A Semantic-based Method for Unsupervised Commonsense Question Answering

Abstract

Unsupervised commonsense question answering is appealing since it does not rely on any labeled task data. Among existing work, a popular solution is to use pre-trained language models to score candidate choices directly conditioned on the question or context. However, such scores from language models can be easily affected by irrelevant factors, such as word frequencies, sentence structures, etc. These distracting factors may not only mislead the model to choose a wrong answer but also make it oversensitive to lexical perturbations in candidate answers.In this paper, we present a novel SEmanticbased Question Answering method (SEQA) for unsupervised commonsense question answering. Instead of directly scoring each answer choice, our method first generates a set of plausible answers with generative models (e.g., GPT-2), and then uses these plausible answers to select the correct choice by considering the semantic similarity between each plausible answer and each choice. We devise a simple, yet sound formalism for this idea and verify its effectiveness and robustness with extensive experiments. We evaluate the proposed method on four benchmark datasets, and our method achieves the best results in unsupervised settings. Moreover, when attacked by TextFooler (Jin et al., 2020) with synonym replacement, SEQA demonstrates much less performance drops than baselines, thereby indicating stronger robustness.

Great~Truths~are ~Always ~Simple: A Rather Simple Knowledge Encoder for Enhancing the Commonsense Reasoning Capacity of Pre-Trained Models A Rather Simple Knowledge Encoder for Enhancing the Commonsense Reasoning Capacity of Pre-Trained Models

“…Therefore, we report the experimental results on their development sets for a fair comparison (Shwartz et al, 2020). For CoPA that only provides development and test sets, we follow Niu et al (2021) to train models on the development set and evaluate the performance on the test set. For commonsense KG, we adopt Con-ceptNet (Speer et al, 2017), a general-domain and task-agnostic CSKG, as our external knowledge source G for all the above models and tasks.…”

CASE: Commonsense-Augmented Score with an Expanded Answer Space

“…This is computed as the conditional probability of the answer given a domain-specific prefix such as "The sentiment of the movie is" for sentiment analysis or "The answer is" for general QA tasks. SEQA (Niu et al, 2021) mitigates the sensitivity to word choice by generating answers using GPT-2, and selecting the answer choice most similar to the generated answers.…”

“…In multiple-choice question answering (MCQA) tasks, zero-shot methods typically rely on the language model (LM) probabilities as a proxy for plausibility, predicting the answer choice with the highest probability conditioned on the question. LM score is a naïve proxy for plausibility, since it confounds factors such as length, unigram frequency, and more (Holtzman et al, 2021;Niu et al, 2021). Indeed, in Figure 1, a GPT-2 based LM score incorrectly predicts that the woman hired a lawyer because she decided to run for office, rather than because she decided to sue her employer.…”

“…We further adopt the strategy suggested by Niu et al (2021) to expand the answer space by using a LM to generate additional answers and then mapping semantically-similar generated answers into the original space. This mitigates the LM score's sensitivity to infrequent words.…”

Commonsense Knowledge Reasoning and Generation with Pre-trained Language Models: A Survey

“…Semantic similarity. Niu et al (2021) show that semantic similarity matching can be used to make PLMs robust against irrelevant factors such as word frequencies. Specifically, we can first use PLMs to generate plausible answers so that we can compute the similarity between each generated answer and each of the provided answer candidates, and then we can select the answer candidate that has the highest similarity score as the correct answer.…”