"... In PAGE 4: ... The first time a noun is encountered it is identified in bold text. In Table1 , the identified nouns are represented in the order they appear, along with the heuristics that have been applied in their grammatical treatment for the purpose of deriving a minimal set of entities. The system should allow learners to search for and make a booking on short courses being run in their area.... ..."

"... In PAGE 12: ... RHS Constituents Unannotated Constituents Unannotated CD 44936 110 44068 1 DT 101135 240 99349 37 IN 121760 35 121213 1 JJ 75257 376 72234 19 NN 163923 772 157887 10 NNP 114041 305 107031 40 NNS 73955 461 69972 1 NP 277909 481 257704 33 PP 64128 212 62528 25 PRP 21357 2 21343 1 RB 38125 152 36958 12 S 58890 463 52052 14 SBAR 25703 42 24847 4 TO 27443 5 27424 1 VB 32545 33 32094 4 VBD 37468 17 37329 8 VBN 24854 32 24537 1 VBZ 26423 33 26254 20 VP 179992 226 168166 8 Table 4: Percentage of unannotated constituents in rule token RHSs The general results obtained with our automatic annotation algorithm are summarised in Table 5. Note again that the Penn-II treebank consists of 49,167 trees, but the results in Table5 exclude trees containing FRAG and X constituents, these being the only constituents which have so far been excluded from our annotation procedure. This leaves us with 48,440 trees.... In PAGE 12: ...ssigned two f-structure fragments, compared to just 0.799% here. Finally, the previously reported gures contained trees which received up to 11 f-structure fragments, whereas the largest fragmentation has currently been reduced to 8 fragments for 6 trees. The results given in Table5 provide us with our rst handle on the fatal annotation errors made by our automatic procedure. We can identify through rigorous linguistic inspection attribute-value clashes, cyclical f-structures, uninstantiated variables and the like, but it is a good deal harder to spot cases of wrong grammatical function assignment, the oblique/adjunct distinction and so on.... ..."

"... In PAGE 12: ... This paper merely sketches the complex linguistic variation detailed there, for the present purpose of examining covariation with social mobility factors.) Table9 shows this linguistic stratification for each variable. Solid lines within a column represent sharp differences of quantitative patterning, while dashed lines represent minor distinctions.... In PAGE 12: ...) Finally, Table 10 brings this all together, performing a qualitative correlation on a quantitative set of rankings. The two central columns give the relative rankings of each speaker on the language scale (derived from Table9 ) and the social status scale (from Table 8). A few speakers who appeared earlier are left out here; data on only one or two variables was analyzed for them, and the purpose of this table is to make the broadest comparison possible.... ..."

"... In PAGE 34: ... Table1 reports the average bandwidth parameter obtained by these methods, and the resulting confidence interval for a t-statistic to test whether the true mean is equal to zero. It can be seen that the Newey-West procedure is better able to detect the higher- order serial correlation, chooses a higher bandwidth parameter, and consequently has better inference properties.... In PAGE 34: ...rocedure. This issue will be discussed further in Section 5. Finally, this simulation experiment highlights the danger of viewing any particular data-dependent bandwidth selection procedure as being fully automatic . As documented in Table1 , the average bandwidth parameter chosen by QS-PW is less than one. When such a low bandwidth parameter is obtained for a sample of 128 observations, it would be useful to check whether the resulting inferences are sensitive to an increase in the bandwidth parameter.... ..."

"... In PAGE 1: ... The intent of this particular project was to begin develop- ing a list of verified heuristics. A comprehensive list of game heuristics (see Table1 ) was developed and com- paratively tested against traditional user study methodolo- gies during the critique of a new game design. The results were examined to evaluate the face validity of the individ- ual heuristics, as well as identify the strengths, weak- Copyright is held by the author/owner(s).... In PAGE 2: ... The displays consisted of screen shots that allowed users to navigate throughout the shell of the game but did now allow any game play. Heuristic Evaluation for Playability (HEP) The HEP heuristics (see Table1 ) were based on the cur- rent literature and reviewed by several playability experts and game designers. The playability evaluator performed the Heuristic Evaluation for Playability (HEP) while fo- cusing on how each heuristic was supported or violated and then defined the playability issue.... In PAGE 2: ... In the Game Play category there were two overlapping issues, which were mainly focused on player goals and motivations. HEP identified four additional issues, which were Game Play Heuristics 4, 6, 7 and 13 (see Table1 for specific heuristics). For example, when using HEP, Game Play Heuristic 13: The first player ac- tion is painfully obvious and should result in immediate positive feedback, identified guidelines regarding the pace of game play that user studies did not directly indicate.... ..."

