Changes from NSP version 0.55 to 0.57 ------------------------------------- Amruta Purandare, pura0010@umn.edu Ted Pedersen, tpederse@umn.edu University of Minnesota, Duluth July 3, 2003 Utils ----- (1) Total rewrite of Utils/kocos.pl to make it more efficient and reliable. kocos allows a user to find Kth order co-occurrences of a given target word or target regexes. The process of identifying "cycles" in the co-occurrence chains has been significantly improved, and problems that existed in handling special characters such as . and , have been resolved. In addition kocos now only displays the Kth order co-occurrences. Previously it displayed the 1st through Kth order co-occurrences. ADP.05 (2) Because of (1) program socs.pl has been removed from NSP. It was provided in .55 as a fast alternative to kocos for 2nd order co-occurrences only. Since kocos is now much faster it is no longer needed and is superseded. (3) Added Utils/combig.pl. This program converts output from count.pl such that bigrams are treated as unordered word pairs. count views a bigram "W1 W2" as W1 preceding W2, while combig.pl views bigram "W1 W2" as W1 and W2 occurring together in either order. (4) Added script/drivers for combig.pl (combig-script.sh). (5) Changed version number of rank.pl from 0.5 to 0.01 to make it consistent with current numbering scheme. rank.pl has not been modified since the very early days of BSP. count.pl -------- (1) display value of --remove in --extended output TDP.57.1 (2) make sure source and destination file exist TDP.57.2 (3) minor rewording of warning/error messages statistic.pl ------------ (1) make sure source and destination file exist TDP.57.3 (2) minor re-wording of warning/error messages Testing ------- (1) Added test cases for new Util combig. (2) Added new tests cases for kocos.pl. (3) Modified test case for count.pl --extended for TDP.57.1 (4) Modified test cases for count.pl and statistic.pl due to rewording of warning and error messages. [Ted - 07/03/2003]