Y. Manolopoulos, G. Kapetanakis, Overlapping B trees for temporal data, in: Proceedings Fifth Jerusalem Conference on Information Technology (JCIT#90) Jerusalem, Israel, 1990, pp. 491--498.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....RT tree grow considerably. Furthermore, time information stored with nodes plays a complementary role and RT trees are not able to answer temporal queries such as find all objects that exist in the database within a given interval . MR trees and HR trees use the concept of overlapping B trees [32]. They have separate index structures for each time point where a change occurs in an object position within the video data. It is space e#cient if the number of objects changing their locations is low because index structures may have some common paths for those objects that have not moved. ....

Y. Manolopoulos and G. Kapetanakis. Overlapping B+ trees for temporal data. In Proceedings of the 5th Jerusalem Conference on Information Technology (JCIT), pages 491--498, 1990.

....bounding regions that expand with the growing data regions. An evaluation of three representations of the R tree, 3 D R tree [37] 2 3 R tree and HR tree [20] is provided in [21] Tzouramanis et. al [39] use the overlapping linear quadtree approach based on the concept of overlapping trees [19]. Other related research from moving objects include topics such as uncertainty in positions of moving objects. Wolfson et. al [41] discuss this imprecision problem and how DBMS can provide a bound on the error. Pfoser and Jensen [25] point out that given a speed limit, the possible locations ....

Y. Manolopoulos and G. Kapetanakis. Overlapping B+ Trees for Temporal Data. In Proceedings of the 6th JCIT Conference, 1990.

....role. This makes RT tree inefficient when handling time related queries, as it is likely that all the objects satisfying the spatial predicate are retrieved before being filtered according to the temporal predicate. Historical R trees (HR trees) NS98] apply the concept of overlapping B trees [MK90] to R trees. The main idea is to construct an R tree for each timestamp in history. However, since consecutive R trees can make use of common paths if objects do not change their positions, new branches are created only for objects that have moved. HR trees are very efficient for timestamp ....

Manolopoulos, Y., Kapetanakis, G. Overlapping B+trees for Temporal Data. JCIT, 1990.

....to two branches 2 old version newversion shared shared old shared new root of root of Figure 1: EXODUS tree versioning of the design. Some of the subtrees are shared and some are exclusively owned by only one of the trees. A similar idea was used in the overlapping B tree [MK90] In one of the two new structures we introduce, the BT Forest, the branching is supported by an EXODUS like discipline with new roots for each branch. The temporal aspect is supported by the Write Once B tree or WOBT ( Eas86] Although the WOBT was originally designed to work on ....

Y. Manolopoulos and G. Kapetanakis. Overlapping B + -trees for temporal data. In Proc. 5th JCIT Conf., pages 491--498, Jerusalem, Israel, 1990.

....a new version. Temporal data structures (without branching) are partially persistent. The idea of overlapping trees is an old one. Overlapping trees of disk pages were used for shadow page recovery in System R [AO76] They were used to construct a temporal structure from B trees in ( CDRS86, MK90] However they have never before been used to build a fully persistent access structure from partially persisitent access structures. In the BT Forest, the basic building blocks are already temporal structures and the overlapped structure is branched and temporal. To do this, pages are not ....

Y. Manolopoulos and G. Kapetanakis. Overlapping B + -trees for temporal data. In Proc. 5th JCIT Conf., pages 491--498, Jerusalem, Israel, 1990.

....have some (potentially many) identical nodes. The HR tree explores this, by keeping all previous states of the two dimensional R tree only logically instead of physically. This is achieved by allowing consecutive instances of R tree to overlap, i.e. to share nodes. This idea was also proposed in [MK90] but for B trees in the context of (single dimension) temporal databases. As an illustration consider the two consecutive (with respect to their timestamps) R trees in Figure 2(a) and (b) which can be represented in a much more compact manner as the HR tree shown Figure 2(c) In this ....

Y. Manolopoulos and G. Kapetanakis. Overlapping B + -trees for temporal data. In Proc. of the 5th Jerusalem Conf. on Information Technology, pages 491--498, August 1990.

....that the advantage of using one or the other approach is a matter of trade off based on how much often purely spatial or temporal queries are posed relatively to spatiotemporal ones. The Overlapping Linear Quadtrees, the MR trees and the HR trees are all based on the concept of overlapping trees [MK90] The basic idea is that, given two trees where the younger one is an evolution 1 of the older one, the second one is represented incrementally. As such only the modified branches are actually stored, the branches that do not change are simply re used (we discuss such an idea in more details ....

