Towards Linguistically Generalizable NLP Systems: A Workshop and Shared Task

By Allyson Ettinger, Sudha Rao, Hal Daumé Iii, Emily M. Bender. 2017

95 citations recorded by the source corpus.

This paper presents a summary of the first Workshop on Building Linguistically Generalizable Natural Language Processing Systems, and the associated Build It Break It, The Language Edition shared task. The goal of this workshop was to bring together researchers in NLP and linguistics with a carefully designed shared task aimed at testing the generalizability of NLP systems beyond the distributions of their training data. We describe the motivation, setup, and participation of the shared task, provide discussion of some highlighted results, and discuss lessons learned.

Subfields: Artificial Intelligence, Information Systems, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Natural Language Processing

Citations in the corpus

GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding (2018, methodology): “, 1996) and the recent BuildIt-Break-It competition (Ettinger et al., 2017), we include a small, manually-curated test set to allow for fine-grained analysis of system performance on a broad range of linguistic phenomena.”
Adversarial NLI: A New Benchmark for Natural Language Understanding (2019, methodology): “Given this, human annotators—be they seasoned NLP researchers or non-experts— might easily be able to construct examples that expose model brittleness.”
AI and the Everything in the Whole Wide World Benchmark (2021, background): “Finally, adversarial testing seeks to explore the edges of a system’s competence by finding minimally contrasting pairs of examples where the system being tested succeeds on one member of the pair and fails on the other [Ettinger et al., 2017].”
Build it Break it Fix it for Dialogue Safety: Robustness from Adversarial Human Attack (2019, methodology): “As mentioned in the introduction, our approach takes inspiration from “Build it Break it” approaches which have been successfully tried in other domains (Ruef et al., 2016; Ettinger et al., 2017).”
CommonsenseQA 2.0: Exposing the Limits of AI through Gamification (2021, neutral): “Examples for this include 368 [39, 40, 41, 12, 42, 34, 43, 38, 44], with recent works featuring re-training during dataset collection 369 (Beat-the-AI [22], StrategyQA [23] and Dynabench [21]).”
Models in the Loop: Aiding Crowdworkers with Generative Annotation Assistants (2021, background): “Separately, Ettinger et al. (2017) provide breakers with the ability to minimally edit original data to identify the boundaries of system capabilities, while Potts et al. (2021) analyse the use of prompts to assist crowdworkers in beating a model in the loop for sentiment analysis.”
Beat the AI: Investigating Adversarial Human Annotation for Reading Comprehension (2020, methodology): “Examples of this method are realised in Build It Break It, The Language Edition (Ettinger et al., 2017), HotpotQA (Yang et al.”
Dynabench: Rethinking Benchmarking in NLP (2021, background): “…are susceptible to adversarial attacks, and despite impressive task-level performance, state-of-the-art systems still struggle to learn robust representations of linguistic knowledge (Ettinger et al., 2017), as also shown by work analyzing model diagnostics (Ettinger, 2020; Ribeiro et al., 2020).”
Evaluating Models’ Local Decision Boundaries via Contrast Sets (2020, background): “Challenge Sets The idea of creating challenging contrastive evaluation sets has a long history (Levesque et al., 2011; Ettinger et al., 2017; Glockner et al., 2018; Naik et al., 2018; Isabelle et al., 2017).”
Adversarial Example Generation with Syntactically Controlled Paraphrase Networks (2018, background): “Recent work has shown it is possible to easily “break” many learned models by evaluating them on adversarial examples (Goodfellow et al., 2015), which are generated by manually introducing lexical, pragmatic, and syntactic variation not seen in the training set (Ettinger et al., 2017).”
Assessing Composition in Sentence Vector Representations (2018, background): “, 2016), and work submitting adversarial examples to systems in order to identify weaknesses (Zhao et al., 2017; Jia and Liang, 2017; Ettinger et al., 2017).”
The FEVER2.0 Shared Task (2019, background): “Such vulnerabilities were also the motivation behind Ettinger et al. (2017)’s NLP shared task that was inspired by the Build It, Break It, Fix It competition 1 .”
What Will it Take to Fix Benchmarking in Natural Language Understanding? (2021, background): “The Buildit-Break-it challenge (Ettinger et al., 2017), the Open Reading Benchmark (Dua et al.”
Probing What Different NLP Tasks Teach Machines about Function Word Comprehension (2019, methodology): “Ettinger et al. (2016, 2017); Zhu et al. (2018), i.a., use a task-based approach similar to ours, where tasks that require a speciﬁc subset of linguistic knowledge are used to perform qualitative evaluation.”
Entity-Based Knowledge Conflicts in Question Answering (2021, background): “Overstability is also relevant to minimal pairs (Ettinger et al., 2017), contrast sets (Gardner et al., 2020), and counterfactually-created data (Kaushik et al., 2020).”
Cycle-Consistency for Robust Visual Question Answering (2019, background): “Robustness has also been studied in natu-ral language processing (NLP) systems [8, 13] in contexts of bias [39, 38], domain-shift [25] and syntactic variations [16].”
Achieving Verified Robustness to Symbol Substitutions via Interval Bound Propagation (2019, background): “Recent work (Jia and Liang, 2017; Belinkov and Bisk, 2018; Ettinger et al., 2017) also exposed the vulnerabilities of neural NLP models, e.”