"... In PAGE 8: ...ssignment for this variable. And so forth. Although developed independently, this al- gorithm is, in fact, a special case of preference-based search [5], where the criteria on which search is based form a total order. Performance comparisons are given in Table1 . These are for problems with pa- rameters BO20,10,0.... In PAGE 12: ...variable problems listed in Table1 : (i) CYCSCY = 10, tightness = 0.35, (ii) CYCSCY = 20, tightness = 0.... ..."

"... In PAGE 7: ...1 Main Results The main result is that the multi-probe LSH method is much more space efficient than the basic LSH and entropy- based LSH methods to achieve various search quality levels and it is more time efficient than the entropy-based LSH method. Table2 shows the average results of the basic LSH, entropy- based LSH and multi-probe LSH methods using 100 random queries with the image dataset and the audio dataset. We have experimented with different number of hash tables L (for all three LSH methods) and different number of probes T (i.... In PAGE 7: ...ges to achieve a 0.93 recall. On the other hand, when the same amount of main memory is used by the multi-probe LSH indexing data structures, it can deal with about 60- million images to achieve the same search quality. The results in Table2 also show that the multi-probe LSH method is substantially more space and time efficient than the entropy-based approach.... In PAGE 10: ... datasets are due to the characteristics of the datasets. As shown in Table2 , for the image data, a 0.90 recall corre- sponds to a 1.... ..."

"... In PAGE 12: ...ultimedia searching using the standard textual interface method for general Web searching (i.e., entering the query in the text box). Table6 shows a comparison of the means, standard deviation, and significance test results using data from this research and reported values from [21]. If we view increased query length and increased session length as indicators of searching complexity, we see from Table 6 that the use of a relative simple interface (i.... In PAGE 12: ... Table 6 shows a comparison of the means, standard deviation, and significance test results using data from this research and reported values from [21]. If we view increased query length and increased session length as indicators of searching complexity, we see from Table6 that the use of a relative simple interface (i.e.... In PAGE 12: ...ndependent t-test (critical value of 1.64) at the 0.05 level of significance. The t-test is a parametric evaluation; however, with large sample sizes the t-test is fairly robust to non-normality [40, 41]. From Table6 , of the six comparisons of with and without radio buttons, only the audio was significantly ... ..."

"... In PAGE 4: ...nterface methodology for general Web searching (i.e., text box). Table2 presents a comparison of the means and standard deviation using data from Jansen and colleagues [6]. If we view increased query length and increased session length as indicators of searching complexity, we see from Table 2 that the use of a relative simple interface (i.... In PAGE 4: ...nterface methodology for general Web searching (i.e., text box). Table 2 presents a comparison of the means and standard deviation using data from Jansen and colleagues [6]. If we view increased query length and increased session length as indicators of searching complexity, we see from Table2 that the use of a relative simple interface (i.e.... ..."

"... In PAGE 10: ...it can be resolved in CWBDB7CY hops and reach the destination peer in CWBDB7CYB7BD hops with a high probability. Table2 shows the the number of peers touched, the number of peers foreseen, and the number of messages produced at each hop along the neighbor links. As long as the shortest distance between the query source peer and the pre- destination peer that has a successful local matching is not longer than CWBD hops in the friend overlay plus CWBE hops in the neighbor overlay, this query is satisfied by our algorithm.... ..."

"... In PAGE 6: ... 7. RESULTS Table2 reports efficiency and effectiveness results for the base- line experiments. Content-based search was more efficient than match-based search (only 77 messages per query and 38 nodes reached by each query, compared with 424 and 145, respectively).... In PAGE 7: ... AverageNumberofMessages 0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 0123456789 NofDeactivated CB+Flood CB+CB MB+Flood MB+MB Figure 6: Average number of messages per query for the dif- ferent search mechanisms. 2503 for match-based experiments ( Table2 ); hence experiments that deactivate 3 or more Ultra Peers produce different numbers of new connections in the network. As expected, the higher the num- ber of deactivations, the higher the number of new connections.... ..."