def generate_words(cluster_sz,
english_vector=english_vector,
normed_eng=normed_eng):
print "generating english vector of cluster size", cluster_sz
# generate english vector
#english_vector = random_idx.generate_RI_text_words(N, RI_letters, './lang_texts/texts_english/eng.txt')
# generate new string of letters
length = 30
alphy = utils.generate_ordered_clusters(alph, cluster_sz=cluster_sz)
gstr = alph[np.random.randint(len(alph))]
temp_str = gstr
for i in xrange(length):
max_idx = 0
maxabs = 0
for j in xrange(len(alphy)):
temp_str = gstr + alphy[j]
temp_id = random_idx.generate_RI_str(N, RI_letters, cluster_sz,
ordered, temp_str)
#temp_id += 1e1*np.random.randn(1,N)
temp_id /= np.linalg.norm(temp_id)
absy = np.abs(temp_id.dot(normed_eng.T))
#print temp_str, absy
if absy > maxabs:
max_idx = j
maxabs = absy
gstr += alphy[max_idx]
print len(gstr), maxabs, gstr
def generate_words(cluster_sz, english_vector=english_vector, normed_eng=normed_eng):
print "generating english vector of cluster size", cluster_sz
# generate english vector
#english_vector = random_idx.generate_RI_text_words(N, RI_letters, './lang_texts/texts_english/eng.txt')
# generate new string of letters
length = 30
alphy = utils.generate_ordered_clusters(alph, cluster_sz=cluster_sz)
gstr = alph[np.random.randint(len(alph))]
temp_str = gstr
for i in xrange(length):
max_idx = 0
maxabs = 0
for j in xrange(len(alphy)):
temp_str = gstr + alphy[j]
temp_id = random_idx.generate_RI_str(N,RI_letters,cluster_sz,ordered,temp_str)
#temp_id += 1e1*np.random.randn(1,N)
temp_id /= np.linalg.norm(temp_id)
absy = np.abs(temp_id.dot(normed_eng.T))
#print temp_str, absy
if absy > maxabs:
max_idx = j
maxabs = absy
gstr += alphy[max_idx]
print len(gstr), maxabs, gstr
def log_generate_RI_str(N, RI_letters, cluster_sz, ordered, string, alph=alphabet):
# generate RI vector for string
text_vector = random_idx.generate_RI_str(N, RI_letters, cluster_sz, ordered, string, alph)
text_vector = np.log2(text_vector)
return text_vector
def log_generate_RI_str(N, RI_letters, cluster_sz, ordered, string, alph=alphabet):
# generate RI vector for string
text_vector = random_idx.generate_RI_str(N, RI_letters, cluster_sz, ordered, string, alph)
text_vector = np.log2(text_vector)
return text_vector