NOTES.md

where I left: try to provide the module with minimum data and directory structure necessary to run some tests.

Tests

• testing good practices

For nicer CLIs

from termcolor import colored
colored('test', 'red')
print(colored('test', 'red'))
print(colored('✓', 'red'))
print(colored('✓', 'green'))

Distributing the package

• see http://pythonhosted.org/setuptools/setuptools.html

Installation problems:

to install SciPy on Ubuntu one needs:

sudo apt-get install gfortran libopenblas-dev liblapack-dev

then SciPy, then scikit-learn

Notes to implement Supervised Relation Detection

• working with many languages makes it more complicated to work with syntactic features as chunkers do not exist for all the languages we considered ()

• the training set should contain both positive and negative examples; to create a negative example out of a positive relation, e.g. "rel(arg1,arg2)" is enough to invert it, "rel(arg2,arg1)"

  class= (scope_pos | scope_neg)

ML CitationMatcher

cfr this thread in SO

to output a probability for each classification by SVM pass probabilities=True when

self._classifier = svm.SVC(
    kernel='linear',
    C=C,
    cache_size=cache_size
)

return the probabilities from citation_extractor.ned.ml::predict()