"... In PAGE 5: ... One can identify three areas of importance in carrying out the abstract sequence, mentioned above. They are expressed as questions in the rst column of Table1 . The questions have, in turn, subordinate aspects, shown in the second column.... In PAGE 17: ...Quality Focus Distribution and causes of retransmissions is Layer Focus Transport measured? Environment Private LAN How Tra c Source generated loads is it Acquisition endpoint active involved measured? Intrusiveness Non-intrusive How is it Acquisition Outside of protocol cycles processed? Analysis Online Table 9: Network Measurement of VMTP { VMTP performance IP Local SIMP both to Satellite echo and through to remote SIMP IP Local Gateway to local SIMP, remote SIMP, and remote gateway IP Local Host to local gateway, local SIMP, remote SIMP, remote gateway TCP Local Host to local SIMP, remote SIMP, remote HOST Table1 0: SATNET Protocols and links disordering ratio was quite close to the predicted value derived for 2 channels. The packet loss rate, between 9 and 106 lost per 10000, was in a agreement with previously measured bit error rates.... In PAGE 18: ...Quality Focus Average and peak throughput of TCP over SATNET Channel is Layer Focus Transport and relation with data link and network measured? Environment Production-use and idle channels How Tra c Source In nite source/sink is it Acquisition Active involvement measured? Intrusiveness (mostly) Non-intrusive How is it Acquisition Run Round Trip test drivers processed? Analysis Online Table1 1: Distributed Testing and Measurement across the Atlantic Packet Satellite Network (SATNET) 3.2.... In PAGE 18: ... This showed that CPU use was the bottleneck. What Quality Focus RPC Transport Latency amp; Throughput is Layer Focus Transport protocol with details down to the hardware level measured? Environment Laboratory LAN How Tra c Source Existing External Network Tra c is it Acquisition Externally observed measured? Intrusiveness Non-intrusive How is it Acquisition Client processed? Analysis online Table1 2: Performance of Fire y RPC... In PAGE 20: ...Quality Focus Throughput amp; Latency is Layer Focus Transport amp; Network measured? Environment Laboratory simulation of small internet How Tra c Source Simulated Network Load is it Acquisition Externally observed measured? Intrusiveness Non-intrusive How is it Acquisition Client processed? Analysis Minimal processing, usually online yielding direct results Table1 3: Congestion Avoidance and Control 4.... In PAGE 21: ... The highest latencies were on the order of 4 times the lowest. What Quality Focus Throughput and latency is Layer Focus Transport measured? Environment Laboratory LANs How Tra c Source Generated loads is it Acquisition Externally observed measured? Intrusiveness Non-intrusive How is it Acquisition Live processed? Analysis Online Table1 4: Measured Performance of Transport Service in LANs 3.2.... In PAGE 22: ...Quality Focus Throughput is Layer Focus Transport measured? Environment 800Mbit/second point-to-point channel How Tra c Source Generated load is it Acquisition Externally observed measured? Intrusiveness Non-intrusive How is it Acquisition Outside protocol time processed? Analysis Online Table1 5: High Speed Networking at Cray Research 3.3 Application Performance This group of papers comprises studies of systems and applications { software with which people directly interact.... In PAGE 24: ...Quality Focus Application Throughput and Latency is Layer Focus Application but extended as necessary down to hardware measured? Environment 4Mbit/second private token ring How Tra c Source Custom load generator is it Acquisition Participate from above measured? Intrusiveness E ectively non-intrusive How is it Acquisition Timestamps obtained outside measured activities processed? Analysis Online, outside of measured activities Table1 6: Analysis of Transaction Management Performance The rst RFC concentrates on performance measurements time protocols. It de- scribes several experiments.... In PAGE 25: ... The other measurement parameters are given in table 17. What Quality Focus Distribution of time service results and performance of NTP is Layer Focus Application measured? Environment Internet How Tra c Source Active queries is it Acquisition Client observation measured? Intrusiveness Non-intrusive How is it Acquisition Capture and transfer to storage processed? Analysis Modest, o ine Table1 7: Network Time Protocol { RFC 1128, 1129 3.3.... In PAGE 26: ...were su ciently close to those of [our] trace-based simulations . . . . quot; What Quality Focus Distribution of responses and response times (latencies) is Layer Focus Transport measured? Environment Internet How Tra c Source Created is it Acquisition Client measured? Intrusiveness Non-intrusive How is it Acquisition Capture and transfer to storage processed? Analysis O ine Table1 8: Accessing Replicated Data in a Large Scale Distributed System 3.3.... In PAGE 27: ...Quality Focus Throughput over link amp; Distribution of paging is Layer Focus Transport amp; Application measured? Environment Laboratory RLAN How Tra c Source Introduced Tra c amp; Workload is it Acquisition Client amp; Externally observed measured? Intrusiveness Non-intrusive amp; Intrusive How is it Acquisition Time online amp; Capture and transfer to storage processed? Analysis Online results amp; O ine Table1 9: Adaptive Remote Paging for Mobile Computers 3.4 Other Studies This group of papers has two subgroups { those related because of some aspect of their work crossing empirical measurement of systems and those which are so broad that they do not neatly fall in any of the preceding three groups.... ..."