def get_dbank_text_features(datum):
feat_dict = pos_phrases.get_all(datum)
feat_dict.update(pos_syntactic.get_all(datum))
feat_dict.update(
psycholinguistic.get_cookie_theft_info_unit_features(datum))
feat_dict.update(psycholinguistic.get_psycholinguistic_features(datum))
feat_dict.update(psycholinguistic.get_spatial_features(datum, 'halves'))
feat_dict.update(psycholinguistic.get_spatial_features(datum, 'strips'))
feat_dict.update(psycholinguistic.get_spatial_features(datum, 'quadrants'))
return feat_dict
def process_blog(blog, name):
posts = []
for post_id in blog:
post = blog[post_id]
if post:
print 'Processing %s' % post_id
feat_dict = pos_phrases.get_all(post)
feat_dict.update(pos_syntactic.get_all(post))
feat_dict.update(
psycholinguistic.get_psycholinguistic_features(post))
feat_dict["number_of_sentences"] = len(post)
feat_dict['blog'] = name
feat_dict['id'] = post_id
posts += [feat_dict]
return posts
def process_blog(blog, name):
posts = []
total = len(blog)
for i, post_id in enumerate(blog):
post = blog[post_id]
if post:
print 'Processing %s (%s / %s)' % (post_id, i + 1, total)
feat_dict = pos_phrases.get_all(post)
feat_dict.update(pos_syntactic.get_all(post))
feat_dict.update(
psycholinguistic.get_psycholinguistic_features(post))
feat_dict["number_of_sentences"] = len(post)
feat_dict['blog'] = name
feat_dict['id'] = post_id
posts += [feat_dict]
return posts
def get_blog_text_features(datum):
feat_dict = pos_phrases.get_all(datum)
feat_dict.update(pos_syntactic.get_all(datum))
feat_dict.update(psycholinguistic.get_psycholinguistic_features(datum))
return feat_dict