### Table 1 Papers According to Author apos;s (or First Author apos;s) Nationality and ICOTS Series ICOTS Nationality

"... In PAGE 2: ... The ICOTS conference is now 20 years old and 6 meetings have taken place, thus the time has come to analyse if and how a line of thought exists which has developed from 1982 to 2002. To examine carefully the Proceedings of the previous five ICOTS is quite an impossible task as the number of published papers is 783, and ICOTS 6 is in preparation ( Table1 ). A possible solution to the problem posed comes from statistics itself, in particular from quot;textual statistics quot; (Lebart, Salem, Berry, 1998), a field of research helpful to describe, compare and classify sets of texts.... ..."

### Table 2 displays the results for these statistics on both the TREC-8 and CBC data sets. The results for the TREC data are considerably lower than the results for the CBC data. One explanation may be that in the CBC data, only sentences from one document containing the answer are considered. In the TREC data, as in the TREC task, it is not known beforehand which documents contain answers, so irrelevant documents may contain high-scoring sentences that distract from the correct sentences. The max results show that high performance is possible using word overlap as a scoring function: 79% of CBC questions and 65% of TREC-8 questions can be answered correctly. However, these same numbers can be turned around to reveal an inherent limitation of word overlap: 21% of CBC questions and 35% of TREC- 8 questions are impossible to answer correctly, even when making perfect choices. This result illustrates the bene t of using the MaxOset formalism: MaxOsets allow us to identify the answer candidates that are impossible to nd because they will

2001

"... In PAGE 13: ...13 Table2 . Maximum Overlap Analysis of Scores exp.... In PAGE 13: ...1% maximal(ow;q) if 8ov;q 2 q; w 6 v Mq = fow;q 2 q j maximal(ow;q)g Cq = fsjs correctly answers qg We can use these de nitions to give upper and lower bounds on the performance of word weighting functions on our two data sets. Table2 shows the results. The max statistic is the percentage of questions for which at least one member of its MaxOsets is correct.... In PAGE 14: ...8ow 2 Mq; 9s 2 ow s.t. s 2 Cq) Impossible to get it wrong 159 24% (8ow 2 Mq; 8s 2 ow; s 2 Cq) There is no chance to get it right 137 21% (8ow 2 Mq; 8s 2 ow; s 62 Cq) There are no correct answers with any overlap with Q 66 10% (8s 2 d;s is incorrect or s has 0 overlap) There are no correct answers (auto scoring error) 12 2% (8s 2 d; s is incorrect) always be ranked lower than incorrect candidates, no matter what weighting scheme is used. Table2 also shows the min and expected max results. The lower bound is 24% for the CBC data and 10% for the TREC-8 data, which tells us the percentage of ques- tions that are trivially easy to answer using the word overlap scoring function (i.... ..."

Cited by 40

### Table 2: Mission Impossible 2 - Bitsizes for frames and GOPs

"... In PAGE 4: ...3 Analysis results: Mission Impossible 2 Here is the data for the movie Mission Impossible 2. Table2 sumarizes GOP and frame size properties for the movie. Minumum, maximum and average size is given in bits.... ..."

### Table 1. Summary of possibilities (a152 ) and impossibilities (a153 )

1999

"... In PAGE 8: ... 4.3 Summary Table1 summarizes our impossibility and possibility results. Table 1.... ..."

Cited by 1

### Table 2 displays the results for these statistics on both the TREC-8 and CBC data sets. The results for the TREC data are considerably lower than the results for the CBC data. One explanation may be that in the CBC data, only sentences from one document containing the answer are considered. In the TREC data, as in the TREC task, it is not known beforehand which documents contain answers, so irrelevant documents may contain high-scoring sentences that distract from the correct sentences. The max results show that high performance is possible using word overlap as a scoring function: 79% of CBC questions and 65% of TREC-8 questions can be answered correctly. However, these same numbers can be turned around to reveal an inherent limitation of word overlap: 21% of CBC questions and 35% of TREC- 8 questions are impossible to answer correctly, even when making perfect choices.

2001

"... In PAGE 13: ...13 Table2 . Maximum overlap analysis of scores exp.... In PAGE 13: ...1% ow;q = fsjs \ q = wg q = all unique overlap sets for q maximal(ow;q) if 8ov;q 2 q; w 6 v Mq = fow;q 2 q j maximal(ow;q)g Cq = fsjs correctly answers qg We can use these de nitions to give upper and lower bounds on the performance of word weighting functions on our two data sets. Table2 shows the results. The max statistic is the percentage of questions for which at least one member of its MaxOsets is correct.... In PAGE 14: ...8ow 2 Mq; 9s 2 ow s.t. s 2 Cq) Impossible to get it wrong 159 24% (8ow 2 Mq; 8s 2 ow; s 2 Cq) There is no chance to get it right 137 21% (8ow 2 Mq; 8s 2 ow; s 62 Cq) There are no correct answers with any overlap with Q 66 10% (8s 2 d;s is incorrect or s has 0 overlap) There are no correct answers (auto scoring error) 12 2% (8s 2 d; s is incorrect) This result illustrates the bene t of using the MaxOset formalism: MaxOsets allow us to identify the answer candidates that are impossible to nd because they will always be ranked lower than incorrect candidates, no matter what weighting scheme is used. Table2 also shows the min and expected max results. The lower bound is 24% for the CBC data and 10% for the TREC-8 data, which tells us the percentage of ques- tions that are trivially easy to answer using the word overlap scoring function (i.... ..."

Cited by 40

### Table 1. Complexity of 6-Round Impossible Differential Attack

"... In PAGE 10: ...ryptions, or 293.5 plaintexts, 2110.5 encryptions and 2104.5 bytes of memory. We summarize the complexities of our attacks together with those of previous works in Table1 . We expect that this method can be applied to other block ciphers... ..."

### Table 1: Parallelism between square and impossible differential characteris- tics

### Table 1. Impossible state transitions. 1 a93

### Table4. NetworkstoMap

"... In PAGE 9: ...llegal organizations. This data is occasionally difficult to unearth with cooperating clients. With covert criminals, the task is enormous, and may be impossible to complete. Table4 below lists multiple project networks and possible data sources about covert collaborators. Table4.... ..."