### Table 4 shows recall for the three approaches to sense weighting. Surprisingly, assigning lower weights to senses for highly ambiguous words ( Fractional ) does not improve per- formance over the standard approach ( Synsets ). Taking the number of different source words in the density measure ( Words ) produces an improvement, but nearly negligible.

2001

"... In PAGE 9: ... Table4 : Effects of sense weighting 3.7 Behavior on different text categories The Semcor documents, a fraction of the Brown Corpus [FK82], are classified accord- ing to a set of predefined domains (Press, General Fiction, Romance, Humor, etc.... ..."

Cited by 3

### Table II shows the gap fractions obtained for the four event samples. The values range from (0:22 0:05)% for central jets at ps = 1800 GeV, to (1:19 0:08)% for forward jets at ps = 630 GeV. Uncertainties are dominated by those on the t parameters. Additional small uncertainties from the dependence on the range of multiplicity used in the ts were added in quadrature. Potential sources of systematic error, such as the number of t parameters, jet energy scale, trigger turn-on, tower threshold, luminosity, residual noise, and jet quality, yield only negligible variations in the gap fractions [11]. Table II shows that the gap fractions at ps = 630 GeV are larger than gap fractions at ps = 1800 GeV and that gap fractions for forward jets are larger than for central jets. Table II also lists predicted gap fractions for several possible pomeron structure functions (discussed below).

54

### Table 1 shows the sizes of the graphs, and the results obtained for the treewidth lower bounds without contraction. These bounds are the exact parameters apart from the values for the three RD-heuristics. As the computation times for , 2 and R are negligible, we omit them in the table. Also the D can be computed within a fraction of a second. The computational complexity of 2D is O(n) larger than the one of D which is re ected in the CPU times for this parameter.

"... In PAGE 15: ... Table1... In PAGE 16: ...rie M. C. A. Koster, Thomas Wolle, and Hans L. Bodlaender Table 2 shows the results for the same graphs as in Table1 . Furthermore, in Table 2, we give the treewidth lower bounds according to the parameters that involve contraction.... In PAGE 16: ...4 Discussion. The results of algorithms and heuristics that do not involve edge-contractions ( Table1 ) show that the degeneracy lower bounds (i.e.... In PAGE 16: ...ethods involving computing lower bounds many times (e.g. branch amp; bound). Even though the 2D algorithm has much higher running times than the other algorithms in Table1 , it is still much faster than some heuristics with contraction. Furthermore, we expect that its running time could be improved by a more e cient implementation.... In PAGE 16: ... No further investigations about parameters without contraction have been carried out as the parameters with contraction are of considerably more interest. We can see that when using edge-contractions, the treewidth lower bounds can be signi cantly improved (compare Table 2 with Table1 ). The results show that values for 2C are typically equal or only marginal better than the value for C.... ..."

### Table 2 shows that for version A, most of the I/O time is spent opening and reading files. Because the default M UNIX mode is used, all reads during phase one are seri- alized. This inefficiency makes reads a high fraction of the total I/O time. However, because all writes occur through node zero, there is no contention, and the time spent on writes and seeks is negligible.

"... In PAGE 4: ... Table 1 summarizes the major differ- ences among the three versions: the number of nodes that actively participate in I/O duringeach of the four application phases and the PFS I/O mode used in each phase. Similarly, Table2 shows the fraction of time attributable to each I/O operation type. The I/O activity for version A reflects the influence of the Concurrent File System (CFS) on the Intel Touchstone Delta, where the code was first developed.... In PAGE 4: ... Table2 : Aggregate I/O performance summaries (ESCAT) In version B, a significant reduction in read time was achieved via code restructuring. In this version, node zero reads the input data and broadcasts it to the remainder of the nodes; see Table 1.... ..."

### Table 4.2 shows the sizes of the graphs, treewidth upper bounds and the results obtained from the treewidth lower bound algorithms without edge-contraction. These bounds are the exact parameters apart from the values for the three RD heuristics. As the computation times for , 2 and R are negligible, we omit them in the table. Also the D can be computed within a fraction of a second. The computational complexity of 2D is a factor of O(n) larger than the one of D which is re ected in the CPU times for this parameter.

### Table 1: Number of decays, e ciencies and background fractions for the decay channels ! e and ! . The variables are explained in Section 2.

1995

"... In PAGE 6: ... This led to a 4:1% loss of tracks around the boundary regions between the six azimuthal TPC sectors, which was well described by simulation. The details of the selected decay sample are listed in Table1 . There were 31325 decays selected, with a background fraction, b , of 0:013 0:004.... In PAGE 8: ... There were 5059 ! e decays identi ed. The identi cation e ciency and back- grounds are summarised in Table1 . The contribution to the background from ! decays was negligible.... In PAGE 9: ... There were 6586 identi ed ! decays. The identi cation e ciency and back- grounds are summarised in Table1 . The contribution to the background from ! e decays was negligible.... In PAGE 9: ... The contribution to the background from ! e decays was negligible. 7 Results and Conclusions From the numbers listed in Table1 the leptonic branching fractions were found to be B( ! e ) = (17:51 0:23stat 0:31sys)%; (4) B( ! ) = (17:02 0:19stat 0:24sys)%: (5) The contributions to the systematic errors are listed in Table 2, and are propagated from the errors on the quantities as given in Table 1. Due to the precise measurement of the polarisation in Z0 decays from DELPHI [5], the uncertainty from this source was estimated to be negligible.... ..."

### Table 8: Compression, fidelity and loss in reconstruction of grey scale images.

"... In PAGE 10: ...percentage of the affected pixels (next to last column of Table8 , and entirely negligible (a fraction of a percent, see last column in Table 8) when averaged over all pixels. This is demonstrated in the reconstructed figures, that show little perceptible change.... In PAGE 10: ...percentage of the affected pixels (next to last column of Table 8, and entirely negligible (a fraction of a percent, see last column in Table8 ) when averaged over all pixels. This is demonstrated in the reconstructed figures, that show little perceptible change.... ..."

### Table 1: Basis modes for the 2002 t to branching fraction data.

"... In PAGE 2: ... The basis modes form an exclusive set whose branching fractions are constrained to sum exactly to one. The list of 31 basis modes selected for the 2002 t are listed in Table1 . There are two changes from the 2000 basis set: 1) the mode ? ! ?K0K0 has been split into the two modes ? ! ?K0 SK0 S and ? ! ?K0 SK0 L , with the assumption that B( ? ! ?K0 SK0 S ) = B( ? ! ?K0 LK0 L ); and 2) the mode ? ! h?h?h+ 3 0 has been replaced by ? ! h?h?h+3 0 , with the assumption that B( ? ! h?h?h+ 4 0 ) is negligible.... ..."

### Table 2 Fractions of trees which are cospectral with respect to the matrices A, L and L

"... In PAGE 9: ... We therefore confine our attention to A, L and L. The results are summarised in Figure 2 and Table2 . The fractions here refer to the number of trees which do not have a unique spectrum.... In PAGE 10: ... These results are very encouraging since they suggest that for both trees and graphs, the cospectrality problem is negligible even for moderately sized objects. The final column in Table2 shows the number of pairs of trees which are cospectral in A and L at the same time. Interestingly, cospectral pairs for A and L seem to be unrelated to each other, and so combining the two spectra leads to very few cospectral graphs.... ..."