### Table 1: Intensity importance of factors in the pairwise comparison process Intensity of

"... In PAGE 9: ... This comparison technique is useful in this paper to find the weight factor of each indexing terms of the domain. The expert assigns an importance intensity number from Table1 that represents, in a 1-9 scale, the relative importance of each term ti g206 Kd with respect to other term tj g206 Kd to describe the domain. If n terms have been selected by the expert to describe the domain, an n g180 n matrix A is derived.... In PAGE 11: ... The AHP allows us to compute a vector vi = {vij} that ranks the terms tj g206 Ti according to their capacity for representing the same concept as ti. The first stage to derive this vector consists of defining an m g180 m pairwise matrix Pi, in which the element pijk is the importance intensity number, in a 1-9 scale (the same as Table1 ), which represents the relative importance of each term tj with respect to other term tk to describe the same concept as ti. Then, the eigenvector vi, associated to the matrix Pi, is computed following the strategy above described to find the weight factor of keywords of the domain.... ..."

### Table 1: Facts About the Alvesta and the Orlando Field Exercises

2004

"... In PAGE 12: ... This exercise also included an AAR less than 3 hours after the conclusion of the exercise in which the participants were able to reflect on their actions in relation to the unfolding of the scenario and the factual recorded mission history. Table1 shows a comparison between the two exercises in terms of a number of important factors. Both exercises provided a unique opportunity for the responding task forces to train together for a plausible emergency.... ..."

### Table 9: Pairwise Comparisons for Qualifications 95% Confidence Interval for

"... In PAGE 9: ...009 50.170 Table9 helps to clarify the nature of these effects by showing the pairwise comparison for the main effect of qualification corrected using a Bonferroni adjustment. This table indicates that the significant main effect reflects a significant difference (p lt; 0.... ..."

### Table 4 about here

1999

"... In PAGE 22: ... It is also possible to see, using the closed-form expressions for these quantities, that the degree of skewness and kurtosis the stochastic volatility model can generate for common parametrizations is less than that of jump-di#0Busions. Table4 summarizes the degree of conditional skewness and excess kurtosis for a range of possible values of the relevant parameters: #1A 2f,0:25; 0g, #14 2f1; 5g, a 2f0:75; 1; 1:25g, and #11 2f0:1; 0:4g. The table reveals two important points.... In PAGE 31: ... In the remainder of this section, we discuss the numbers in these tables, with a particular focus on relating the behavior of the implied volatility smiles to the behavior of excess kurtosis in this model. First, wehave seen earlier #28 Table4 #29 that the stochastic volatility model has only a limited ability to generate excess kurtosis unless the value of #11 is very high. Re#0Decting this, it can be seen from Tables 8#7B13 that at the realistic value of #11 =0:1, the depth of the volatility smile #28the highest minus the lowest implied volatilityinanyrow#29 is insubstantial, rarely exceeding 1#25 in Tables 8#7B10, and never exceeding 2#25 in Tables 11#7B13.... In PAGE 31: ...onths in some cases #28e.g., when #14 = 1 and #11 =0:4inTable 8#29. Third, the introduction of skewness into the model by considering #1A = ,0:25 results in a signi#0Ccant asymmetry in the volatility smile, and also increases the depth of the smile. #28The latter e#0Bect arises from the fact that kurtosis also increases in this case as #1A changes; see Table4 .#29 This is immediate from a comparison of the otherwise identical Tables 8 and 11, Tables 9 and 12, or Tables 10 and 13.... In PAGE 31: ...#29 This is immediate from a comparison of the otherwise identical Tables 8 and 11, Tables 9 and 12, or Tables 10 and 13. Fourth, Table4 also showed that the degree of excess kurtosis in the stochastic volatility model decreases as the ratio a = v 0 =#12 of initial volatility to its long-term mean increases. As a consequence, the smiles in Tables 8#7B10 also exhibit a montonic pattern, decreasing in... In PAGE 51: ...Table4 : Conditional Skewness and Kurtosis in the Stochastic Volatility Model This table presents the values of conditional skewness and excess kurtosis at various horizons in the stochastic volatility model for a range of values of the four relevant parameters: the correlation #1A between the returns and volatility processes, the coe#0Ecient #14 of mean-reversion in the volatility process, the ratio a of currentvolatility to its long- term mean, and the volatilityofvolatility #11. Parameters Skewness Excess Kurtosis #1A #14 a #11 1week 1 month 3 months 1week 1 month 3 months 0 1 0.... ..."

Cited by 25

### Table 2: Pairwise comparison of the relative performance of the protocols as measured by the total

"... In PAGE 18: ... One protocol signi cantly outperformed the other when it piggybacked on average fewer bits and the 95% con dence intervals did not overlap. Table2 shows the pairwise comparison of the six protocols. The values in the table represent the number of points in the BBL space where the protocol in the column outperformed the protocol in the row.... In PAGE 18: ... Because they are not competitive, we exclude protocols + jLogj and + Log from the rest of the discussion and concentrate on the performance of the four remaining protocols. Protocol Det Table2 shows that over all the points sampled, no protocol ever piggybacks signi cantly fewer bits than Det . This suggests that if no knowledge about the application apos;s characteristic is known then Det is a good choice.... In PAGE 20: ...2% fewer determinants than Det , not enough to reduce the extra cost associated with this protocol. One might, in fact, argue from Table2 that Det is slightly better than jLogj , since there are 59 points where Det signi cantly outperforms Log and only 56 points where jLogj signi cantly outperforms Log . Similarly, Det does better more often in comparison to + Det than jLogj .... ..."

