N. Kline and R. Snodgrass, "Computing Temporal Aggregates", Proc. of ICDE, 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....array. Temporal aggregation can then be performed independently for each interval, using any algorithm. Results for all intervals are combined together and with the meta array. This algorithm is complementary to our approach and could be used to parallelize our algorithms. Kline and Snodgrass [12] developed a structure called the aggregation tree based on the binary segment tree [16] Aggregation trees support incremental computation of temporal aggregates. In particular, their segment tree features allow efficient processing of tuples with long valid intervals. This point will be ....

.... usable as index (lookup time) support for cumulative aggregates basic [20] all disk O(n ) no no no balanced tree SUM COUNT AVG memory O(n log n) no no no endpoint sort (Section 5) SUM COUNT AVG disk O(n log n) no no no merge sort MIN MAX disk O(n log n) no no no aggregation tree [12] all memory O(n ) O(n) O(n) no if k ordered) no SB tree (Sections 3, 4.1) all disk O(n log n) O(log n) O(log n) fixed window offset dual SB trees and JSB tree SUM COUNT AVG disk O(n log n) O(log n) O(log n) arbitrary window offset MSB tree MIN MAX disk O(n log n) O(log n) O(log n) ....

N. Kline and R. T. Snodgrass. Computing temporal aggregates. In Proc. of the 1995 Intl. Conf. on Data Engineering, pages 222--231, Taipei, Taiwan, March 1995.

....a aP tree that uses O(n log B n) disk blocks and answers a range aggregate query using O(log B n) I Os. As temporal aggregation is being included in most of the temporal languages, there has been a flurry of activity on answering temporal aggregate queries. After several early results, including [14, 13], Yang and Widom [21] proposed an indexing scheme called the SB tree that stores a set of N time intervals in R using O(n log B n) disk blocks so that for a time interval , the aggregate over all keys that are valid at some time in the interval can be computed using O(log B n) I Os. ....

N. Kline and R. T. Snodgrass. Computing temporal aggregates. In Proc. of Intl conference on Data Engineering, pages 222--231, 1995.

....computes a temporal aggregate in O(mn) time, where m is the number of result tuples (at worst, m is O(n) but in practice it is usually much less than n) Note that this two step approach can be used to compute range temporal aggregates, however the full database scans makes it inefficient. [KS95] used the aggregation tree, a main memory tree (based on the segment tree [PS85] to incrementally compute temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. KS95] also presented a variant of the ....

....full database scans makes it inefficient. KS95] used the aggregation tree, a main memory tree (based on the segment tree [PS85] to incrementally compute temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. [KS95] also presented a variant of the aggregation tree, the k ordered tree, which is based on the k orderness of the base table; the worst case behavior though remains O(n) GHR 99, YK97] introduced parallel extensions to the approach presented in [KS95] MLI00] presented an improvement by ....

[Article contains additional citation context not shown here]

N. Kline and R. Snodgrass, "Computing Temporal Aggregates", Proc. of ICDE, 1995.

....they are more general; thus the term temporal aggregation implies cumulative temporal aggregation . Furthermore, in this paper we focus on the SUM aggregate but our solutions apply to COUNT and AVG as well. Many approaches have been recently proposed to address temporal aggregation queries [28,19, 29,11,22,30,31]. They are classi ed into two categories: approaches that compute a temporal aggregate when the aggregate is requested (usually by sweeping through related data) and those that maintain a specialized aggregate index [30,31] The latter approaches dynamically precompute aggregates and store them ....

....approach computes a temporal aggregate in O(mn) time, where m is the number of result tuples (at worst, m is O(n) but in practice it is usually much less than n) Note that this two step approach can be used to compute range temporal aggregates, however the full database scans make it inecient. [19] used the aggregation tree, a mainmemory tree (based on the segment tree [23] to incrementally compute instantaneous temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. 19] also presented a variant of the ....

[Article contains additional citation context not shown here]

N. Kline and R. Snodgrass, \Computing Temporal Aggregates", Proc. of ICDE, 1995.

....for ef ciently aggregating planar points. 32] addresses the problem to evaluate multiple range sums progressively. Temporal Aggregation and Objects with Extent. The instantaneous temporal aggregation query nds the aggregate value over all records whose intervals contains a given time instant. [20] provided the aggregation tree which incrementally computes temporal aggregates. 17] introduced parallel extensions, while [23] presented an improvement by utilizing a red black balanced tree. The cumulative temporal aggregation query nds the aggregate value over all records whose intervals ....

N. Kline and R. Snodgrass, \Computing Temporal Aggregates ", Proc. of ICDE, 1995.

....they are more general; thus the term temporal aggregation implies cumulative temporal aggregation . Furthermore, in this paper we focus on the SUM aggregate but our solutions apply to COUNT and AVG as well. Many approaches have been recently proposed to address temporal aggregation queries [26, 19, 27, 11, 22, 28, 29]. They are classi ed into two categories: approaches that compute a temporal aggregate when the aggregate is requested (usually by sweeping through related data) and those that maintain a specialized aggregate index [28, 29] The latter approaches dynamically precompute aggregates and store them ....

