Authorship attribution with syntactic fragments. Cf. http://staff.science.uva.nl/~acranenb/clfl2012.pdf
Requirements:
Usage:
- run
runexpfederalist.sh
to download the federalist papers and evaluate on the disputed and co-authored papers.
alternatively, evaluate on larger set of texts with cross-validation:
- prepare a directory
books/
with one directory for each author, each containing one.txt
file per work in UTF-8. - edit
parseworks.sh
andparsefiles.sh
to set the right paths for Java and the Stanford parser - edit and run
runexpbooks.sh