Social Chemistry 101: Learning to Reason about Social and Moral Norms (2011.00620v3)

Abstract: 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.

An Overview of Social Chemistry 101: Learning to Reason about Social and Moral Norms

The paper presented in "Social Chemistry 101: Learning to Reason about Social and Moral Norms" undertakes the significant task of modeling human norms—specifically, the social and moral norms that permeate everyday interactions and narratives. Norms are imperative for comprehending social behaviors, providing unspoken rules that guide human interaction. This paper introduces Social Chemistry as a novel framework for understanding such norms, proposing a systematic way to catalog these social expectations through intuitive and applicable rules-of-thumb (RoTs).

Core Contributions

1. Social-Chem-101 Corpus: This work introduces the Social-Chem-101 corpus, encompassing 292,000 rules-of-thumb across a wide array of social contexts. These are derived from 104,000 real-world situations, amassing over 4.5 million annotations. Each rule provides an evaluation or judgment about a social action, grounded through 12 distinct dimensions such as moral judgment, legality, and cultural pressure.
2. Neural Norm Transformer: The authors present a model built on state-of-the-art neural LLMs equipped to generate and reason about these RoTs. The model, termed the Neural Norm Transformer, learns from the Social-Chem-101 corpus to predict norms applicable to unseen situations.
3. Rule-of-Thumb Breakdown: Each RoT is decomposed into actions and judged across multiple dimensions, revealing perspectives that help illuminate why a particular norm is applied to a scenario. By disaggregating each RoT in terms of key attributes, this paper provides an enriched understanding of social norms.

Empirical Results and Challenges

The research showcases state-of-the-art approaches, utilizing architectures like GPT-2, BART, and T5. The authors conduct extensive experiments investigating how these models can predict plausible norms. Although these models articulated relevant RoTs and actions, the comprehensive conditioning on multiple attributes poses substantial difficulties. Human evaluations highlighted a range of scores, illustrating that despite progress, AI systems still grapple with the intricate task of adhering to varied conditional constraints in language understanding and generation.

Theoretical and Practical Implications

The implications of this research are multifaceted. Theoretically, Social Chemistry frameworks promise new insights into the intersection of computational linguistics, social psychology, and ethical reasoning. Practically, the development of AI systems that understand and respect social norms holds potential for applications in areas ranging from dialogue systems to human-AI collaboration tools. However, it is noteworthy that social norms can be culturally sensitive, emphasizing the need for careful cultural contextualization in training datasets and models.

Future Directions

Speculation on future directions stemming from this research includes the expansion of the Social-Chem-101 framework to encompass diverse cultural norms and the refinement of neural models that accurately capture the multi-dimensionality of social rules across varied contexts. Moreover, integrating such models into practical systems requires addressing biases, ensuring fairness, and maintaining ethical considerations.

In summary, "Social Chemistry 101" presents a compelling exploration into the computational modeling of social norms, providing foundational resources and methodologies that mirror the complexities and nuances intrinsic to human social interactions.