....approach computes a temporal aggregate in O(mn) time, where m is the number of result tuples (at worst, m is O(n) but in practice it is usually much less than n) Note that this two step approach can be used to compute range temporal aggregates, however the full database scans make it inecient. [19] used the aggregation tree, a main memory tree (based on the segment tree [23] to incrementally compute instantaneous temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. 19] also presented a variant of the ....

[Article contains additional citation context not shown here]

N. Kline and R. Snodgrass, \Computing Temporal Aggregates", Proc. of ICDE, 1995.

....they are more general; thus the term temporal aggregation implies cumulative temporal aggregation . Furthermore, in this paper we focus on the SUM aggregate but our solutions apply to COUNT and AVG as well. Many approaches have been recently proposed to address temporal aggregation queries [26, 19, 27, 11, 22, 28, 29]. They are classi ed into two categories: approaches that compute a temporal aggregate when the aggregate is requested (usually by sweeping through related data) and those that maintain a specialized aggregate index [28, 29] The latter approaches dynamically precompute aggregates and store them ....

....approach computes a temporal aggregate in O(mn) time, where m is the number of result tuples (at worst, m is O(n) but in practice it is usually much less than n) Note that this two step approach can be used to compute range temporal aggregates, however the full database scans make it inecient. [19] used the aggregation tree, a main memory tree (based on the segment tree [23] to incrementally compute instantaneous temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. 19] also presented a variant of ....

[Article contains additional citation context not shown here]

N. Kline and R. Snodgrass, \Computing Temporal Aggregates", Proc. of ICDE, 1995.

....approach computes a temporal aggregate in O(mn) time, where m is the number of result tuples (at worst, m is O(n) but in practice it is usually much less than n) Note that this twostep approach can be used to compute range temporal aggregates, however the full database scans make it inecient. [KS95] used the aggregation tree, a main memory tree (based on the segment tree [PS85] to incrementally compute instantaneous temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. KS95] also presented a variant of ....

....scans make it inecient. KS95] used the aggregation tree, a main memory tree (based on the segment tree [PS85] to incrementally compute instantaneous temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. [KS95] also presented a variant of the aggregation tree, the k ordered tree, which is based on the k orderness of the base table; the worst case behavior though remains O(n) GHR 99, YK97] introduced parallel extensions to the approach presented in [KS95] MLI00] presented an improvement by ....

[Article contains additional citation context not shown here]

N. Kline and R. Snodgrass, \Computing Temporal Aggregates", Proc. of ICDE, pp. 222-231, 1995.

....answers. By traversing the index, a rough approximation is obtained from the values at the higher levels, which is progressively refined as the search continues towards the leaves. Similar structures have also been developed in the context of main memory temporal databases: Kline and Snodgrass [KS95] propose the aggregation tree, for computing aggregations over temporal data. The tree indexes constant intervals, i.e. the maximum continuous intervals where the value of the aggregation function is constant. The nodes store the difference of the aggregation value between the current and the ....

Kline, N., Snodgrass, R. Computing Temporal Aggregates. ICDE, 1995.

....of unnecessary effort, seriously compromising performance. A solution for the problem is to store aggregate information in the nodes of specialized index structures. Such aggregate trees have already been employed in the context of temporal databases for computing aggregates over temporal data [KS95, KKK99, YW01, ZMT 01] In order to improve the performance of WA queries in OLAP applications involving multidimensional ranges, Jurgens and Lenz [JL98] proposed the storage of summarized data in the nodes of the R tree [BKS 90] used to index the fact table. Each entry of the resulting ....

Kline, N., Snodgrass, R. Computing Temporal Aggregates. IEEE ICDE, 1995.

....computes a temporal aggregate in O(mn) time, where m is the number of result tuples (at worst, m is O(n) but in practice it is usually much less than n) Note that this two step approach can be used to compute range temporal aggregates, however the full database scans makes it inefficient. [KS95] uses the aggregation tree, a main memory tree (based on the segment tree [PS85] to incrementally compute temporal aggregates. However the structure can become unbalanced which implies O(n) worstcase time for computing a scalar temporal aggregate. KS95] also presents a variant of the aggregation ....

....full database scans makes it inefficient. KS95] uses the aggregation tree, a main memory tree (based on the segment tree [PS85] to incrementally compute temporal aggregates. However the structure can become unbalanced which implies O(n) worstcase time for computing a scalar temporal aggregate. [KS95] also presents a variant of the aggregation tree, the k ordered tree, which is based on the k orderness of the base table; the worst case behavior though remains O(n) GHR 99, YK97] introduce parallel extensions to the approach presented in [KS95] MLI00] presents an improvement by considering a ....

