"... In PAGE 14: ...Table 2: The bipartite-regular hypermaps on the sphere. Based on the knowledge of regular hypermaps on the sphere, we display in Table2 all the possible values (up to duality) for the bipartite-type of the bipartite-regular hypermaps on the sphere and the unique hypermap (up to isomorphism) with such a bipartite-type. Notice that the map of bipartite-type (1; n; 2; 2n) can be constructed from Dn either via a Wal transformation W al(D(02)(Dn)) or via a Pin transformation P in(D(12)(Dn)).... In PAGE 17: ...nd n give rise to a spherical type (cf. Table 1). If not we choose the second greatest, the third greatest and so forth. For each bipartite-regular hypermap K in Table2 , where (l1; l2; m; n) is our (r; s; u; v), taking the greatest values for the triple (l; m; n) we get a spherical type. To check if such triple determines a hypermap covered by K we take a half-turn in the middle of each hyperedge of K; these half-turns determine a covering K 7! K .... ..."

"... In PAGE 37: ...5.13% 88.00% 81.00% 85.00% Table2 - Companies Online Sophistication: Serbia 2007 and EU 2006 7.4 Cluster Online sophistication 7.... In PAGE 83: ... The proposed method was used on a set of OO projects in order to explore the acknowledged effort on real projects. Table2 summarizes the metrics values for three groups of projects. In the first group are smaller student projects written in Java.... In PAGE 83: ... MI is calculated at the project level, thus only one value is in the table. Table2 : The Values for Selected Code Quality Metrics DIT CBO LCOM MI GROUP 1 Project 1.1 3,35 2,47 11,82 11,23 95,56 13,33 65,19 Project 1.... In PAGE 113: ... We report average performance (precision, recall) over the three models. We first ran 3-fold cross validation on manually labeled dataset, using only the first paragraph from the article text ( Table2 a) and using the plain text from the entire article (Table 2b). Precision Recall F1 BEP People 92.... In PAGE 113: ... We report average performance (precision, recall) over the three models. We first ran 3-fold cross validation on manually labeled dataset, using only the first paragraph from the article text (Table 2a) and using the plain text from the entire article ( Table2 b). Precision Recall F1 BEP People 92.... In PAGE 113: ...5.96% 49.17% 33.91% 28.05% Table2 a: Results from cross validation on sample set using only first paragraphs of article text. Precision Recall F1 BEP People 85.... In PAGE 113: ...7.63% 37.73% 41.59% 43.11% Table2 b: Results from cross validation on sample set using plain text from entire article. *Since the sample set was too unbalanced in this category, the cross validation was run with bias misclassification cost (SVM parameter j=5).... In PAGE 113: ... The text extracted from the first paragraph of the articles was used in all subsequent classification experiments presented in this paper since it is significantly computationally less expensive. However, the results obtained using the entire text of the articles ( Table2 b) are better than the results of using the first paragraph only (Table 2a). We also tried to use the text ... In PAGE 113: ... The text extracted from the first paragraph of the articles was used in all subsequent classification experiments presented in this paper since it is significantly computationally less expensive. However, the results obtained using the entire text of the articles (Table 2b) are better than the results of using the first paragraph only ( Table2 a). We also tried to use the text ... In PAGE 114: ... Bhole et al. are a few percent better than those in Table2 b, but this needs further analysis. The diversity of the sample set can be explained to be a cause of the low recall in case of people and places, where the SVM may have misclassified some article alien to it.... In PAGE 123: ... As seen from Table 1, each module has its own strong and weak points. The integrated system without the visual learning classified as OK 88% of all the tested regular entries and as alarm 69% of all the irregular entries as presented in Table2 and Table 3. In practical experiments of several additional scenarios, the expert-defined rules and the video learning module proved quite successful on their own, and the overall performance improved as well.... ..."

"... In PAGE 4: ... The confidence score represents the probability that two proteins interact with each other [12]. The size and global topological measures for genetic, con- gruence, and protein networks are provided ( Table1 ). The average degree is the number of edges per node, and the clustering coefficient measures the interconnectivity around a node.... In PAGE 7: ... We have found good agree- ment between motif enrichments (Fig. 3A) and average clustering coefficients ( Table1 ), i.e.... ..."

"... In PAGE 39: ... Using an automated expert system, we were able to benefit from the additional information with little effort. Table12 gives the top 10 double, triple and quadruple feature combinations, which provided the lowest miss rates. The feature numbers listed in the table correspond to the feature numbers in Table 11.... In PAGE 40: ... This is particularly true when the 13th feature (NotTaken Successor Ends, feature 25) is added in the Best First Fortran line. This result, in combination with the data in Table12 , strongly suggests that for Fortran, at least, a small feature set containing the most important features, results in the best miss rates for the technique. Any additional features appear to have little effect and sometimes reduce the effectiveness of the technique.... ..."

"... In PAGE 27: ... Using an automated expert system, we were able to benefit from the additional information with little effort. Table12 gives the top 10 double, triple and quadruple feature combinations, which provided the lowest miss rates. The feature numbers listed in the table correspond to the feature numbers in Table 11.... In PAGE 28: ... This is particularly true when the 13th feature (NotTaken Successor Ends, feature 25) is added in the Best First Fortran line. This result, in combination with the data in Table12 , strongly suggests that for Fortran, at least, a small feature set containing the most important features, results in the best miss rates for the technique. Any additional features appear to have little effect and sometimes reduce the effectiveness of the technique.... ..."