Using NLP to quantify the environmental cost and diversity benefits of in-person NLP conferences

Findings of the Association for Computational Linguistics: ACL 2022, pages 3853 - 3863 May 22-27, 2022 c⃝2022 Association for Computational Linguistics Using NLP to quantify the environmental cost and diversity benefits of in-person NLP conferences Piotr Przybyła Institute of Computer Science, Polish Academy of Sciences Warsaw, Poland piotr.przybyla@ipipan.waw.pl Matthew Shardlow Department of Computing and Mathematics, Manchester Metropolitan University Manchester, UK m.shardlow@mmu.ac.uk Abstract The environmental costs of research are pro- gressively important to the NLP community and their associated challenges are increasingly debated. In this work, we analyse the carbon cost (measured as CO2-equivalent) associated with journeys made by researchers attending in-person NLP conferences. We obtain the nec- essary data by text-mining all publications from the ACL anthology available at the time of the study (n=60,572) and extracting information about an author’s affiliation, including their address. This allows us to estimate the cor- responding carbon cost and compare it to pre- viously known values for training large models. Further, we look at the benefits of in-person conferences by demonstrating that they can in- crease participation diversity by encouraging attendance from the region surrounding the host country. We show how the trade-off between carbon cost and diversity of an event depends on its location and type. Our aim is to foster further discussion on the best way to address the joint issue of emissions and diversity in the future. 1 Introduction Figure 1 shows the increase in travel to the ACL an- nual meeting over the past 40 years. Whereas con- ferences used to be the privilege of a few academics, they are now attended by participants from com- panies, research institutes and universities across the world. This comes with an increase in the total volume of work published, and with it an increase in the carbon emissions attributed to travelling to in-person events. In this study we seek to quantify the impact of conferences that are increasingly diverse in terms of participation and location (undoubtedly benefi- cial) on the increased carbon emissions (undoubt- edly detrimental). We base our analysis on publica- tions spanning 55 years (1965–2020), taken from (a) ACL 1979: La Jolla, California, USA (b) ACL 1999: College Park, Maryland, USA (c) ACL 2019: Florence, Italy Figure 1: Visualisation of estimated journeys to the ACL annual meetings over 40 years. Maps for all major NLP conferences are included in the supplementary material. the ACL Anthology1. We use NLP tools to parse each document and identify the locations of the conference venues and lead researcher’s institution. We answer the following questions: 1. Where is NLP research performed and pre- sented? 2. What are the environmental costs? 3. Do conferences increase local participation? 4. Which events attract a diverse audience and how do they compare to non-physical venues? To the best of our knowledge, our work is the first to quantitatively explore the relationship be- tween the location of conferences in a research field 1https://www.aclweb.org/anthology/ 3853 and diversity of participation. We make our dataset and code available2 to enable further discussion on …

Content truncated to 500 words