[Article contains additional citation context not shown here]

N. Kline and R. Snodgrass, "Computing Temporal Aggregates", Proc. of ICDE, pp. 222-231, 1995.

....array. Temporal aggregation can then be performed independently for each interval, using any algorithm. Results for all intervals are combined together and with the meta array. This algorithm is complementary to our approach and could be used to parallelize our algorithms. Kline and Snodgrass [7] developed a data structure called the aggregation tree based on the binary segmenttree [11] Aggregation trees support incremental computation of temporal aggregates. In particular, their segmenttree features allow efficient processing of tuples with long valid intervals. This point will be ....

.... time) support for cumulative aggregates basic [15] all disk O(n 2 ) no no no balanced tree [10] SUM COUNT AVG memory O(n log n) no no no endpoint sort (see full version [17] SUM COUNT AVG disk O(n log n) no no no merge sort [10] MIN MAX disk O(n log n) no no no aggregation tree [7] all memory O(n 2 ) O(n) O(n) no if k ordered) no SB tree (Sections 3 and 4) all disk O(n log n) O(log n) O(log n) fixed window offset dual SB trees (Section 4) SUM COUNT AVG disk O(n log n) O(log n) O(log n) arbitrary window offset MSB tree (Section 4) MIN MAX disk O(n log n) ....

N. Kline and R. T. Snodgrass. Computing temporal aggregates. In Proc. of the 1995 Intl. Conf. on Data Engineering, pages 222--231, Taipei, Taiwan, March 1995.

....computes a temporal aggregate in O(mn) time, where m is the number of result tuples (at worst, m is O(n) but in practice it is usually much less than n) Note that this two step approach can be used to compute range temporal aggregates, however the full database scans makes it inefficient. [KS95] uses the aggregation tree, a main memory tree (based on the segment tree [PS85] to incrementally compute temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. KS95] also presents a variant of the ....

....full database scans makes it inefficient. KS95] uses the aggregation tree, a main memory tree (based on the segment tree [PS85] to incrementally compute temporal aggregates. However the structure can become unbalanced which implies O(n) worst case time for computing a scalar temporal aggregate. [KS95] also presents a variant of the aggregation tree, the k ordered tree, which is based on the k orderness of the base table; the worst case behavior though remains O(n) GHR 99, YK97] introduce parallel extensions to the approach presented in [KS95] MLI00] presents an improvement by considering a ....

[Article contains additional citation context not shown here]

N. Kline and R. Snodgrass, "Computing Temporal Aggregates", Proc. of ICDE, pp. 222-231, 1995.

....go into a meta array. Temporal aggregation can then be performed independently for each interval, using any algorithm. Results for all intervals are combined together and with the meta array. This algorithm is complementary to our approach and could be used to parallelize our algorithms. KS95] developed a data structure called the aggregation tree based on the binary segment tree [PS85] Aggregation trees support incremental computation of temporal aggregates. In particular, their segmenttree features allow efficient processing of tuples with long valid intervals. This point will be ....

....with the aggregation tree is that it is unbalanced. In the worst case, it takes O(n 2 ) to compute a temporal aggregate from a base table with n tuples, O(n) to process an insertion into the base table, and O(n) to perform a lookup of the aggregate value by time. To circumvent the problem, KS95] proposed a variant of the aggregation tree called the k ordered aggregation tree, which takes advantage of the k orderedness of the base table to enable garbage collection of tree nodes. However, garbage collection makes it impossible to use the aggregate tree as an index. Moreover, ....

[Article contains additional citation context not shown here]

N. Kline and R. T. Snodgrass. Computing temporal aggregates. In Proc. of the 1995 Intl. Conf. on Data Engineering, pages 222--231, Taipei, Taiwan, March 1995.

....is the most common and challenging temporal aggregation. Computing instant aggregates is expensive because it is necessary to know which tuples overlap each instant, and simply considering each tuple in order in a sortedby time relation will not be sufficient due to the varying interval lengths [9]. For example, computing the time varying maximum salary of employees involves computing the temporal extent of each maximum value, which requires determining the tuples that overlap each temporal instant. Figure 1(a) shows a sample Employees table with two temporal attributes, which represent ....

....38,000 Accounting 18 21 (a) Input Database Tuples Count Max Begin End 1 35,000 7 8 2 45,000 8 12 1 45,000 12 18 3 46,000 18 20 2 46,000 20 21 1 46,000 21 31 (b) Temporal Aggregation Results Figure 1. Sample Database and Its Temporal Aggregation oped for computing temporal aggregates [7, 9, 12, 14, 15], they suffer from serious limitations such as the size of aggregation restricted by available memory and requirement of a priori knowledge about the orderedness of an input database. In this paper, we propose a variety of temporal aggregation algorithms that overcome major drawbacks of previous ....

