A Gradually Soft Multi-Task and Data-Augmented Approach to Medical Question Understanding
8762Authors
nan
Abstract
AbstractUsers of medical question answering systems often submit long and detailed questions, making it hard to achieve high recall in answer retrieval. To alleviate this problem, we propose a novel Multi-Task Learning (MTL) method with data augmentation for medical question understanding. We first establish an equivalence between the tasks of question summarization and Recognizing Question Entailment (RQE) using their definitions in the medical domain. Based on this equivalence, we propose a data augmentation algorithm to use just one dataset to optimize for both tasks, with a weighted MTL loss. We introduce gradually soft parameter-sharing: a constraint for decoder parameters to be close, that is gradually loosened as we move to the highest layer. We show through ablation studies that our proposed novelties improve performance. Our method outperforms existing MTL methods across 4 datasets of medical question pairs, in ROUGE scores, RQE accuracy and human evaluation. Finally, we show that our method fares better than single-task learning under 4 low-resource settings.
Keywords
Full Content
Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing, pages 128–138 August 1–6, 2021. ©2021 Association for Computational Linguistics 128 Conversations Are Not Flat: Modeling the Dynamic Information Flow across Dialogue Utterances Zekang Li12, Jinchao Zhang3, Zhengcong Fei12, Yang Feng12∗, Jie Zhou3 1 Key Laboratory of Intelligent Information Processing Institute of Computing Technology, Chinese Academy of Sciences (ICT/CAS) 2 University of Chinese Academy of Sciences 3 Pattern Recognition Center, WeChat AI, Tencent Inc, China {lizekang19g,feizhengcong,fengyang}@ict.ac.cn {dayerzhang,withtomzhou}@tencent.com Abstract Nowadays, open-domain dialogue models can generate acceptable responses according to the historical context based on the large-scale pre- trained language models. However, they gen- erally concatenate the dialogue history directly as the model input to predict the response, which we named as the flat pattern and ig- nores the dynamic information flow across di- alogue utterances. In this work, we propose the DialoFlow model, in which we introduce a dynamic flow mechanism to model the con- text flow, and design three training objectives to capture the information dynamics across di- alogue utterances by addressing the semantic influence brought about by each utterance in large-scale pre-training. Experiments on the multi-reference Reddit Dataset and DailyDi- alog Dataset demonstrate that our DialoFlow significantly outperforms the DialoGPT on the dialogue generation task. Besides, we pro- pose the Flow score, an effective automatic metric for evaluating interactive human-bot conversation quality based on the pre-trained DialoFlow, which presents high chatbot-level correlation (r = 0.9) with human ratings among 11 chatbots. Code and pre-trained mod- els will be public. 1 1 Introduction Recent intelligent open-domain chatbots (Adiwar- dana et al., 2020; Bao et al., 2020; Smith et al., 2020) have made substantial progress thanks to the rapid development of the large-scale pre-training approaches (Devlin et al., 2019; Radford et al., 2019; Brown et al., 2020) and the large amount of conversational data (Dinan et al., 2019; Baum- gartner et al., 2020; Smith et al., 2020). However, ∗Joint work with Pattern Recognition Center, WeChat AI, Tencent Inc. Yang Feng is the corresponding author. Work was done when Zekang Li and Zhengcong Fei were intern at WeChat AI. 1https://github.com/ictnlp/DialoFlow When was this break-in? I would like to report a break-in. It was at 7 am. Ik <latexit sha1_base64="MsQDzSneT4sAofd9WG3mLm5EQT4=">ACz3icjVHLSsNAFD3GV62vqks3wSK4KokIuiy60V0L9gF tKcl02g7Ni2SilFJx6w+41b8S/0D/wjtjCmoRnZDkzLn3nJl7rxt5IpGW9bpgLC4tr6zm1vLr