Good Intentions Beyond ACL: Who Does NLP for Social Good, and Where?

TL;DR

This study maps the landscape of NLP for Social Good (NLP4SG) across ACL and non-ACL venues, revealing that a majority of NLP4SG work is produced outside ACL and by non-ACL authors. By augmenting a large corpus with NLP relevance, venue type, and author classifications, the authors show pronounced distributional, topical, and methodological differences between venue types and disciplines. They demonstrate that ACL authors publish more NLP4SG outside ACL than inside, while non-ACL authors drive NLP4SG in external venues, with external NLP4SG work focusing more on health and education and less on neural methods than internal ACL work. The findings have governance implications for the ACL community, suggesting concrete steps like theme tracks and keynote inclusion to better integrate NLP4SG into ACL conferences, and they provide a replicable dataset to support future cross-venue analysis. Overall, the work highlights the breadth of NLP4SG beyond ACL and calls for intentional agenda-setting to balance impact across venues and disciplines.

Abstract

The social impact of Natural Language Processing (NLP) is increasingly important, with a rising community focus on initiatives related to NLP for Social Good (NLP4SG). Indeed, in recent years, almost 20% of all papers in the ACL Anthology address topics related to social good as defined by the UN Sustainable Development Goals (Adauto et al., 2023). In this study, we take an author- and venue-level perspective to map the landscape of NLP4SG, quantifying the proportion of work addressing social good concerns both within and beyond the ACL community, by both core ACL contributors and non-ACL authors. With this approach we discover two surprising facts about the landscape of NLP4SG. First, ACL authors are dramatically more likely to do work addressing social good concerns when publishing in venues outside of ACL. Second, the vast majority of publications using NLP techniques to address concerns of social good are done by non-ACL authors in venues outside of ACL. We discuss the implications of these findings on agenda-setting considerations for the ACL community related to NLP4SG.

Figures (5)

• Figure 1: A higher ratio of NLP papers outside of the ACL Anthology (external) are characterized as social good than in acl and acl-adjacent venues. Moreover, NLP papers by non-acl authors are more likely to focus on social good questions than those by acl authors across all venue types.
• Figure 2: Schematic of the metadata augmentation used in conducting our analyses. Papers are labeled for relevance to social good as defined by UN SDGs, author association with ACL, venue type, and neural vs. non-neural methods.
• Figure 3: Proportion of NLP papers identified as NLP4SG by venue type and year. Applications of NLP techniques to social good questions are increasing as a share of all NLP papers across all venues.
• Figure 4: Proportions of SDG topics across venue types. The targets of social good research have different distributions between acl-associated venues and other conferences. acl venues have a greater focus on topics related to peace, innovation, and inequality while external venues have a greater focus on those related to health and education.
• Figure 5: Distribution of neural methods within papers across venue and author types since 2017. Within the acl Anthology, NLP4SG papers are just as likely to rely on neural methods as those not focused on social good. In external venues, NLP4SG papers are relatively less likely to use neural methods.