NormBank: A Knowledge Bank of Situational Social Norms

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 7756–7776 July 9-14, 2023 ©2023 Association for Computational Linguistics NORMBANK: A Knowledge Bank of Situational Social Norms Caleb Ziems Jane Dwivedi-Yu Yi-Chia Wang Alon Y. Halevy Diyi Yang Stanford University, Meta AI {cziems, diyiy}@stanford.edu, {janeyu, yichiaw, ayh}@fb.com Abstract We present NORMBANK, a knowledge bank of 155k situational norms. This resource is designed to ground flexible normative reason- ing for interactive, assistive, and collabora- tive AI systems. Unlike prior commonsense resources, NORMBANK grounds each infer- ence within a multivalent sociocultural frame, which includes the setting (e.g., restaurant), the agents’ contingent roles (waiter, customer), their attributes (age, gender), and other phys- ical, social, and cultural constraints (e.g., the temperature or the country of operation). In total, NORMBANK contains 63k unique con- straints from a taxonomy that we introduce and iteratively refine here. Constraints then apply in different combinations to frame so- cial norms. Under these manipulations, norms are non-monotonic — one can cancel an in- ference by updating its frame even slightly. Still, we find evidence that neural models can help reliably extend the scope and coverage of NORMBANK. We further demonstrate the utility of this resource with a series of transfer experiments. For data and code, see https: //github.com/SALT-NLP/normbank 1 Introduction As AI systems continually evolve for human as- sistance and collaboration, they will increasingly operate within cultural and social spaces, and re- quire increasingly robust and flexible knowledge of social norms (Carlucci et al., 2015). From dia- logue systems (Molnár and Szüts, 2018; Vaidyam et al., 2019; Bavaresco et al., 2020; Grossman et al., 2019) to socially interactive robots (Fong et al., 2003; Deng et al., 2019) and augmented or mixed reality technologies (Anderson and Rainie, 2022), each could benefit from understanding how humans effectively communicate, make decisions, engage with requests, and broadly interact with others (Sun- stein, 1996; Sherif and Sherif, 1953). Work done at Meta AI Research Figure 1: What is special about NORMBANK? Norms are grounded by situational constraints—environmental and personal attributes, as well as roles and other behav- iors. In this example, drinking coffee is an encouraged activity in its prototypical context, for a customer in a cafe, but it is counternormative for a working barista to do so in the same cafe, or for a child-age student do so in a classroom. These represent only some of the non- monotonic normative inferences that are represented in NORMBANK. 7756 Natural language is flexible and highly expres- sive; thus it is a promising medium for encoding knowledge of social norms (Sap et al., 2019a). The goal of this work is to construct NORMBANK, a nat- ural language bank of social norms that will allow AI systems to reason about social situations under complex constraints. NORMBANK encodes 155k norms via scalable human annotation, bootstrapped with implicit knowledge from large pre-trained lan- guage models (LLMs). NORMBANK factors in two important consider- ations that have been previously overlooked. First: norms are not rigid …

Content truncated to 500 words