In this paper, we explore a new approach based on discourse analysis for the task of intent segmentation.
Particularly, we design a two-stage learning method to effectively train the model using non-parallel data.
Joint intent detection and slot filling has recently achieved tremendous success in advancing the performance of utterance understanding.
We propose a learning-exploring method to generate sentences as learning objectives from the learned data distribution, and employ reinforcement learning to combine these new learning objectives for model training.
Unsupervised text style transfer is full of challenges due to the lack of parallel data and difficulties in content preservation.
Named Entity Recognition (NER) in domains like e-commerce is an understudied problem due to the lack of annotated datasets.
Ellipsis and co-reference are common and ubiquitous especially in multi-turn dialogues.
This paper focuses on learning both local semantic and global structure representations for text classification.