### TABLE I COMPARISON BETWEEN ALGEBRAIC AND SAMPLING BASED APPROACH

### Table III. STRAT versus Other Sampling-Based Approaches AQP Leverages Models Lifted Handles Data Handles Optimizes

in and

### Table 2: Algorithm for Term-based sampling approach

### Table 10: Estimated values using acceptance sampling based on Weak simulation runs, with the strong prior as revised prior.

"... In PAGE 23: ... However, this would introduce dependency between the simulation runs, and the e ective number of observations in the sample would not increase correspondingly. As shown in Table10 12 only part of the original simulation runs are retained. The proportion retained ranged from 20:6% to 3:9% and corresponds to the average weight value, and depends on the di erence between the sampling and the target density.... In PAGE 24: ... However, this is because some of the observations are replicated. The apos;e ective apos; number of observations is comparable to the accepted observations in Table10 12. As shown in Table 13 15 the results using this approach are similar to the previous.... ..."

### Table 2: The breakdown of the query processing time of the conversion approach into the range VPT traversal time, the sequential search time, and the overhead. All performance measurements are in seconds. Figure 6 shows the speedup of the sampling approach with respect to the exact approach versus the sampling step size used in the sampling-based distance function computation. Again there are 13

"... In PAGE 14: ... In the end, the bene t of reducing individual distance function computation time is overshadowed by the performance penalty due to more distance func- tion calls. Table2 shows that this is indeed the case by providing a detailed breakdown of the query computation time. The Total Time is the sum of the range VPT traversal time, which is the product of the number of distance function calls and the average distance function computation time, the Sequential Search Time, which is the last linear search phase of range VPT traversal using the original distance function, and the Overhead.... ..."

### Table 2. Normalization factors for Leg 168 bulk-powder samples, based on XRD pe using 11 standard mixtures.

"... In PAGE 1: ... Figure 2 shows the positions and angular ranges of all peaks used in quantitative analysis. The numerical technique of Fisher and Un- derwood (1995) allows one to assign either positive or negative nor- malization factors to relate each indicator mineral to each target phase ( Table2 ). Normalization factors for all shipboard analyses were derived from 11 mixtures of mineral standards, each of which 1Davis, E.... In PAGE 2: ...inerals is 0.8% (see CD-ROM, back pocket, for additional data). We also tested the validity of a model in which all normalization fac- tors must be positive in sign, but this approach resulted in a weaker match between calculated and measured weight percentages of the mineral standards. A separate set of normalization factors was gener- ated for samples analyzed at the University of Lille after Leg 168 dances of samples from Sites 1030, 1031, and 1032 ( Table2 ), and they are based on analyses of splits of the 11 shipboard mixtures of mineral standards. Sixteen clay-sized fractions from Site 1027 were analyzed in ad- ditional detail.... ..."

### TABLE I SAMPLING-BASED ROADMAP OF TREES (SRT) ALGORITHM.

2005

Cited by 8

### TABLE III PARALLEL SAMPLING-BASED ROADMAP OF TREES (SRT) ALGORITHM.

2005

Cited by 8

### Table 8: Selected terms generated for each collection using the new proposed term-based random sampling approach

in Edited by:

2005

"... In PAGE 87: ...0775 0.0909 Table8 : The precision/recall value for correction of character reversal errors (.GOV).... ..."