"... In PAGE 14: ... This more complex queue implementation adds some extra overhead, but pays back by avoiding unnecessary locks and enabling a higher degree of parallelism instantly. The queries investigated are marked by the parameters given in Table1 . For each con gu- ration, we rst generate the base relations according to the query speci cations, then build the hash tables sequentially and after that, execute the strategy considered.... In PAGE 14: ... The base relation sizes were chosen randomly from our portfolio and one of the ve distribution types was used to generate the attribute value distribution. For each query, all distributions were of the same type; the particular parameters are chosen as given in Table1 . All experiments were carried out on 4 processors.... ..."

"... In PAGE 14: ... This more complex queue implementation adds some extra overhead, but pays backby avoiding unnecessary locks and enabling a higher degree of parallelism instantly. The queries investigated are marked by the parameters given in Table1 . For each conn0cgu- ration, we n0crst generate the base relations according to the query specin0ccations, then build the hash tables sequentially and after that, execute the strategy considered.... In PAGE 14: ... The base relation sizes were chosen randomly from our portfolio and one of the n0cve distribution types was used to generate the attribute value distribution. For each query, all distributions were of the same type; the particular parameters are chosen as given in Table1 . All experiments were carried out on 4 processors.... ..."

"... In PAGE 1: ... Relevant documents corre- sponded these topics were given by the Executive Committee. Table3 describes key parameters used in query expansion and their values utilized in the formal runs. The aim of using virtual word collocations is to catch context more accurately.... ..."

"... In PAGE 11: ... In the following, we present our analytical cost model for estimating the cost of executing a query. Table1 lists some database parameters used in the cost analysis.... ..."

"... In PAGE 50: ...00 7.00 0 100 200 300 400 500 600 700 800 900 Number of Food Items R e s p o n s e Ti m e ( s ec) hash join strategy full semantic enrichment Figure 4a: Performance comparison of four query execution strategies Figure 4b: Performance com- parison of hash join and full semantic enrichment execution strategies The diagrams are based on the experimental results in Table2 and the experiment was made by using the real values to actually retrieve the results through web service operations. VG, NF, S1, S2, S3, and S4 denote the value used for parameter gpcode, ... In PAGE 51: ... This is clearly very slow. Table2 . Experimental results VG NF S1 S2 S3 S4 0900 303 1985.... In PAGE 69: ... As a consequence this is the best source for the user profile creation. Table2 . Lattes CV Metadata Element Subset, adapted from [13].... In PAGE 70: ...Table2 shows a Lattes CV metadata elements subset. It presents the categories used in this work to support the recommendation process and their associated descriptions.... In PAGE 99: ... The times were averaged with the first run eliminated to ensure that results are warm cache numbers. Table2 displays the execution times for eXist, the Extended Schema Repos- itory (ESR), and in the final column the factor at which the ESR out-performs eXist. A value of 1 indicates that both are equal and anything less than 1 indi- cates that the ESR is slower than eXist.... ..."

"... In PAGE 14: ... Additionally, user adjustable parameters are provided for the simulation such as execution length, the number of mobile hosts and their query frequency, the number of POIs, etc. Table2 lists all of the simulation parameters. The simulation is initialized by randomly choosing a starting location for each MH within the simulation area.... ..."

"... In PAGE 12: ...31 1 0 R*-treeSR 101.71 1 0 Table 4 Performance results obtained from the building phase of indexing structure Table5 gives the performance results for the multidimensional access methods that were obtained from executing queries of type point query. These results indicated that the efficiency... ..."

"... In PAGE 8: ... The times were averaged (with the first run elimination) to ensure that all results are warm cache numbers. Table2 displays the execution times for each of the seven queries in each support mode. Some of the queries (Q2, Q6) in the eXist mode failed to return any results (R2, R6), as eXist continually returned an out of memory error.... ..."