def classify_proba(self, form, threshold=0.0):
"""
Return dict with probabilities of ``form`` and its fields belonging
to various form and field classes::
{
'form': {'type1': prob1, 'type2': prob2, ...},
'fields': {
'name': {'type1': prob1, 'type2': prob2, ...},
...
}
}
``form`` should be an lxml HTML <form> element.
Only classes with probability >= ``threshold`` are preserved.
"""
form_types_proba = self.form_classifier.classify_proba(form, threshold)
form_type = max(form_types_proba, key=lambda p: form_types_proba[p])
field_elems = get_fields_to_annotate(form)
xseq = fieldtype_model.get_form_features(form, form_type, field_elems)
yseq = self._field_model.predict_marginals_single(xseq)
return {
'form': form_types_proba,
'fields': {
elem.name: thresholded(probs, threshold)
for elem, probs in zip(field_elems, yseq)
},
}
def classify_proba(self, form, threshold=0.0, fields=True):
"""
Return dict with probabilities of ``form`` and its fields belonging
to various form and field classes::
{
'form': {'type1': prob1, 'type2': prob2, ...},
'fields': {
'name': {'type1': prob1, 'type2': prob2, ...},
...
}
}
``form`` should be an lxml HTML <form> element.
Only classes with probability >= ``threshold`` are preserved.
If ``fields`` is False, only information about the form is returned::
{
'form': {'type1': prob1, 'type2': prob2, ...}
}
"""
form_types_proba = self.form_classifier.classify_proba(form, threshold)
res = {'form': form_types_proba}
if fields:
form_type = max(form_types_proba,
key=lambda p: form_types_proba[p])
field_elems = get_fields_to_annotate(form)
xseq = fieldtype_model.get_form_features(form, form_type,
field_elems)
yseq = self._field_model.predict_marginals_single(xseq)
res['fields'] = {
elem.name: thresholded(probs, threshold)
for elem, probs in zip(field_elems, yseq)
}
return res
def _probs2dict(self, probs, threshold):
return thresholded(dict(zip(self.classes, probs)), threshold)
def _probs2dict(self, probs, threshold):
return thresholded(dict(zip(self.classes, probs)), threshold)