Cited by 3

### Table 6 Number of articles in each journal in each year (for journals with more than seven articles)

"... In PAGE 13: ...erely 18. And this may be an area where researchers want to investigate further. Table 5 Number of published articles for each theme Themes Number of articles Implementation General Case study Critical success factors Change management Focused stage in the implementation process Cultural (national) issues 135 61 17 15 11 16 17 Using ERP General Decision support Focused function in ERP Maintenance 44 21 4 11 8 Extension 37 Value 24 Trends and perspectives General In a particular sector 55 48 7 Education 18 Note: The total number of journal papers over sub-themes may be greater than the number for a corresponding theme due to the fact that certain articles may be designated for more than one sub-theme. Top 13 journals in terms of the number of articles in ERP are listed in Table6 . These 13 journals have published a total of 179 articles or 57% of the total.... ..."

### Table 4: Dataset Compression Rates for SCE and Wilson Editing .

2004

"... In PAGE 5: ...4.5 that was run using its default parameter settings. Table 3 reports the accuracies obtained by the four classifiers evaluated in our experiments. Table4 reports the average dataset compression rates for supervised clustering editing and Wilson editing. Due to the fact that the supervised clustering algorithm has to be run 10 times, once for each fold, different numbers of representatives are usually obtained for each fold.... In PAGE 5: ...1 the (rounded) average number of representatives was 27, the maximum number of representatives during the 10 runs was 33 and the minimum number of representatives was 22; supervised clustering editing reduced the size of the original dataset O by an average of 96.5%, as displayed in Table4 . The NR classifier classified 73.... In PAGE 5: ...able 4. The NR classifier classified 73.6% of the testing examples correctly, as indicated in Table 3. Table4 only reports average compression rates for Wilson editing. Minimum and maximum compression rates observed in different folds are not reported, because the deviations among these numbers were quite small.... In PAGE 6: ...Table 4: Dataset Compression Rates for SCE and Wilson Editing . More importantly, looking at Table4 , we notice that with the exception of the Glass and the Segmentation datasets, SCE accomplishes compression rates of more than 95% without a significant loss in prediction accuracy for the other 6 datasets. For example, for the Waveform dataset, a 1-NN classifier that only uses 28 representatives outperforms the traditional 1-NN classifier that uses all 4500 training examples4 by 7.... In PAGE 7: ... Prior to conducting the experiments we expected that the NR classifier would perform better for lower compression rates. However, as can be seen in Table4 , this is not the case: for six of the eight datasets, the highest accuracies were obtained using b=0.1 or b=0.... ..."

Cited by 4

### Table 2. Comparison for special cases

2004

"... In PAGE 12: ... Table2 illustrates the importance of checking for special cases. These include in- variants, syntactically safe properties, bounded LTL properties, and liveness properties of the form F p, where p is a propositional formula.... In PAGE 12: ...ng properties. The column labeled k has the same meaning as in Table 1. If the value of k is 0, the corresponding property is an inductive invariant. In Table2 , the general method is slower. There are two reasons for that: The rst is that using the termination criteria of Theorem 1 and (3a00) generate more clauses for a given value of k.... ..."

Cited by 10

### Table 2.1: Tools apos; general description Feature Apollo LinkFactory OILEd OntoEdit

"... In PAGE 89: ...2 hosts a comparison of RDF/S query languages according to five axes: Modeling constructs supported, Ontology Querying, Data Querying, Data/Ontology Querying and Additional Features provided. Regarding the first comparison axis (Modeling constructs), we can observe from Table2 that all query languages taking part in the comparison, i.e.... In PAGE 89: ... Furthermore, the ability of query languages to perform Data querying is of major importance. Based on this set of criteria, we can note from Table2 that all query language provide constructs for finding the extent of a class or property, either directly or transitively. What most query languages do not support is set-based operations (union, intersection, difference) as well as arithmetic operations on data values.... In PAGE 89: ...ime. The basic criterion for judging this ability is the use of generalized path expressions. Generalized path expressions are very useful primitives because they allow data and ontology to be uniformly queried. As indicated from Table2 , only RQL is capable of incorporating knowledge from ontologies into data querying. In fact, RQL features generalized path expressions with variables on labels of both nodes and edges.... In PAGE 89: ... To evaluate the effectiveness of the query languages when used in large-scale Semantic Web applications we can also use a set of criteria referring to Additional Features supported by the query languages. Table2 records whether the query languages under evaluation support aggregate, grouping and sorting functions. More specifically, RDFQL supports only a count function, VERSA supports min and max, while RQL features min, max, count, average and sum functions, as known from relational databases.... ..."

### Table 9 (continued) Has your firm seriously considered issuing common stock? (if no , please go to the next question) If quot;yes quot;, what factors affect your firm apos;s decisions about issuing common stock?

2001

"... In PAGE 22: ... Among these firms, earnings dilution is the most important concern affecting their decision to issue equity (mean response of 2.84 in Table9 ).15 The popularity of this response is intriguing.... ..."

Cited by 34