Y. Manolopoulos and G. Kapetanakis. Overlapping B + -trees for temporal data. In Proc. of the 5th Jerusalem Conf. on Information Technology, pages 491--498, August 1990.

....that the advantage of using one or the other approach is a matter of trade off based on how much often purely spatial or temporal queries are posed relatively to spatiotemporal ones. The Overlapping Linear Quadtrees, the MR trees and the HR trees are all based on the concept of overlapping trees [MK90] The basic idea is that, given two trees where the younger one is an evolution 1 of the older one, the second one is represented incrementally. As such only the modified branches are actually stored, the branches that do not change are simply re used (we discuss such an idea in more details ....

Y. Manolopoulos and G. Kapetanakis. Overlapping B + -trees for temporal data. In Proc. of the 5th Jerusalem Conf. on Information Technology, pages 491--498, August 1990.

....space in the media where they reside. In fact the data structure would lead one to see the shared VTT as in Figure 18. In the case of Figure 17 the trade off is that maintainance of such shared tree is not trivial, but is accomplishable. There has been some work done in overlapping B trees [MK90, BHK85, B 90] that can be used as a baseline for the above idea. Another work that is logically equivalent, although structurally different, is the based on Fat Node and Fat Field and was presented in Index Time Valid data . data data Index Time Valid data . data data . T2 T1 T0 ....

Y. Manolopoulos and G. Kapetanakis. Overlapping B+--trees for temporal data. In The 5th Jerusalem Conference on Information Technology, pages 491--498, 1990.

No context found.

Y. Manolopoulos, G. Kapetanakis, Overlapping B trees for temporal data, in: Proceedings Fifth Jerusalem Conference on Information Technology (JCIT#90) Jerusalem, Israel, 1990, pp. 491--498.

No context found.

Y. Manolopoulos, G. Kapetanakis, Overlapping B trees for temporal data, in: Proceedings Fifth Jerusalem Conference on Information Technology (JCIT#90) Jerusalem, Israel, 1990, pp. 491--498.

....or variations which are suitable for multicolored images. 2.3. The concept of overlapping The technique of overlapping was initially presented by Burton et al. 23, 24] The authors proposed the use of overlapping trees to manage the evolution of text files. Later the idea was generalized in [25] to manage temporally evolving B trees. Recently, in [20] the idea was extended, THE COMPUTER JOURNAL, Vol. 43, No. 4, 2000 328 T. TZOURAMANIS,M.VASSILAKOPOULOS AND Y. MANOLOPOULOS FIGURE 2. a) Two similar binary images, b) the respective quadtrees and (c) the corresponding overlapping ....

Manolopoulos, Y. and Kapetanakis, G. (1990) Overlapping B + -trees for temporal data. In Proc. 5th Jerusalem Conf. on Information Technology (JCIT'90), Jerusalem, Israel, pp. 491--498. IEEE Press.

....1 1 1 A A A A A A i i i i i i i i i i i i i i i i ( Gamma Gamma Gamma Gamma Gamma Gamma C C C C C C Figure 1: The Overlapping B trees structure. Example Figure 1 shows how this structure was known until today [2, 3, 14]. More specifically, in this OB trees 2 Note that the term TTT has also been used in [17] yet with a different meaning. instance the index and the data node capacity equals 4. Every page contains an additional field reference counter (refcounter) which is used to register the number of ....

....are known in advance. Finally, T tree index nodes are treated differently from leaves, since whenever an internal node changes to historical, no modification ( disk operation) is required. 2 4 Query Processing in OB trees The differences of this new OB trees version over the older one [2, 3, 14] are the following: ffl The space increases only marginally. It is still O(nlog Bn) since new pages are created for the path leading to each leaf which is updated during a new transaction time. However, there is an additional structuring via new pointer fields. ffl Even though more page accesses ....

Y. Manolopoulos and G. Kapetanakis: Overlapping B + -trees for Temporal Data, Proceedings of the 5th JCIT Conference, pp.491-498, Jerusalem, Israel, October 1990.

No context found.

Y. Manolopoulos and G. Kapetanakis. Overlapping B + -trees for temporal data. In Proc. of the 5th Jerusalem Conf. on Information Technology, pages 491#498, August 1990.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC