search) and with bounded error guarantees. ...

Citation Context

...ite satisfactory for moving objects databases, generally entail high cost when applied on streaming locations. The insightful idea of amnesic approximation for streaming time series was introduced in =-=[10]-=-, considering that the importance of each measurement generally decays with time. Their time complexity is independent of the total stream size, but they only handle one-dimensional time series. In th...

Citation Context

...ectories does not take into consideration spatiotemporal peculiarities pertaining to the movement of distinct objects. The necessity of compression techniques for moving objects has been set forth in =-=[7]-=-. Inspired by line generalization that has been widely accepted on spatial data, several off-line and online spatiotemporal extensions were suggested to approximate trajectories, also providing a deta...

Citation Context

...ization techniques have been proposed for traditional databases, but recent work has focused particularly on data stream synopses [1], such as sampling [2], histograms [6], wavelets [5], and sketches =-=[4]-=-. Of these, random sampling is perhaps the most simple and intuitive algorithm for data reduction, aiming to capture the most significant features of the append-only stream. Reservoir algorithms such ...

Citation Context

...ase for trajectories is different due to characteristics inherent in movement: not only spatial locations, but speed and orientation must also not be overlooked when approximating trajectories. SCUBA =-=[9]-=- is a recently suggested technique that maintains clusters of closely located objects and queries moving along a network. Although it is stated that trajectory queries can be supported, this method is...

Citation Context

...ct an acceptable trace of the entire movement. Usually, it is assumed that line segments connect each pair of successive points in the sequence, although a curve-based representation is also possible =-=[13]-=- through higher-order polynomials or splines. This technique is used to provide synchronized trajectory representations by reconstructing original positional measurements at a fixed rate. In practice,...