Social Chemistry 101: Learning to Reason about Social and Moral Norms

By Maxwell Forbes, Jena D. Hwang, Vered Shwartz, Maarten Sap, Yejin Choi. 2020

352 citations recorded by the source corpus.

Social norms---the unspoken commonsense rules about acceptable social behavior---are crucial in understanding the underlying causes and intents of people's actions in narratives. For example, underlying an action such as "wanting to call cops on my neighbors" are social norms that inform our conduct, such as "It is expected that you report crimes." We present Social Chemistry, a new conceptual formalism to study people's everyday social norms and moral judgments over a rich spectrum of real life situations described in natural language. We introduce Social-Chem-101, a large-scale corpus that catalogs 292k rules-of-thumb such as "it is rude to run a blender at 5am" as the basic conceptual units. Each rule-of-thumb is further broken down with 12 different dimensions of people's judgments, including social judgments of good and bad, moral foundations, expected cultural pressure, and assumed legality, which together amount to over 4.5 million annotations of categorical labels and free-text descriptions. Comprehensive empirical results based on state-of-the-art neural models demonstrate that computational modeling of social norms is a promising research direction. Our model framework, Neural Norm Transformer, learns and generalizes Social-Chem-101 to successfully reason about previously unseen situations, generating relevant (and potentially novel) attribute-aware social rules-of-thumb.

Subfields: Artificial Intelligence, Sociology and Political Science, Natural Language Processing, Prompting & Reasoning

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 Sociology and Political Science

In Natural Language Processing

In Prompting & Reasoning

Citations in the corpus

Towards Safer Generative Language Models: A Survey on Safety Risks, Evaluations, and Improvements (2023, background): “To address these ethical challenges, researchers demonstrated that incorporating human moral principles, such as the Rule of Thumb, into models enhances LMs’ transparency and explain-ability when handling ethical issues (Forbes et al., 2020; Ziems et al., 2022; Kim et al., 2022).”
Latent Hatred: A Benchmark for Understanding Implicit Hate Speech (2021, methodology): “%) surrounding social norms (e.g. a dependant is inferior to a supplier ) (Forbes et al., 2020).”
Teach Me to Explain: A Review of Datasets for Explainable NLP (2021, background): “…something about the world do not use the term “explanation” (Zellers et al., 2018; Sap et al., 2019b; Zellers et al., 2019b; Huang et al., 2019; Fan et al., 2019; Bisk et al., 2020; Forbes et al., 2020), a few recent ones do: ART (Bhaga-vatula et al., 2020) and GLUCOSE (Mostafazadeh et al., 2020).”
Thinking Like a Skeptic: Defeasible Inference in Natural Language (2020, methodology): “The SOCIAL-CHEM-101 dataset of social norms (henceforth, Social Norms) compiles a collection of 292K crowdsourced natural language statements about commonsense social judgments made given everyday situations (Forbes et al., 2020).”
DREAM: Uncovering Mental Models behind Language Models (2021, background): “Unlike the “norm” ﬁeld in Moral stories, where a single general “norm” is applicable to both the immoral and moral and actions, our model exploits the richness of the Social Chemistry dataset to learn various social norms that are intended to be more speciﬁc to the given situation.”
On the Machine Learning of Ethical Judgments from Natural Language (2022, background): “C HEM -101 (Forbes et al., 2020), crowd-sourced dataset of rules of thumbs that are paired with an action and a judgment on the action; • M ORAL S TORIES (Emelin et al., 2020), a dataset built on top of SOCIAL - CHEM -101, where annotators were asked to write 7-sentence stories that include “moral”…”
SOTOPIA: Interactive Evaluation for Social Intelligence in Language Agents (2023, methodology): “Specifically, we select 20 vignettes from Social Chemistry (Forbes et al., 2020), 20 from Social IQa (Sap et al., 2019), 10 from Deal-or-no-Deal (Lewis et al., 2017), and 10 vignettes from Normbank (Ziems et al., 2023) to generate 60 scenarios focusing on general daily-life social interactions.”
ExplaGraphs: An Explanation Graph Generation Task for Structured Commonsense Reasoning (2021, background): “…2018, 2019), cultural, social, and physical commonsense understanding (Lin et al., 2018; Sap et al., 2019a,b; Bisk et al., 2020; Hwang et al., 2020; Forbes et al., 2020), pronoun disambiguation (Sakaguchi et al., 2020; Zhang et al., 2020), abductive commonsense reasoning (Bhagavatula et al., 2019)…”
Second Thoughts are Best: Learning to Re-Align With Human Values from Text Edits (2023, methodology): “To tackle this misalignment problem, prior work has proposed using binary answers [27, 57], rankings [17, 4], or ratings [70, 39] to model human value preferences.”
The Importance of Modeling Social Factors of Language: Theory and Practice (2021, background): “Consequently, most NLP models still fail to recognize social norms (for an exception, see Forbes et al. (2020)).”
Think about it! Improving defeasible reasoning by first modeling the question scenario. (2021, methodology): “…graph is then used for the QA task on three existing defeasible inference datasets from diverse domains, viz., δ - SNLI (nat-ural language inference) (Bowman et al., 2015), δ - SOCIAL (reasoning about social norms) (Forbes et al., 2020), and δ - ATOMIC (commonsense reasoning) (Sap et al., 2019).”
Explaining Toxic Text via Knowledge Enhanced Text Generation (2022, background): “Forbes et al. (2020) develop conceptual formalisms that rely on annotations about the input to generate text.”
The Moral Integrity Corpus: A Benchmark for Ethical Dialogue Systems (2022, neutral): “Using beam search, T-5 outperforms all other RoT generation models signiﬁcantly, and the success of the same model with nucleus sampling is consistent with Forbes et al. (2020).”
Aligning to Social Norms and Values in Interactive Narratives (2022, methodology): “Delphi is trained on COMMONSENSE NORM BANK, a knowledge repository of everyday ethics, sourced from five existing large-scale datasets, including SOCIAL CHEMISTRY (Forbes et al., 2020), ETHICS Commonsense Morality (Hendrycks et al.”
Natural Language Processing with Commonsense Knowledge: A Survey (2021, background): “C HEM -101 [25] is a large-scale corpus designed to enhance language models’ comprehension of the intents and underlying causes in human narratives.”