William Held, Christopher Hidey, Fei Liu, Eric Zhu, Rahul Goel, Diyi Yang, Rushin Shah. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Semantic parsing (SP) is a core component of modern virtual assistants like Google Assistant and Amazon Alexa. While sequence-to-sequence based auto-regressive (AR) approaches are common for conversational SP, recent studies employ non-autoregressive (NAR) decoders and reduce inference latency while maintaining competitive parsing quality. However, a major drawback of NAR decoders is the difficulty of generating top-k (i.e., k-best) outputs with approaches such as beam search. To address this challenge, we propose a novel NAR semantic parser that introduces intent conditioning on the decoder. Inspired by the traditional intent and slot tagging parsers, we decouple the top-level intent prediction from the rest of a parse. As the top-level intent largely governs the syntax and semantics of a parse, the intent conditioning allows the model to better control beam search and improves the quality and diversity of top-k outputs. We introduce a hybrid teacher-forcing approach to avoid training and inference mismatch. We evaluate the proposed NAR on conversational SP datasets, TOP & TOPv2. Like the existing NAR models, we maintain the O(1) decoding time complexity while generating more diverse outputs and improving top-3 exact match (EM) by 2.4 points. In comparison with AR models, our model speeds up beam search inference by 6.7 times on CPU with competitive top-k EM.

Retraining modern deep learning systems can lead to variations in model performance even when trained using the same data and hyperparameters by simply using different random seeds.This phenomenon is known as model churn or model jitter.This issue is often exacerbated in real world settings, where noise may be introduced in the data collection process.In this work we tackle the problem of stable retraining with a novel focus on structured prediction for conversational semantic parsing.We first quantify the model churn by introducing metrics for agreement between predictions across multiple re-trainings.Next, we devise realistic scenarios for noise injection and demonstrate the effectiveness of various churn reduction techniques such as ensembling and distillation.Lastly, we discuss practical tradeoffs between such techniques and show that co-distillation provides a sweet spot in terms of churn reduction with only a modest increase in resource usage.

Tuhin Chakrabarty, Christopher Hidey, Smaranda Muresan. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.

Christopher Hidey, Tuhin Chakrabarty, Tariq Alhindi, Siddharth Varia, Kriste Krstovski, Mona Diab, Smaranda Muresan. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. 2020.

Tuhin Chakrabarty, Christopher Hidey, Smaranda Muresan, Kathy McKeown, Alyssa Hwang. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

Word pairs across argument spans have been shown to be effective for predicting the discourse relation between them.We propose an approach to distill knowledge from word pairs for discourse relation classification with convolutional neural networks by incorporating joint learning of implicit and explicit relations.Our novel approach of representing the input as word pairs achieves state-of-theart results on four-way classification of both implicit and explicit relations as well as one of the binary classification tasks.For explicit relation prediction, we achieve around 20% error reduction on the four-way task.At the same time, compared to a two-layered Bi-LSTM-CRF model, our model is able to achieve these results with half the number of learnable parameters and approximately half the amount of training time.

Automatic detection of persuasion in online discussion is key to understanding how social media is used. Predicting persuasiveness is difficult, however, due to the need to model world knowledge, dialogue, and sequential reasoning. We focus on modeling the sequence of arguments in social media posts using neural models with embeddings for words, discourse relations, and semantic frames. We demonstrate significant improvement over prior work in detecting successful arguments. We also present an error analysis assessing novice human performance at predicting persuasiveness.