Deep Neural Architecture with Character Embedding for Semantic Frame Detection

摘要

Semantic frame detection has been extensively used for language understanding tasks, such as in dialogue systems or more recently, in Chat-bots. Traditionally, this involves two separate tasks: the detection of the semantic frame (i.e. intent detection), and the detection of frame elements (i.e. slot-filling). Recent efforts have attempted to combine the two tasks using recurrent neural networks. However, there is still room for improvement as these efforts do not efficiently model temporal dependencies. In this paper we propose a deep neural network architecture that uses long-short term memory (RNN-LSTM) with character embedding for joint modeling of intent detection and slot filling. Our results show significant improvement in slot-filling and intent detection at the sentence level over state-of-the-art semantic frame detection methods.