Frustratingly Simple Few-Shot Slot Tagging

Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 1028–1033 August 1–6, 2021. ©2021 Association for Computational Linguistics 1028 Frustratingly simple few-shot slot tagging Jianqiang Ma∗ Platform and Content Group, Tencent alexanderma@tencent.com Zeyu Yan∗ AI Dept, Souche Inc. yanzeyu@souche.com Chang Li∗ AI Dept, Souche Inc. lichang@souche.com Yang Zhang AI Dept, Souche Inc. zhangyang@souche.com Abstract We propose a simple and effective few-shot model for slot tagging. Recent work shows that it is promising to extend standard few- shot classification methods to sequence label- ing with CRF-specific augmentations. Such methods show strengths in encoding slot name semantics and slot dependencies. However, we find these strengths can be obtained by a much simpler method, which casts slot tagging into machine reading comprehension (MRC). We fine-tune a standard BERT-based MRC model with a mixture of source domain and (few-shot) target domain data. Such simple method outperforms state-of-the-art methods by a large margin on the SNIPS dataset. 1 Introduction This paper considers the task of few-shot slot tag- ging. Slot tagging (Zhang and Wang, 2016; Hai- hong et al., 2019) is a core component for task- oriented dialog systems (Papineni et al., 2001), where the goal is to provide a fine-grained, struc- tured description of user request for a given intent. Example (1) shows the input-output of a slot tag- ging module for the book restaurant intent, where the module yields the semantic analysis for the input query in terms of slot label-value pairs such as restaurant type is ‘brasserie’ and time range is ‘15 minutes’, etc. (1) Query: I want to book a far brasserie that serves minestrone in PA for a party of 9 in 15 minutes. Tagged Slots: {restaurant type: ‘brasserie’, time range: ‘15 minutes’, state: ‘PA’, party size number: ‘9’, served dish: ‘minestrone’} In real world systems, slot taggers are required to rapidly cover new domains to address increasing ∗Equal contribution. Order decided by tossed coins. user needs. A key challenge here is that labeled data are often scarce in new domains and the high cost of manually annotating large-scale data be- comes a major obstacle for domain adaptation. An attractive alternative is few-shot learning, which aims at achieving reasonable good results using only a few labeled instances in the new domain. Although there are many successful few-shot classification methods, especially meta learning ones (Bapna et al., 2017; Luo et al., 2018; Fritzler et al., 2019), directly adapting them to slot tag- ging often yields unsatisfactory results (Hou et al., 2020). This is due to the sentence-level, extrac- tive nature of slot tagging, where token dependen- cies within sentences are important but ignored by token-level classification models. Recent work (Hou et al., 2020) tackles this problem by extending meta learning methods to sequence transduction within the BERT-CRF framework, using several CRF-specific augmentations. Such augmentations show strengths in both encoding slot name seman- tics and modeling slot dependencies, which are key elements for effective few-shot slot tagging. This paper shows that we can enjoy similar strengths with a frustratingly simple method. …

Content truncated to 500 words