def make_luminoso_space(docDir, studyDir):
if not os.path.isdir(studyDir):
model = luminoso2.make_common_sense(studyDir, 'en')
else:
model = luminoso2.load(studyDir)
model.learn_from(docDir,studyDir)
return model
from csc import divisi2
import numpy as np
import json, luminoso2, time, urllib2
from divisi2 import DenseMatrix
from csc_utils.ordered_set import OrderedSet
from charm_exceptions import *
#model = luminoso2.load('pldb_2011_may')
model = luminoso2.load('hack_2011_oct/Model')
sponsormat = divisi2.load("sponsors.dmat")
doc_matrix = model.get_doc_matrix('pldb')
tag_matrix = model.get_tag_matrix()
tag_matrix = model.get_tag_matrix()
tag_matrix = DenseMatrix.concatenate(tag_matrix, sponsormat)
for i in xrange(doc_matrix.shape[0]):
doc_matrix.row_labels[i] = doc_matrix.row_labels[i].replace('hack_2011_oct/Documents', 'PLDBDocs')
# print doc_matrix.row_labels[i]
def get_related_sponsors(email, n=10):
if not ('sponsor', email) in tag_matrix.row_labels:
return []
vec = tag_matrix.row_named(('sponsor', email))
got = divisi2.dot(tag_matrix, vec)
results = []
for tag, weight in got.top_items(len(got)):
key, value = tag
if key == 'sponsor':
results.append((value, weight))
if len(results) >= n: