[Moose-dev] MSE models repository

Hi folks, At the following address you will find a repository of MSE files derived from the Qualitas Corpus: => http://scg.unibe.ch/pangea/ The purpose of this repository is providing a common and easy to setup dataset for (moose-based) analyses. The files can be freely downloaded and used. The repository has the following structure: - QualitasCorpus - yyyymmdd(e | r): qualitas corpus with MSE files - yyyymmdd(e | r)-compressed: original qualitas corpus (as downloaded from http://qualitascorpus.com/) - extractor: tool used to generate MSE files (verveineJ SVN rev. 192) - mse: generated MSE files - scripts: shell script for setting up the analysis data folder How to setup the analysis data folder ? 1. download all files available at http://scg.unibe.ch/pangea/scripts/ 2. chmod +x 3a. ./fetchQCr.sh for downloading and uncompressing Qualitas Corpus "recent release" + MSE files 3b. ./fetchQCe.sh for downloading and uncompressing Qualitas Corpus "evolution release" + MSE files at the end you should have an exact clone of http://scg.unibe.ch/pangea/QualitasCorpus/20120401r/ and/or http://scg.unibe.ch/pangea/QualitasCorpus/20120401e/ How to run an analysis ? Use a script like this> ----------------------------------------------- cog="path/to/CogVM" moose="path/to/moose.image" smalltalk="path/to/SmallTalkCode.st" for m in $(find . -name 'model.mse'); do src=${m/model.mse/src} ln -s $m model.mse # link to model file ln -s $src src # link to src file $cog -nodisplay $moose $smalltalk rm model.mse rm src done ----------------------------------------------- where SmallTalkCode.st is something like> ----------------------------------------------- | stream model | model := MooseModel new. stream:= MultiByteFileStream newFrom:(FileStream readOnlyFileNamed: 'model.mse'). model name:(FileDirectory baseNameFor: stream localName). model importFromMSEStream: stream. model size > 0 ifTrue: [ model install ]. model rootFolder:'src'. "… analysis ..." WorldState addDeferredUIMessage: [ SmalltalkImage current snapshot: false andQuit: true ]. ----------------------------------------------- Andrea