"... In PAGE 61: ...On questionnaire students estimated the usefulness of contact lessons, videoconferencing or chat in learning environment as an optional method for a Web-Based programming course. As we can see in Table4 even 53 % of students wished for contact-lessons (face-to-face) besides of Web-Based lessons. When we compare the information in Table 3 and Table 4... In PAGE 62: ... Only exception is arrays: even 40% of students believed that chat in learning environment might help to learn arrays. Quite expectedly the needs for support mainly correlates the feelings of di culties (com- paring Table 2 and Table4 ). But even students found loops not so bad to learn, still 40% of them want contact-lessons to support the learning of loops.... In PAGE 62: ... We found it important that not only the best students but also those students who own problems with learning process also realize the bene t of support tools like chat or discussing forum in Web-Based learning environment and start to use them as a natural way of learning. As we can see in Table4 students can learn very much from each other and in distance education contacts with other students via Web can be as fruitful as discussing in face-to-face. 5 Conclusions As we can clearly see high school students had di culties with studying independent via Internet.... In PAGE 74: ... The both parts are summarized in Table 4. Table4 : Results for internal students (group 1). Activity Students Registered for the course 89 Passed the project 73 Passed the course with the rst exam 64 Passed the course within one school year 73 Study average for the whole study group and school year 4.... ..."

"... In PAGE 4: ... Due to the soft integration of the tagger, though, the parser is not forced to accept its predictions unchallenged, but can override them if the wider syntactic context suggests this. In our experiments (line 1 in Table1 ) this happens 75 times; 52 of these cases were actual errors com- mitted by the tagger. These advantages taken to- gether made the tagger the by far most valuable in- formation source, whithout which the analysis of arbitrary input would not be feasible at all.... In PAGE 5: ... For compar- ison with previous work, we used the next-to-last 1,000 sentences of the NEGRA corpus as our test set. Table1 shows the accuracy obtained.1 The gold standard used for evaluation was de- rived from the annotations of the NEGRA tree- bank (version 2.... In PAGE 5: ...ion guidelines (e.g. with regard to elliptical and co-ordinated structures, adverbs and subordinated main clauses.) To illustrate the consequences of these corrections we report in Table1 both kinds of results: those obtained on our WCDG-conform annotations (reannotated) and the others on the raw output of the automatic conversion (trans- 1Note that the POS model employed by TnT was trained on the entire NEGRA corpus, so that there is an overlap be- tween the training set of TnT and the test set of the parser. However, control experiments showed that a POS model trained on the NEGRA and TIGER treebanks minus the test set results in the same parsing accuracy, and in fact slightly better POS accuracy.... ..."

"... In PAGE 4: ...88, indicating near- perfect agreement [22]. The final set of categories and (summarized) definitions are shown in Table1 . Based on these categories, we coded all messages in the entire corpus of messages from the groups (4242 messages), which forms the basis of all analyses that follow.... ..."

"... In PAGE 4: ...88, indicating near- perfect agreement [22]. The final set of categories and (summarized) definitions are shown in Table1 . Based on these categories, we coded all messages in the entire corpus of messages from the groups (4242 messages), which forms the basis of all analyses that follow.... ..."

"... In PAGE 5: ... The multiple voices solution is investigated further in section five of the discussion. The remaining options attempt to mimic the ranking system in Table7 . One option is to markup messages individually with ranked live settings.... In PAGE 5: ... A MIN ranked message could be given a polite live setting. Below is an example of what the chat in Table7 would look like using the described markup. Figure 2: Live Region Ranking Markup Several design decisions were made when using this markup.... In PAGE 7: ... Reef Chat can be used to do basic ranking or filtering of messages based on importance. The markup described in Table7 could be read by a screen reader and used to give a volume to a message, effectively using different volume levels as an imperfect, but perhaps passable, substitute for different voices for the ear to latch onto. A message marked with a MAX flag, would be given a 100% volume setting, a MID flagged message would be given a volume between 50-80%, and a MIN lagged message a volume of 50%.... In PAGE 7: ... Future work is needed to study how many audio conversations the average user could follow without being overloaded with information. Also, the ranking algorithm used in Table7 is fairly simple and more complex algorithms would be required to best support information processing. Several other technical barriers also remain as well, especially if this model is to be extended to support the wide diversity of Ajax applications that exist beyond chat.... ..."

"... In PAGE 4: ... This affected fewer than twenty labels on the development data, and added only about 0:1 to the overall f-measure. 4 Results The results on the test section of the CoNLL 2004 data are presented in Table1 below. The overall result, an f- score of 60:66, is considerably below results reported for systems using a parser on a comparable data set.... ..."

"... In PAGE 7: ... Our system performed better than the commercial systems, but this has to be interpreted with caution, since our system was trained and tested on sentences from the same lexically limited corpus #28but of course without overlap#29, whereas the other systems were developed on and for texts from a larger varietyof domains, making lexical choices more di#0Ecult in par- ticular. Table5 shows the correlation between various parse and translation metrics. Labeled precision has the strongest correlation with both the syntactic and semantic translation evaluation grades.... ..."

"... In PAGE 7: ... Our system performed better than the commercial systems, but this has to be interpreted with caution, since our system was trained and tested on sentences from the same lexically limited corpus #28but of course without overlap#29, whereas the other systems were developed on and for texts from a larger varietyof domains, making lexical choices more di#0Ecult in par- ticular. Table5 shows the correlation between various parse and translation metrics. Labeled precision has the strongest correlation with both the syntactic and semantic translation evaluation grades.... ..."