RNN_word_embeddings_Named_Entity_Classification_sample.py

#create an index of all the words (words2idx)
...
#create an index of all the labels (labels2idx)
...
#query to classify
q="hotel amsterdam wifi"
#labels to use "prop", "dest", "fac"

#this will contain all our training data
matrix=[]

#this array will contain all the indexes of the words
words=[]
#this array will contain all the indexes of the labels
labels=[]
#this will contain the arrays of indexes
sentence=[]

words.append(words2idx["hotel"])
labels.append(labels2idx['proptype'])

words.append(words2idx["amsterdam"])
labels.append(labels2idx['dest'])

words.append(words2idx["wifi"])
labels.append(labels2idx['fac'])

sentence.append(words)
sentence.append(labels)
sentence.append(labels)
matrix.append(sentence)

...
#the RNN settings
s = { 'fold':3, # 5 folds 0,1,2,3,4
      'lr':0.0627142536696559,
      'verbose':1,
      'decay':False, # decay on the learning rate 
      'win':7, # number of words in the context window
      'bs':9, # number of backprop through time steps
      'nhidden':100, # number of hidden units
      'seed':345,
      'emb_dimension':100, # dimension of word embedding
      'nepochs':50}
         
# instanciate the model
numpy.random.seed(s['seed'])
random.seed(s['seed'])
rnn = model(    nh = s['nhidden'],
                nc = nclasses,
                ne = vocsize,
                de = s['emb_dimension'],
                cs = s['win'] )
#separate in train, test, validation
...
#train
for i in xrange(nsentences):
    train_lex_list=train_lex[i].tolist()[0]
    cwords = contextwin(train_lex_list, s['win'])
    #print cwords
    words  = map(lambda x: numpy.asarray(x).astype('int32'),\
    minibatch(cwords, s['bs']))
    labels = train_y[i]
    cnt_w=0
    for word_batch , label_last_word in zip(words, labels):
    #print word_batch, label_last_word
    #cnt_w+=1
    #print cnt_w
    rnn.train(word_batch, label_last_word, s['clr'])
    rnn.normalize()
    if s['verbose']:
    print '[learning] epoch %i >> %2.2f%%'%(e,(i+1)*100./nsentences),'completed in %.2f (sec) <<\r'%(time.time()-tic),
    sys.stdout.flush()