How to classify sentences using word2vec and keras?

Question

I have 2 lists of sentences. First list contains different questions, second contains different statements.

Little example:

1st list:
[
    "What are cultures that are associated with core values?",
    "How do bumblebees fly?",
    "It is possible that avocado seed can be a tea?",
    ...
]

2nd list:
[
    "The population was 388 at the 2010 census.",
    "Grevillea helmsiae is a tree which is endemic to Queensland in Australia.",
    "He played youth football for Tynecastle Boys Club.",
    ...
]

I want to write program which will be able to classify this 2 types of sentences. For this, I can create neural network and train it on my 2 lists. I guess, this must be recurrent neural network.

I have transformed each sentence to array of word2vec vectors. And now I want to set up keras recurrent neural network with LSTM layers. But I don't know how to do that correctly. Can you write keras model for this problem?

UPDATE

the form of above sentences after transforming it by word2vec is like this:

[
    [vector_of_floats_for_word_"what", vector_of_floats_for_word_"are", vector_of_floats_for_word_"cultures", vector_of_floats_for_word_"that", ...],
    [vector_of_floats_for_word_"how", vector_of_floats_for_word_"do", vector_of_floats_for_word_"bumblebees", ...]
]

and so on. each vector has 300 dimensions.

here is my model:

X = []
Y = []

for i in range(1000):
    X.append(questions_vectors[i])
    Y.append([1, 0])
    X.append(statements_vectors[i])
    Y.append([0, 1])

model = Sequential()
model.add(LSTM(128, input_shape=(2000, None, 300)))
model.add(Dense(2, activation='softmax'))
model.compile(loss='binary_crossentropy', optimizer=RMSprop(lr=0.01))

there you can see magic numbers 2000 and 300. 2000 is 1000 questions + 1000 statements, 300 - word vector length

but I'm sure that my model is wrong. also I'm getting the error:

ValueError: Input 0 is incompatible with layer lstm_1: expected ndim=3, found ndim=4

do you really need a neural net to classify questions from statements? looking at your dataset, just the last character would be enough to classify. Anyway, to help you with the model part, we need to see the code you used for the beginning, how you transformed these sentences. — Nassim Ben, Mar 17 '17 at 12:09
Of course, I deleted the last characters from the statements. — Dmytro Nalyvaiko, Mar 17 '17 at 12:18

How to classify sentences using word2vec and keras?

0 Answers0