Persian Link Grammar -------------------- By Jon Dehdari Brigham Young University Linguistics Department 2003-2005 This README is for 'persianlg', a Persian (Farsi) implementation of the Link Grammar (LG) syntactic parser. The LG Parser is a new approach to parsing sentences, and was developed at Carnegie Melon University. The parser works by simply linking words together. Directionality and distance are the primary constraints given, and are found in the 4.0.dict file. The Persian PersianLG implementation is the first full non-English implementation, to my knowledge. It works a little differently than the English version, due to the more complex word structure. Morphological pre-parsing is required, either by hand, using the included stemmer (stemmer.pl) [1], or using the Persian-pckimmo morphology engine. The LG-Soar framework, developed in the BYU Soar research group, has tied both the LG parser and the PC-Kimmo engine together, to provide seamless integration of morphology and syntax. Thus manual morphology pre-parsing is not necessary, nor is cutting and pasting pckimmo output to the link-parser input. For example, an unvoweled text like 'mn midAnm kh tu rfti.' (I know that you went), should go into the LG parser as 'mn mi dAn m kh tu rf t i.' See the orthography.txt file for the romanization scheme used. This package is Free Software, licensed under Lesser General Public License v.2.1 (www.fsf.org). See LICENSE for details. Note: the original PersianLG was made available under GPLv2; but, as of 2014, Jon Dehdari has consented to changing the license to LGPLv2.1 so that it can be included with the main Link Grammar project. [1] The stemmmer is located in the /morphology/fa directory. An example usage is: ./persianparse.sh 'man midAnam keh tu rafti.' or, equivalently: echo -e 'man midAnam keh tu rafti.\n' | ./stemmer.pl -u | link-parser fa