def KMeansPredict(self,details,vec): # Details: centers , xtoc , distances return Kmeans_l1.nearestcentres(X = [vec], centres = details[0], metric="cosine")
mv = len(model.vocab)
level = 1
while threshold < mv:
# KMeans
print()
print()
print(str(numInter) + " more iterations left ...")
print()
t0 = time.time()
curVecs = list(set(curVecs) - set(Rseeds))
#L2# kmeans = cluster.KMeans(n_clusters=len(Rseeds), init=np.array(Rseeds), max_iter = 1)
#kmeans.fit(np.array(curVecs))
centers, xtoc, distances = Kmeans_l1.kmeans(np.array(curVecs),
np.array(Rseeds),
maxiter=1,
metric='cosine')
#print(kmeans)
print(str(time.time() - t0))
#print("Level "+ str(level) + "\nIntertia = " + str(kmeans.inertia_))
print("Level " + str(level) + " Done.")
kmeans = [centers, xtoc]
save_obj(kmeans, path + "Level" + str(level))
print("Mapping Centroids to Labels")
centroid2Word = dict()
for c in Rseeds:
w = words[allseeds.index(tuple(c))]
centroid2Word[Rseeds.index(c)] = w
save_obj(centroid2Word, path + "Level" + str(level) + "c2w")
threshold = len(Rseeds)
mv = len(model.vocab)
level = 1
while threshold < mv:
# KMeans
print()
print()
print(str(numInter) + " more iterations left ...")
print()
t0 = time.time()
curVecs = list(set(curVecs) - set(Rseeds))
#L2# kmeans = cluster.KMeans(n_clusters=len(Rseeds), init=np.array(Rseeds), max_iter = 1)
#kmeans.fit(np.array(curVecs))
centers , xtoc , distances = Kmeans_l1.kmeans(np.array(curVecs), np.array(Rseeds), maxiter=1, metric='cosine')
#print(kmeans)
print(str(time.time()-t0))
#print("Level "+ str(level) + "\nIntertia = " + str(kmeans.inertia_))
print("Level "+ str(level) + " Done.")
kmeans = [centers,xtoc]
save_obj(kmeans,path+"Level"+str(level))
print("Mapping Centroids to Labels")
centroid2Word = dict()
for c in Rseeds:
w = words[allseeds.index(tuple(c))]
centroid2Word[Rseeds.index(c)] = w
save_obj(centroid2Word,path+"Level"+str(level)+"c2w")
print()