The proposed model makes use of a dependency parser to generate pair of words with their dependency relation.
Then each word is converted to a vector using word embeddings, and each dependency relation to a 1-hot-encoding vector. Thus a triple
(e_0,rel, e_1)
is created, where e_0
and e_1
are the word-embeddings for the 2 words, and rel
is the encoding of the dependency relation.
Therefore, the input of our model is dual: a sequence of triples which represents the question and another sequence for the comment. These are then passed to a sentence encoder, which is a Recurrent Neural Network (RNN), that is used to return a single output aiming to represent the entire sequence. The RNN output of both questions and comments, along with a vector made up of additional features, are the inputs of the final feed-forward layers that perform the classification.
The following figure shows an outline of the model:
Dependency parsing of two sentences taken from a question and a comment in the training set. In this example the first input x(t)
of the RNN is going to be: ("is", SUBJ,"there","is", SUBJ,"It")
.