Repository for the development of the Python-language successor to the TABARI event-data coding program. You can find the documentation for PETRARCH here. You can read more about event data here.

We seem to have settled on a few core functions at this point. Each function is designed to do a slightly different task.

#####parse

The parse command is the key functionality of PETRARCH. The expected input is the standard TABARI-formatted lead sentences. If the input is longer than one sentence, then the program will skip it. To run the parse command:

python petrarch.py parse -i SENTENCE_FILE -o output.txt 

#####parallel_parse

The parallel_parse command does the same thing as the parse command, but makes use of multiple cores. This command requires the Parallel Python library (pip install pp). To run:

python petrarch.py parallel_parse -i SENTENCE_FILE -o output.txt

#####batch_parse

The batch_parse command is designed to take as input a directory of article-length news stories, and parse them in the same manner as the single sentences in the other commands. The main difference between batch_parse and the other parse commands is the manner in which the program interacts with StanfordNLP. To run the command:

python petrarch.py batch_parse -i ~/sentence_directory -o output.txt