### Table 14 - Helpful Comment Summary

"... In PAGE 9: ...able 13 - Comment Percentages by Subject.......................................................................................66 Table14 - Helpful Comment Summary .... ..."

### Table 1. Conjectures proved and lemmata speculated by reuse Acknowledgements: We thank Stefan Gerberding for helpful comments.

1997

"... In PAGE 3: ...ne the alternatives of our reuse method w.r.t. analysis techniques and heuristics for retrieval and adaptation by performing experiments with Plagiator. Table1 illustrates some of these experiments.1 For each of the given conjec- tures in the second column, the third column points to the statement whose proof is reused and the fourth column shows the lemma speculated by reuse resp.... ..."

Cited by 4

### Table 1: Percentage of out of order packets ( Ben-Aroya amp; Schuster alg. ) 8 Acknowledgements We would like to thank Friedhelm Meyer auf der Heide for his helpful comments on an early draft of this paper.

### Table 3: Number of NN identi ed for Reuters data have very di erent variance). Acknowledgement.The authors would like to thank Yoav Freund for stimulating discussions and help- ful comments.

1997

"... In PAGE 9: ... When the nearest neighbors have higher values a smaller number of samples is needed. In Table3 we list results for several choices of H and k0. For each run, we list the running time, and the fraction of queries for which at least k Nearest Neighbors were correctly identi ed (for k = 1; : : :; 9).... ..."

Cited by 25

### Table 15 - Unhelpful Comment Summary

"... In PAGE 9: ...able 14 - Helpful Comment Summary ...............................................................................................68 Table15 - Unhelpful Comment Summary .... ..."

### Table 7: The numbers of iterations and total time for Sjac, S and Sc The convergence rate of the CG method with Sc is better than that for Sjac, though it is expensive. The CG method with S terminates at the 300-th step without convergence. Acknowledgement The authors gratefully acknowledge Prof. G. Meurant for providing the subrou- tine INV(1) at the very beginning of this work and Dr. M. Pester and Dr. R. Nabben for many helpful comments during the implementation of our method.

### Table 3: Optimal release times (in minutes) and tardiness probabilities mental editor for their helpful comments and suggestions. This work was supported, in part, by grant Grant DMI-9713878 from the National Science Foundation. References Ahuja, R.K., T.L. Magnanti and J.B. Orlin, Network Flows, Prentice-Hall, 1993. Bazaraa, M.S., H.D. Sherali and C.M. Shetty, Nonlinear Programming: Theory and Algorithms, John Wiley amp; Sons, 1993. Danskin, J.M., The Theory of Max-Min and Its Applications to Weapons Allocation Problems, Springer, New York, 1967. 29

"... In PAGE 29: ...8) is veri ed here. Table3 shows the optimal release times together with the respective due dates and tardiness probabilities computed by (7.... ..."

### Table 2: Heap Cell Sharing Statistics. memory since there is little contention for these locks. Although an ideal shared-memory has been simulated, the simulator is exible and can be used to study arbitrarily complex designs. Acknowledgements Enormous thanks are due to our collaborators on the FAST project, John Wild and Hugh Glaser at Southampton University, Pieter Hartel at the University of Amsterdam, and J Liu and Frank Taylor at Imperial. We gratefully acknowledge Erik Hagersten of the Swedish Institute for Computer Science, whose helpful comments led us to the feasibly e cient simulation technique described. This work was supported by an SERC research studentship (for AJB), and partially by SERC project number GR/F/35081. References

1992

Cited by 1

### Tables 1 and 2 quantify the improvement on the previous bound for the Zipf distribution, where pi = 1=iHn. The Zipf distribution often governs the reference process, especially when keys are drawn from a text le (as in Lisp [2]). In Tables 1 and 2 we present the ratio between the stopping points achieved by the Chebyshev and Hoe ding bounds and m as computed in (16), respectively. As shown in Table 2, for any , the bound presented in Theorem 5 improves the value obtained by the Hoe ding bound by at least a factor of . In comparison with Chebyshev (Table 1), which also uses the variance for computing the bound, the improvement increases as becomes smaller. Acknowledgment: We would like to thank Micha Hofri, Kurt Mehlhorn and Ofer Zeitouni for helpful comments and suggestions.

### Table 16 - Aggregated Net Helpfulness Results

"... In PAGE 9: ...able 15 - Unhelpful Comment Summary ...........................................................................................69 Table16... ..."