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

Summary

This paper investigates how much work in Natural Language Processing (NLP) is focused on solving social problems, often called 'NLP for Social Good'. It looks at who is doing this work – researchers typically involved in the main NLP conferences versus others – and where they are publishing their findings.

What's the problem?

While there's growing interest in using NLP to address issues like those outlined in the UN's Sustainable Development Goals, it wasn't clear *who* was actually doing this work and *where* it was being shared. Specifically, the researchers wanted to know if the core NLP research community was actively contributing to 'NLP for Social Good' and if their contributions were happening within their usual conferences or elsewhere.

What's the solution?

The researchers analyzed a large collection of NLP papers to see how often topics related to social good were addressed. They then categorized the authors based on whether they regularly present at the major NLP conference (ACL) and the venues where the papers were published (ACL conferences versus other places). This allowed them to compare the amount of 'NLP for Social Good' work done by ACL insiders versus outsiders, and within versus outside of ACL venues.

Why it matters?

The study found that researchers affiliated with the main NLP community are much more likely to work on social good problems when they publish *outside* of the main NLP conferences. Furthermore, the vast majority of 'NLP for Social Good' research is actually done by researchers *not* typically involved in the core NLP conferences and published in other venues. This suggests the main NLP community might need to rethink how it prioritizes and supports work addressing social issues, and that important contributions are happening beyond the usual channels.