G5tb24Wd3XoSpjHjNRZ6Ydx0nYR7IuA1KaTHm1HMHd/1eMdXah4bHiQiDazmOeMd3BoHoC+ZIotpt35FDtz+5mnZH3ULRKl6mfPAzkAR2aqEhRe0UMIhQ+O AJIwh4cJPS0YMNCRFwHE+JiQkLHOabIkzalLE4ZDrEj+g5o18rYgPbKM9FqRqd49MakNHFImpDyYsLqNFPHU+2s2N+8J9pT3W1Mfzfz8omVGBL7l26W+V+dqkWijzNdg6CaIs2o6ljmkuquqJubX6qS5BARp3CP4jFhpWzPptak+jaVW8dHX/TmYpVe5blpnhXt6 QB2z/HOQ/qxyXbKtnVk2 L5PBt1Dvs4wBHN8xRlXKCGnlHeMQTno2qcWvcGfefqcZCptnDt2U8fACMX5RM</latexit> <latexit sha1_base64="MsQDzSneT4sAofd9WG3mLm5EQT4=">ACz3icjVHLSsNAFD3GV62vqks3wSK4KokIuiy60V0L9gF tKcl02g7Ni2SilFJx6w+41b8S/0D/wjtjCmoRnZDkzLn3nJl7rxt5IpGW9bpgLC4tr6zm1vLr G5tb24Wd3XoSpjHjNRZ6Ydx0nYR7IuA1KaTHm1HMHd/1eMdXah4bHiQiDazmOeMd3BoHoC+ZIotpt35FDtz+5mnZH3ULRKl6mfPAzkAR2aqEhRe0UMIhQ+O AJIwh4cJPS0YMNCRFwHE+JiQkLHOabIkzalLE4ZDrEj+g5o18rYgPbKM9FqRqd49MakNHFImpDyYsLqNFPHU+2s2N+8J9pT3W1Mfzfz8omVGBL7l26W+V+dqkWijzNdg6CaIs2o6ljmkuquqJubX6qS5BARp3CP4jFhpWzPptak+jaVW8dHX/TmYpVe5blpnhXt6 QB2z/HOQ/qxyXbKtnVk2 L5PBt1Dvs4wBHN8xRlXKCGnlHeMQTno2qcWvcGfefqcZCptnDt2U8fACMX5RM</latexit> <latexit sha1_base64="MsQDzSneT4sAofd9WG3mLm5EQT4=">ACz3icjVHLSsNAFD3GV62vqks3wSK4KokIuiy60V0L9gF tKcl02g7Ni2SilFJx6w+41b8S/0D/wjtjCmoRnZDkzLn3nJl7rxt5IpGW9bpgLC4tr6zm1vLr G5tb24Wd3XoSpjHjNRZ6Ydx0nYR7IuA1KaTHm1HMHd/1eMdXah4bHiQiDazmOeMd3BoHoC+ZIotpt35FDtz+5mnZH3ULRKl6mfPAzkAR2aqEhRe0UMIhQ+O AJIwh4cJPS0YMNCRFwHE+JiQkLHOabIkzalLE4ZDrEj+g5o18rYgPbKM9FqRqd49MakNHFImpDyYsLqNFPHU+2s2N+8J9pT3W1Mfzfz8omVGBL7l26W+V+dqkWijzNdg6CaIs2o6ljmkuquqJubX6qS5BARp3CP4jFhpWzPptak+jaVW8dHX/TmYpVe5blpnhXt6 QB2z/HOQ/qxyXbKtnVk2 L5PBt1Dvs4wBHN8xRlXKCGnlHeMQTno2qcWvcGfefqcZCptnDt2U8fACMX5RM</latexit> <latexit sha1_base64="MsQDzSneT4sAofd9WG3mLm5EQT4=">ACz3icjVHLSsNAFD3GV62vqks3wSK4KokIuiy60V0L9gF tKcl02g7Ni2SilFJx6w+41b8S/0D/wjtjCmoRnZDkzLn3nJl7rxt5IpGW9bpgLC4tr6zm1vLr G5tb24Wd3XoSpjHjNRZ6Ydx0nYR7IuA1KaTHm1HMHd/1eMdXah4bHiQiDazmOeMd3BoHoC+ZIotpt35FDtz+5mnZH3ULRKl6mfPAzkAR2aqEhRe0UMIhQ+O AJIwh4cJPS0YMNCRFwHE+JiQkLHOabIkzalLE4ZDrEj+g5o18rYgPbKM9FqRqd49MakNHFImpDyYsLqNFPHU+2s2N+8J9pT3W1Mfzfz8omVGBL7l26W+V+dqkWijzNdg6CaIs2o6ljmkuquqJubX6qS5BARp3CP4jFhpWzPptak+jaVW8dHX/TmYpVe5blpnhXt6 QB2z/HOQ/qxyXbKtnVk2 L5PBt1Dvs4wBHN8xRlXKCGnlHeMQTno2qcWvcGfefqcZCptnDt2U8fACMX5RM</latexit> Figure 1: Illustration of the dynamic information flow in the semantic space. Ck (gray dotted line) denotes the dense representation of dialogue history which we named as context. Ik (pink solid line) denotes the se- mantic influence brought about by the k-th utterance, which is the difference between Ck and Ck+1. effectively modeling the dialogue history in large- scale dialogue pre-training is still challenging. Most of the previous work on dialogue history modeling mainly fall into two groups. One group of works generally concatenate the dialogue history as the model input and predict the response …
Content truncated to 500 words
Labeled Text
AbstractUsers of medical question answering systems often submit long and detailed questions, making it hard to achieve high recall in answer retrieval. To alleviate this problem, we propose a novel Multi-Task Learning (MTL) method with data augmentation for medical question understanding. We first establish an equivalence between the tasks of question summarization and Recognizing Question Entailment (RQE) using their definitions in the medical domain. Based on this equivalence, we propose a data augmentation algorithm to use just one dataset to optimize for both tasks, with a weighted MTL loss. We introduce gradually soft parameter-sharing: a constraint for decoder parameters to be close, that is gradually loosened as we move to the highest layer. We show through ablation studies that our proposed novelties improve performance. Our method outperforms existing MTL methods across 4 datasets of medical question pairs, in ROUGE scores, RQE accuracy and human evaluation. Finally, we show that our method fares better than single-task learning under 4 low-resource settings.
Publication Details
- Paper ID:
- 8762
- Year:
- 2021
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan