Skyler Hallinan, Faeze Brahman, Ximing Lu, Jaehun Jung, Sean Welleck, Yejin Choi. Findings of the Association for Computational Linguistics: EMNLP 2023. 2023.

Typical diets include an assortment of unprocessed, processed, and ultra-processed foods, along with culinary ingredients. Linear programming (LP) can be used to generate nutritionally adequate food patterns that meet pre-defined nutrient guidelines. The present LP models were set to satisfy 22 nutrient standards, while minimizing deviation from the mean observed diet of the Seattle Obesity Study (SOS III) sample. Component foods from the Fred Hutch food frequency questionnaire comprised the market basket. LP models generated optimized 2000 kcal food patterns by selecting from all foods, unprocessed foods only, ultra-processed foods only, or some other combination. Optimized patterns created using all foods contained less fat, sugar, and salt, and more vegetables compared to the SOS III mean. Ultra-processed foods were the main sources of added sugar, saturated fat and sodium. Ultra-processed foods also contributed most vitamin E, thiamin, niacin, folate, and calcium, and were the main sources of plant protein. LP models failed to create optimal diets using unprocessed foods only and ultra-processed foods only: no mathematical solution was obtained. Relaxing the vitamin D criterion led to optimized diets based on unprocessed or ultra-processed foods only. However, food patterns created using unprocessed foods were significantly more expensive compared to those created using foods in the ultra-processed category. This work demonstrates that foods from all NOVA categories can contribute to a nutritionally adequate diet.

While text style transfer has many applications across natural language processing, the core premise of transferring from a single source style is unrealistic in a real-world setting. In this work, we focus on arbitrary style transfer: rewriting a text from an arbitrary, unknown style to a target style. We propose STEER: Unified Style Transfer with Expert Reinforcement, a unified frame-work developed to overcome the challenge of limited parallel data for style transfer. STEER involves automatically generating a corpus of style-transfer pairs using a product of experts during decoding. The generated offline data is then used to pre-train an initial policy before switching to online, off-policy reinforcement learning for further improvements via fine-grained reward signals. STEER is unified and can transfer to multiple target styles from an arbitrary, unknown source style, making it particularly flexible and efficient. Experimental results on a challenging dataset with text from a diverse set of styles demonstrate state-of-the-art results compared to competitive baselines. Remarkably, STEER outperforms the 175B parameter instruction-tuned GPT-3 on overall style transfer quality, despite being 226 times smaller in size. We also show STEER is robust, maintaining its style transfer capabilities on out-of-domain data, and surpassing nearly all baselines across various styles. The success of our method highlights the potential of RL algorithms when augmented with controllable decoding to overcome the challenge of limited data supervision.

Even to a simple and short news headline, readers react in a multitude of ways: cognitively (e.g. inferring the writer's intent), emotionally (e.g. feeling distrust), and behaviorally (e.g. sharing the news with their friends). Such reactions are instantaneous and yet complex, as they rely on factors that go beyond interpreting factual content of news. We propose Misinfo Reaction Frames (MRF), a pragmatic formalism for modeling how readers might react to a news headline. In contrast to categorical schema, our free-text dimensions provide a more nuanced way of understanding intent beyond being benign or malicious. We also introduce a Misinfo Reaction Frames corpus, a crowdsourced dataset of reactions to over 25k news headlines focusing on global crises: the Covid-19 pandemic, climate change, and cancer. Empirical results confirm that it is indeed possible for neural models to predict the prominent patterns of readers' reactions to previously unseen news headlines. Additionally, our user study shows that displaying machine-generated MRF implications alongside news headlines to readers can increase their trust in real news while decreasing their trust in misinformation. Our work demonstrates the feasibility and importance of pragmatic inferences on news headlines to help enhance AI-guided misinformation detection and mitigation.