P. M. Aoki. Algorithms for index-assisted selectivity estimation. In ICDE, page 258, 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....complete distance matrix to be given, since it only computes those distances when necessary. Algorithm StringMap Input: N strings: t[1; N ] d: Dimensionality of Euclidean space. M : Metric function on strings. Output: N corresponding objects in the new space. Variables: PA[1,2][1, N] 2 d pivot strings. coord[1, N] 1, d] object coordinates. Method: for (h = 1 to d) f (p1 ; p2) ChoosePivot(h,M) choose pivot strings PA[1,h] p1 ; PA[2,h] p2 ; store them dist = GetDistance(p1 , p2 , h, M) if (dist = 0) f set al..l coordinates in ....

....step of testing other attributes takes relatively much less time than the step of finding the candidate record pairs, thus we mainly focus on the time of doing the similarity join that finds the candidate pairs. We can use existing techniques on estimating the performance of spatial joins (e.g. [2, 19]) and choose the attribute that takes the least time to do the corresponding similarity join. This attribute is called the most selective attribute for this conjunctive clause. Notice that similar to [15] we could also search along multiple attributes of the conjunction to improve accuracy. ....

P. M. Aoki. Algorithms for index-assisted selectivity estimation. In ICDE, page 258, 1999.

No context found.

P.M. Aoki, "Algorithms for Index-Assisted Selectivity Estimation," Tech. Rep. UCB//CSD-98-1021, Univ. of California, Berkeley, CA, Oct. 1998.

....traversal and index assisted sampling. Sections 4 and 5 explain our experimental infrastructure, procedures and results. Section 6 reviews related work. We conclude in Section 7. Additional algorithmic issues, experimental results and discussions of future work are contained in the full paper [4]. 2. Background and assumptions In this section, we briefly review the concepts and assumptions that underlie our approach. First, we give an overview of the approach. We then describe the specific index structure used. Specifically, we discuss the concepts of the generalized search tree and the ....

....level of update overhead. node uc S c 0 S c S u S (a) 5 13 31 26 (c) 15 7 12 20 13 (b) 11 9 12 17 8 (f) 3 9 11 14 5 (g) 2 7 10 12 3 (e) 2 9 9 10 1 (d) 1 9 9 90 (c) b) d) e) f) g) h) a) query (a) b) h) g) c) d) e) f) 5 9 16] 5 8 15] 2 2 3][2 3 4][3 4 6] 2 3 4] 3 4 6] c c 0 c ] valid record c lower bound c 0 center value c upper bound 50 overlap Notes: Node (h) is pruned immediately after node (c) is visited. Ties between nodes with equal u values are broken arbitrarily. a) Tree structure with pseudo ranks ....

[Article contains additional citation context not shown here]

P.M. Aoki, "Algorithms for Index-Assisted Selectivity Estimation," Tech. Rep. UCB//CSD-98-1021, Univ. of California, Berkeley, CA, Oct. 1998.

....histogram, the pseudo ranked tree has precision that depends on the acceptable update overhead. Our results in this paper assume the use of the example formula for c and c in [ANTO92] this formula provides fixed imprecision bounds for a given tree height. In Sections 3.3 and 5. 4 of [AOKI98b], we explain why the imprecision caused by pseudo ranking is easily computed and small in practice. 3 Use in sampling. In the remainder of the paper, we assume that index assisted sampling is implemented using acceptance rejection (A R) sampling applied to pseudo ranked trees. A detailed ....

....overhead. If we never update the index after an initial bulk load, we need never accept any imprecision at all. 4 A R sampling from an unbalanced pseudo ranked tree (and therefore, by immediate extension, from a pseudo ranked forest) returns each record with equal probability (see Appendix D of [AOKI98b]) We will need this result in the next section because sampling from the frontier of a traversal is essentially sampling from a forest. 5 Pseudo ranking implies that the exact size of the population being sampled is unknown. The simplest way of dealing with this is to perform random sampling ....

[Article contains additional citation context not shown here]

P.M. Aoki, "Algorithms for Index-Assisted Selectivity Estimation," Tech. Rep. UCB//CSD-98-1021, Univ. of California, Berkeley, CA, Oct. 1998.

....bad estimates unless enough samples can be obtained) Since the first factor is not known in advance, we must consider adaptive combination algorithms as well as fixed algorithms. The algorithms, experimental results, and an extensive discussion of both background and related work may be found in [1]. In particular, results from an experimental comparison between our estimation algorithms and several multidimensional estimators (i.e. those based on the uniformity assumption, Hausdorff fractal dimension, correlation fractal dimension and density) have been promising. ....

P. M. Aoki. Algorithms for Index-Assisted Selectivity Estimation. Technical Report UCB//CSD-98-1021, University of California, Berkeley, CA, Oct. 1998.

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