Breaking NLP: Using Morphosyntax, Semantics, Pragmatics and World Knowledge to Fool Sentiment Analysis Systems

Abstract

This paper describes our "breaker" submission to the 2017 EMNLP "Build It Break It" shared task on sentiment analysis. In order to cause the "builder" systems to make incorrect predictions, we edited items in the blind test data according to linguistically interpretable strategies that allow us to assess the ease with which the builder systems learn various components of linguistic structure. On the whole, our submitted pairs break all systems at a high rate (72.6%), indicating that sentiment analysis as an NLP task may still have a lot of ground to cover. Of the breaker strategies that we consider, we find our semantic and pragmatic manipulations to pose the most substantial difficulties for the builder systems.

Stance Detection Benchmark: How Robust is Your Stance Detection?

“…Question-Answering [49], Machine Translation [4], or Fact Checking [1,44]. Unfortunately, preserving the semantics of a sentence while automatically generating these adversarial attacks is difficult, which is why some works have defined small stress tests manually [19,27]. As this is time (and resource) consuming, other work has defined heuristics with controllable outcome to modify existing datasets and to preserve the semantics of the data [31].…”

DagoBERT: Generating Derivational Morphology with a Pretrained Language Model

“…Morphological productivity has been the subject of much work in psycholinguistics since it reveals implicit cognitive generalizations (see Dal and Namer (2016) for a review), making it an interesting phenomenon to explore in PLMs. Furthermore, in the context of NLP applications such as sentiment analysis, productively formed derivatives are challenging because they tend to have very low frequencies and often only occur once (i.e., they are hapaxes) or a few times in large corpora (Mahler et al, 2017). Our focus on productive derivational morphology has crucial consequences for dataset design (Section 3) and model evaluation (Section 4) in the context of DG.…”

Probing What Different NLP Tasks Teach Machines about Function Word Comprehension

“…Using recast NLI, Poliak et al (2018a) probe for semantic phenomena in neural machine translation encoders. Staliūnaite and Bonfil (2017); Mahler et al (2017);Ribeiro et al (2018) use similar strategies to our structural mutation method, although their primary goal was to break existing systems by adversarial modifications rather than to compare different models. Ribeiro et al (2018) and our work both test for proper comprehension of the modified expressions, but our modifications are designed to induce semantic changes whereas their modifications are intended to preserve the original meaning.…”