[Article contains additional citation context not shown here]

Nick Kline and Richard T. Snodgrass. Computing temporal aggregates. In Proceedings of the 11th Inter. Conference on Data Engineering, pages 222--231, Taipei, Taiwan, March 1995.

....database operators, aggregate functions are expensivebutvery important components. They are essential to statistical tasks and decision support applications. Although there have been several proposals for temporal aggregate processing, the tree based approach is known to be a promising solution [7]. In the tree based approach, a special tree structure is built for the 2 Jong Soo Kim et al. tuples that are filtered by a given query, and the temporal aggregate results for the qualified tuples are computed by using the constructed tree structure. Here the construction of a new tree and the ....

....each different query generates a differentsetof qualified tuples. Therefore, efficient tree construction and aggregate computation techniques are essential to the improvement of the query response time in the tree based approach. As a tree based temporal aggregate method, the aggregation tree[7] was proposed. However, the aggregation tree has several drawbacks. The aggregation tree can be severely skewed if the time intervals of tuples are in a near sorted order. This situation may occur frequently because tuples are likely to be stored in the order of their time values. The space ....

[Article contains additional citation context not shown here]

Kline, N., Snodgrass, R.T.: Computing Temporal Aggregates. In Proc. of the 11th ICDE. (1995) 222--231

No context found.

N. Kline and R. T. Snodgrass. Computing Temporal Aggregates. In Proceedings of IEEE ICDE,Taipei, Taiwan, pp. 222--231 (1995).

No context found.

N. Kline and R. T. Snodgrass. Computing Temporal Aggregates. In Proceedings of IEEE ICDE, Taipei, Taiwan, pp. 222--231 (1995).

....to sort merge join and computes the aggregate values group by group. Figure 7(b) outlines its pseudo code for computing the COUNT aggregate; the code has to be modified slightly for computing other aggregates. The algorithm is different from the temporal aggregation algorithms presented in [KS95], which used aggregation trees in memory or, during computation, maintained lists of constant periods and their running aggregate values. The cost of temporal aggregation in the middleware depends on the size of the argument and of the result (see Figure 6) For simplicity, the complexity of the ....

N. Kline and R. T. Snodgrass. Computing Temporal Aggregates. In Proceedings of IEEE ICDE, Taipei, Taiwan, pp. 222--231 (1995).

....common time period of tuples that have equal values for the grouping attributes. The tuples of each group are sorted on the time attributes in ascending order. The schema of the result relation follows from the definition of concatenation. Our definition corresponds to the definition given in [KS95]. Let us consider an example query that counts the number of employees working on each project (see relation PROJECT in Figure 3) The query is expressed as T Prj;COUNT(EmpName) PROJECT) and the result is shown in Figure 7. Temporal aggregation is defined next. Prj COUNT(EmpName) T1 T2 P1 ....

N. Kline and R. T. Snodgrass. Computing Temporal Aggregates. In Proceedings of IEEE ICDE, Taipei, Taiwan, pp. 222--231 (1995).

....extent of each maximum value, which requires determining the tuples that overlap each temporal instant. In this paper, we present several new parallel algorithms for the computation of temporal aggregates on a sharednothing architecture [8] Specifically, we focus on the Aggregation Tree algorithm [7] and propose several approaches to parallelize it. The performance of the parallel algorithms relative to various data set and operational characteristics is of our main interest. The rest of this paper is organized as follows. Section 2 gives a review of related work and presents the sequential ....

.... by Epstein [5] A different approach employing program transformation methods to systematically generate efficient iterative programs for aggregate queries has also been suggested [6] Tumas extended Epstein s algorithms to handle temporal aggregates [9] these were further extended by Kline [7]. While the resulting algorithms were quite effective in a uniprocessor environment, all suffer from poor scale up performance, which identifies the need to develop parallel algorithms for computing temporal aggregates. Early research on developing parallel algorithms focused on the framework of ....

[Article contains additional citation context not shown here]

N. Kline and R. T. Snodgrass. Computing temporal aggregates. In the 11th Inter. Conference on Data Engineering, pages 222--231, Taipei, Taiwan, Mar. 1995.

No context found.

N. Kline and R. Snodgrass, "Computing Temporal Aggregates", Proc. of ICDE, 1995.

No context found.

N. Kline and R. Snodgrass, "Computing Temporal Aggregates", Proc. of ICDE, 1995.

No context found.

N. Kline and R. T. Snodgrass. Computing Temporal Aggregates. In Proc. ICDE, pages 222--231, 1995.

No context found.

N. Kline and R. T. Snodgrass. Computing temporal aggregates. In Proc. of Intl conference on Data Engineering, pages 222--231, 1995.

First 50 documents

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