Analyzing Compositionality-Sensitivity of NLI Models

By Yixin Nie, Yicheng Wang, Mohit Bansal. 2018

83 citations recorded by the source corpus.

Success in natural language inference (NLI) should require a model to understand both lexical and compositional semantics. However, through adversarial evaluation, we find that several state-of-the-art models with diverse architectures are over-relying on the former and fail to use the latter. Further, this compositionality unawareness is not reflected via standard evaluation on current datasets. We show that removing RNNs in existing models or shuffling input words during training does not induce large performance loss despite the explicit removal of compositional information. Therefore, we propose a compositionality-sensitivity testing setup that analyzes models on natural examples from existing datasets that cannot be solved via lexical features alone (i.e., on which a bag-of-words model gives a high probability to one wrong label), hence revealing the models’ actual compositionality awareness. We show that this setup not only highlights the limited compositional ability of current NLI models, but also differentiates model performance based on design, e.g., separating shallow bag-of-words models from deeper, linguistically-grounded tree-based models. Our evaluation setup is an important analysis tool: complementing currently existing adversarial and linguistically driven diagnostic evaluations, and exposing opportunities for future work on evaluating models’ compositional understanding.

Subfields: Artificial Intelligence, 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 Natural Language Processing

Citations in the corpus

Right for the Wrong Reasons: Diagnosing Syntactic Heuristics in Natural Language Inference (2019, neutral): “Similar in spirit to our lexical overlap heuristic, Dasgupta et al. (2018), Nie et al. (2018), and Kim et al. (2018) also studied the performance of NLI models on speciﬁc phenomena where word order matters; however, we use a larger set of phenomena with the aim of studying lexical overlap in…”
An Empirical Study on Robustness to Spurious Correlations using Pre-trained Language Models (2020, background): “lifu-tu/Study-NLP-Robustness poorly on the so-called challenging or adversarial datasets where such correlations no longer hold (Glockner et al., 2018; McCoy et al., 2019; Nie et al., 2019; Zhang et al., 2019).”
DynaSent: A Dynamic Benchmark for Sentiment Analysis (2020, background): “These efforts seek to determine whether models have met specific learning targets (Alzantot et al., 2018; Glockner et al., 2018; Naik et al., 2018; Nie et al., 2019), exploit relatively superficial properties of their training data, (Jia and Liang, 2017; Kaushik and Lipton, 2018; Zhang et al.”
ANLIzing the Adversarial Natural Language Inference Dataset (2020, methodology): “These top-level types were introduced for smaller subsets of the ANLI development set in § 5 of Nie et al. (2020a), which we drastically expand both in number and specificity of tag types, as well as in the scope of annotation.”
Towards Robustifying NLI Models Against Lexical Dataset Biases (2020, background): “…set, we also add these samples back to the training set, similar to the concept of adversarial training (Jia and Liang, 2017; Wang et al., 2019c; Niu and Bansal, 2018) where the adversarial examples are added back to the training set so that the resulting model will be more robust to similar…”
Machine Learning Testing: Survey, Landscapes and Horizons (2019, background): “There has also been research on generating adversarial inputs for NLI models [98], [99](Section 8.”
Generating Data to Mitigate Spurious Correlations in Natural Language Inference Datasets (2022, background): “• IS-CS: lexically misleading instances constructed by Nie et al. (2019).”
Right for the Right Reason: Evidence Extraction for Trustworthy Tabular Reasoning (2022, background): “Additionally, NLI models (e.g. Ribeiro et al., 2016, 2018a,b; Zhao et al., 2018; Iyyer et al., 2018; Glockner et al., 2018; Naik et al., 2018; McCoy et al., 2019; Nie et al., 2019; Liu et al., 2019a) must be subjected to numerous test sets with adversarial settings.”
Beyond Leaderboards: A survey of methods for revealing weaknesses in Natural Language Inference data and models (2020, methodology): “…Nie et al., 2019a) Data Improvements (Mitra et al., 2020) HELP Data Improvements (Yanaka et al., 2019b) SNLI Stress-test (Glockner et al., 2018; Nie et al., 2019a; Richardson et al., 2019) Data Improvements (Kang et al., 2018; Mitra et al., 2020; Kaushik et al., 2020) Heuristics (Gururangan et…”
The Curse of Performance Instability in Analysis Datasets: Consequences, Source, and Suggestions (2020, background): “, 2018), some are more syntactic-related (McCoy et al., 2019b; Nie et al., 2019; Geiger et al., 2019), some also involved logical-related study (Minervini and Riedel, 2018; Wang et al.”