Summary of changes made to WordNet between versions 1.5 and 1.6 The most significant change between WordNet 1.5 and 1.6 is the introduction of a new window based WordNet Browser application. The new program is written in Tcl/Tk, which runs on Unix, PC and Macintosh platforms. Using Tcl/Tk for the user interface and the WordNet library functions for accessing the database files enables us to provide a common, uniform user interface on all supported platforms. In developing the new browser, we stepped back from the old, familiar one and, with the wondeful insight of our student programmer, David Slomin, redesigned the look and features provided. While most users complimented us on the ease of use of the "old" WordNet, we feel that the new interface is easier yet, and incorporates additional features that users have been asking for. A summary of the changes: New Tcl/Tk based interface that is uniform on all platforms. New options: show synset location in database file (synset offset) show lexical file information show sense numbers for words in synsets user selectable fonts and sizes Functional Changes: new "Overview" of all senses comes up first substring search (grep) is now a "File" option - not tied to word in syntactic category new search "History" new print search results improved save search results keyboard shortcuts improved "Help" With this release of WordNet we have added a package that contains information about mapping WordNet 1.5 noun senses to their corresponding 1.6 senses. This package is intended primarily to assist users who have semantically tagged text to WordNet 1.5 senses, and who wish to upgrade to the 1.6 database. The WordNet Reference Manual is provided in the form of manual pages. In this release, we have included the documentation as nroff source, ASCII formatted files, Postscript files, and HTML files. There have been many smaller changes in all parts of the WordNet system since the last release. The summary information below is mainly for developers. WordNet database: added tagged_cnt field to index file added verb group pointer type index.sense file: added number of tags for each sense removed list of filenames from database "data" files use DBLICENSE_SIZE to get starting offset of first synset in each data file frequency counts (polysemy) now based on # of WN senses, not Collins new gloss index (index.gloss) file example sentences for verb senses Prolog version of the database: added new file wn_vgp.pl -> verb group pointers added field to end of w_s.pl file -> 1 : sense is tagged 0 : sense is tagged changed wn_at.pl to be semantic relation rather than lexical WordNet Library: new WNDBVERSION environment variable that indicates which version of the database to work on. Defaults to "1.6". If set to "1.5", the new Browser and WordNet library can be used with the WordNet 1.5 database files. new WNHOME environment variable. If "old" WNSEARCHDIR is set, it uses that. If not, and WHOME is set, library code looks for database files in WNHOME/dict. Otherwise the default for your computer platform is used. moved SenseCount, OutSenseCount to new "wnresults" data structure removed "datastart" variable (use DBLICENSE_SIZE instead) new OVERVIEW search WNGREP substring search now done in (fast) library code, and not with 'wngrep' system call example sentences for some verb senses new VERBGROUP searches for grouping verb senses by similarity of meaning Synset structure changed: "wnsense" is gone "lexid" is old "wnsense" (lexicographer id) "wnsns" is WordNet sense number Index structure: added "tagged_cnt" field new SnsIndex structure for parsed "sense index" entry