"... In PAGE 5: ...able 2: Basic Correspondence between EER and OMT ................................................. 17 Table3 : Advantages and Disadvantages of ER/EER and OO/OMT Models:.... In PAGE 16: ... It promotes database integrity and integration, for the object paradigm helps to bridge the semantic gap between databases and applications (Blaha, Premerlani, Rumbaugh, 1988, 425). Table2 and Table3 on the following two pages summarize the comparative studies on model coverage, model connectivity, model advantages and disadvantages between the ER/EER and the OO/OMT models (Elmasri and Navathe, 1999; Blaha and Premerlani, 1998; Blaha et al., 1988; Navathe, 1992; Sanders, 1995; Gray et al.... ..."

"... In PAGE 8: ... Other crossover events are said selectively Neutral (noted N). Table2 reports the frequency of such crossover events. We observe, for N=8 and N=16, that D events prevail with SC and that their frequency is dramatically reduced using MHC.... ..."

"... In PAGE 29: ... This is due to the fact that, for lowvalues, the incumbent is likely to fall behind and #5Cgamble for resurrection, quot; while for high values the Type I mistake is more likely. Table2 reports the average number of periods before switching takes place and the average advantage at the time of switching for a subset of the cases of Table 1. The Table shows that the average delay in adoption decreases with d 0 , and that the average advantage at adoption increases.... In PAGE 30: ... Table 1 shows that the percentage of adoptions are uniformly lower for the p = :7 case. Table2 shows that adoption time increases dramatically for the p = :7 case. Less predictably, a comparison between the advantage at the time of adoption shows that this is lower for p = :7 case.... In PAGE 31: ...2 Replacement We now turn to the question of the persistence of market power and analyze who will be the dominant #0Crm at the end of the episode. Again the certainty game is a useful benchmark against 22 These two results are con#0Crmed by the comparison of the #0Crst with the third sub-column in Table2 for the adoption time and the second with the fourth sub-columns for the advantage at adoption.... In PAGE 35: ... Table 4 shows that the average adoption time is greatly increased by the drop, because the incumbent wants to collect more evidence before being willing to accept the drop in performance. In terms of advantage, we note that, taking the drop into account, the values are in general not below the corresponding ones in Table2 . This is because the incumbent switches only under particularly favorable circumstances for the performance advantage, otherwise he would rather #5Cgamble quot; and not adopt.... ..."

"... In PAGE 3: ... decision= A , then accuracy=0; precision=1, recall=0.5, F1-score=0.67). Table1 compares the performance of (1) n-ary with uniform prior [n-uni], (2) n-ary with estimated joint label priors [n-joint], and (3) multi-label with joint priors [multi]. The col- umn [oracle] shows the best possible accuracy that can be achieved among all proposed hypotheses; the column [ER] reports the accuracy for the case where the hypothesis selec- tion is based on the minimum empirical risk on the labeled data.... ..."