Lydia is a project mainly written in Python, it's free.
Arabic NER pipeline for Lydia system
Modify your bashrc
PYTHONPATH=$PYTHONPATH:/path/to/your/lydia/ export PYTHONPATH
you can change ~/sandbox/ to wherever you are keeping the code
To get a dump of arabic wikipedia http://download.wikimedia.org/arwiki/
the one that I downloaded |wget -c http://download.wikimedia.org/arwiki/20110121/arwiki-20110121-pages-articles.xml.bz2
To fragment the wikipedia dump to different xml files, each contain one article |$wikipedia/fragmenter.py wikiDump.xml outputDirectory The outputDirectory will contain all the files. The files are named after the ids of the wikipedia articles.
In each xml file, the text is contained between the
To extract the text and remove the markup language. $wikipedia/formatter.py inputFile This will generate an output file called inputFile.f
$aner/NER.py folderName It will run over all the files that have .f extension, it will delete all the temp files except .bw.post.NER file which has five columns. The stats will will be stored in folderName/stats
$aner/NER filename It will run on the file passed and make .bw.ner file, it will not delete any temp files.