### Table 8: Fdep for negation words

2007

"... In PAGE 8: ... In order to gain some initial insight into whether dependency parses might be of use here, we calculated the Fdep score for all dependency arcs beginning or ending at any of these words: apos;not apos;, apos;n apos;t apos;, apos;no apos;, apos;none apos;, apos;negative apos;, apos;without apos;, apos;absence apos;, apos;cannot apos;, apos;fail apos;, apos;failure apos;, apos;never apos;, apos;without apos;, apos;unlikely apos;, apos;exclude apos;, apos;disprove apos;, apos;insignificant apos;. The results ( Table8 ) are encouraging and the use of dependency graphs in resolving negations warrants further investiga- tion. The difference between these two parsers is much clearer in this task than in any of the others, and demon- strates that the Charniak-Lease parser may be particularly suited to tackling this problem, as it scores higher than its all-dependencies average while the Bikel parser scores considerably lower.... ..."

### Table 5. Results of Curve Fitting for Number of Failures per Mission. Best Chi-sq Parameter Parameter Std

1996

"... In PAGE 9: ... In order to identify a distribution for the number of failures per mission, the chi-square goodness-of-fit test was applied to three discrete distributions: the binomial, Poisson, and negative binomial. In all cases, the nega- tive binomial was the only acceptable fit with the pa- rameter values as shown in Table5 . However, for the Wing TPS, the large chi-square value indicates the fit was marginal.... ..."

Cited by 6

### Table 5. Results of Curve Fitting for Number of Failures per Mission. Best Chi-sq Parameter Parameter Std

"... In PAGE 9: ... In order to identify a distribution for the number of failures per mission, the chi-square goodness-of-fit test was applied to three discrete distributions: the binomial, Poisson, and negative binomial. In all cases, the nega- tive binomial was the only acceptable fit with the pa- rameter values as shown in Table5 . However, for the Wing TPS, the large chi-square value indicates the fit was marginal.... ..."

### Table 1.1* Summar y of Paths to Belonging and Isolation

### Table 2. Labeling of true/false positive/negative drives. true failure true non-failure

2001

"... In PAGE 5: ... Experimental setup The goal is to obtain a high true positive rate identifying will-fail drives correctly and a low false positive rate of predicting that a good drive will fail. Classification labels are given in Table2 . The true and false positive rates are defined as The true positive rate is the number of true positives divided by the total number of drives identified by the model as anomalies, that is , since there are 9 true failures.... ..."

Cited by 15

### Table 2. Labeling of true/false positive/negative drives. true failure true non-failure

2001

"... In PAGE 5: ... Experimental setup The goal is to obtain a high true positive rate identifying will-fail drives correctly and a low false positive rate of predicting that a good drive will fail. Classification labels are given in Table2 . The true and false positive rates are defined as D8D4D6 BP D8D4BPB4D8D4 B7 CUD2B5 BP D8D4BPBL CUD4D6 BP CUD4BPB4CUD4 B7 D8D2B5 BP CUD4BPBDBLBFBGBM The true positive rate is the number of true positives divided by the total number of drives identified by the model as anomalies, that is D8D4BPB4D8D4 B7 CUD2B5 BP D8D4BPBL, since there are 9 true failures.... ..."

Cited by 15

### Table 2. Labeling of true/false positive/negative drives. true failure true non-failure

2001

"... In PAGE 5: ... Experimental setup The goal is to obtain a high true positive rate identifying will-fail drives correctly and a low false positive rate of predicting that a good drive will fail. Classification labels are given in Table2 . The true and false positive rates are defined as D8D4D6 BP D8D4BPB4D8D4 B7 CUD2B5 BP D8D4BPBL CUD4D6 BP CUD4BPB4CUD4 B7 D8D2B5 BP CUD4BPBDBLBFBGBM The true positive rate is the number of true positives divided by the total number of drives identified by the model as anomalies, that is D8D4BPB4D8D4 B7 CUD2B5 BP D8D4BPBL, since there are 9 true failures.... ..."

Cited by 15

### Table 2. Labeling of true/false positive/negative drives. true failure true non-failure

2001

"... In PAGE 5: ... Experimental setup The goal is to obtain a high true positive rate identifying will-fail drives correctly and a low false positive rate of predicting that a good drive will fail. Classification labels are given in Table2 . The true and false positive rates are defined as D8D4D6 BP D8D4BPB4D8D4 B7 CUD2B5BPD8D4BPBL CUD4D6 BP CUD4BPB4CUD4B7 D8D2B5BPCUD4BPBDBLBFBGBM The true positive rate is the number of true positives divided by the total number of drives identified by the model as anomalies, that is D8D4BPB4D8D4 B7 CUD2B5 BP D8D4BPBL, since there are 9 true failures.... ..."

Cited by 15

### Table 1. Failure data.

"... In PAGE 15: ... Reliability estimation and prediction of dynamic failures. The fail- ure data from Table1 are now input to the estimation and prediction procedure described in Chapter 5. First the failure data are fed into a software reliability growth model to get an estimation of the current reliability as well as a prediction of the future reliability growth.... In PAGE 17: ... More specifically, the results in Table 4 are obtained as follows. First, random numbers are generated in order to put in some more failures among the failure data from the dynamic analysis controlled by the usage model, see Table1 . The failures should be put in to capture failure types that the tool can not find.... In PAGE 17: ...1 failures are put in prior to the first failure in Table 1. This changes the times between failures, hence the time between fail- ures in Table1 are not relevant at this stage. After the failures have been put into the failure history, new times between failures can be calculated.... ..."

### Table 1: Percentage of failures.

"... In PAGE 54: ... This is aimed at testing unimodularity of U(z) without computing det U(z), a numerically di cult task for high values of n and d. Average execution times in seconds are reported in Table1 . A star (?) means that the triangularization failed because the above criteria were not veri ed.... In PAGE 55: ...4 ? 8.0 ? ?? 9 5 29 ? 12 ? ?? 3 9 48 59 17 ? ?? 5 9 220 ? ? ? ?? 7 9 580 ? ? ? ?? 9 9 1200 ? ? ? ?? Table1 : Comparative execution times in seconds. Some comments are in order.... ..."