In this blog post, our cohort2 CDT student Dayyán O’Brien writes about his experience presenting at a EMNLP conference in Suzhou, China, in November 2025.
Recently, I traveled to Suzhou, China, for the Conference on Empirical Methods in Natural Language Processing (EMNLP) to present two pieces of my work.
The first of these was DocHPLT, which I presented at the Conference on Machine Translation (WMT). This is currently the world’s largest multilingual parallel document corpus! I shared this work through both an oral presentation and a poster session. It was lovely meeting people, including those I worked with, from all across the field and learning about all the multilingual research happening.
My second presentation was at the MathNLP workshop, where I shared a non-archival paper on MatheMagic. This is a new type of benchmark designed to minimize the artificial performance gains models get from data leakage. We do this by making the correct answers deliberately contradictory to normal maths, forcing the models to genuinely reason rather than rely on memorized training data.
Outside of the conference itself, I made sure to take in the local sights. I spent time exploring the historic canals throughout Suzhou and wandering through the old districts in Shanghai. I even managed to visit Shanghai Disneyland!
Overall, it was a fantastic trip. I’ve never been to China before, so it was cool to travel somewhere so far and different!
Written by: Dayyán O’Brien
