Joint Slot Filling and Intent Detection in Spoken Language Understanding by Hybrid CNN-LSTM Model

We investigate the usage of hybrid convolutional and long- short-term memory neural networks for joint slot filling and intent detection in spoken language understanding. We propose a novel model that combines between convolutional neural networks, for their ability to detect complex features in the input sequences by applying filters to frames of these inputs, and recurrent neural networks taking in account the fact, that they can keep track of the long- and short- term dependencies in the input sequences. We choose to build a model for joint slot filling and intent detection, because we believe, that there is a strong relation between the intent and the semantic slots. A joint model can reflect this relation, figure it out and make use of it to enhance the prediction results. We use the Airline Travel Information System (ATIS) dataset to measure the performance of our model and compare it with the results of other models, as this dataset has become one of the most popular datasets for spoken language understanding problem.