"... In PAGE 7: ... This would allow to safely exploit the incremental approach in domains characterized by a continuous flow of new documents. Table1 reports the statistics regarding the performance of the two exploited approaches, averaged on the 10 folds, of the classification process in this environment as regards number of clauses that define the concept Clauses, Accuracy on the test set (expressed in percentage) and Runtime (in seconds). The difference in computational time between the two systems is noteworthy, confirming that the incremental approach should be more efficient than the batch one (since it has to just revise theories stepwise, instead of learning them from scratch).... ..."

"... In PAGE 10: ... Therefore, we would like to use a combination of these measures, rather than a single measure, to measure the effectiveness of our document classification system. The measures of recall, precision, and F-measure can be explained with respect to a contin- gency table ( Table1 ), which characterizes typical outputs from a document classification system. Joanna Yi-Hang Pong et al.... In PAGE 12: ...3.1 Micro-averaged F-measure As shown in Table1 , the KNN classifier consistently out-performs the NB classifier in terms of recall, precision, and micro-averaged F-measure in the library setting. The F-measure of the KNN classifier is 0.... ..."

"... In PAGE 7: ... Table1 shows the features used to represent documents using text chunks. Top five features represent how often the phrases are used in a document, the following five features imply how long they are, and the final feature means how long a sentence is on the average.... ..."

"... In PAGE 4: ...Table 1. Results on 100 documents: stroke and document classification rate Table1 shows the classification rates of the proposed system and the reference system on the stroke level and on the document level. The strokes classification rate is the number of correctly assigned strokes divided by the to- tal number of strokes.... ..."

"... In PAGE 7: ..., 1999). Table1 shows the features used to represent documents using text chunks. Top five features represent how often the phrases are used in a document, the following five features implies how long they are, and the final feature means how long a sentence is on the average.... ..."

"... In PAGE 3: ...tation (180 Mhz). It took about 2.5 seconds to classify each document (after page segmentation). The classification ac- curacy for each of the five classes of documents is shown in Table1 . a145a73a134 a134 a146 a106 a147 a134 a148 a109 refers to the percentage of documents whose correct class appeared as the top-choice (top-one ac- curacy).... In PAGE 4: ... whose correct class appeared in either the top or the second choice (top-two accuracy). As seen from Table1 , the top-two accuracy scores are very high for all but one class, demonstrating the effective- ness of the method as an initial screening stage. The ac- curacies for the two-column letter class are relatively low.... ..."

"... In PAGE 3: ...tation (180 Mhz). It took about 2.5 seconds to classify each document (after page segmentation). The classification ac- curacy for each of the five classes of documents is shown in Table1 . Accuracy1 refers to the percentage of documents whose correct class appeared as the top-choice (top-one ac-... In PAGE 4: ... whose correct class appeared in either the top or the second choice (top-two accuracy). As seen from Table1 , the top-two accuracy scores are very high for all but one class, demonstrating the effective- ness of the method as an initial screening stage. The ac- curacies for the two-column letter class are relatively low.... ..."