Faculty and students from the Johns Hopkins Department of Computer Science and affiliated centers presented their recent findings in computational linguistics at the 61st Annual Meeting of the Association for Computational Linguistics, held July 9–14 in Toronto, Canada. The ACL is the premier international scientific and professional society for researchers working in the field of natural language processing and computation.
Among the Hopkins researchers’ accolades are two Outstanding Paper Awards, secured by an intercollegiate team including Daniel Khashabi, an assistant professor of computer science, and by a group of researchers from academia and industry that included recent CS alumnus Aaron Mueller ’20 (MS), ’23 (PhD).
Khashabi—along with UCLA researchers Nikil Roashan Selvam and Kei-Wei Chang and research scientists Sunipa Dev (Google Research) and Tushar Khot (the Allen Institute for AI)—investigated the unreliability of scores obtained from social bias benchmarks in “The Tail Wagging the Dog: Dataset Construction Biases of Social Bias Benchmarks.”
Joined by Meta AI researchers Koustuv Sinha, Keren Fuentes, and Adina Williams and PhD candidates Jon Gauthier (MIT) and Kanishka Misra (Purdue University), Mueller analyzed language modeling acceptability judgements with systematically manipulated contexts. He presented their findings, titled “Language model acceptability judgements are not always robust to context,” in person at the conference.
Additionally, a position paper by Arya McCarthy, a doctoral candidate in the Center for Language and Speech Processing, and Giovanna Maria Dora Dore, an associate teaching professor and the associate director of the Krieger School’s Program in East Asian Studies, received an honorable mention; “Theory-Grounded Computational Analysis” argues for a return to theoretically grounded research questions to promote better integration of NLP and the social sciences.
Other JHU ACL’23 presenters include:
- “On the Blind Spots of Model-Based Evaluation Metrics for Text Generation” by Tianxing He, Jingyu Zhang, Tianle Wang, Sachin Kumar, Kyunghyun Cho, James Glass, and Yulia Tsvetkov
- “Characterization of Stigmatizing Language in Medical Records” by Keith Harrigian, Ayah Zirikly, Brant Chee, Alya Ahmad, Anne R. Links, Somnath Saha, Mary Catherine Beach, and Mark Dredze
“Contrastive Decoding: Open-Ended Text Generation as Optimization” by Xiang Lisa Li, Ari Holtzman, Daniel Fried, Percy Liang, Jason Eisner, Tatsunori Hashimoto, Luke Zettlemoyer, and Mike Lewis
- “The Effect of Alignment Correction on Cross-Lingual Annotation Projection” by Shabnam Behzad, Seth Ebner, Marc Marone, Benjamin Van Durme, and Mahsa Yarmohammadi
- “JHU IWSLT 2023 Multilingual Speech Translation System Description” by Henry Li Xinyuan, Neha Verma, Bismarck Bamfo Odoom, Ujvala Pradeep, Matthew Wiesner, and Sanjeev Khudanpur
“A Measure-Theoretic Characterization of Tight Language Models” by Li Du, Lucas Torroba Hennigen, Tiago Pimentel, Clara Meister, Jason Eisner, and Ryan Cotterell
- “Privacy-Preserving Domain Adaptation of Semantic Parsers” by Fatemehsadat Mireshghallah, Yu Su, Tatsunori Hashimoto, Jason Eisner, and Richard Shin
- “Task-Oriented Dialogue as Dataflow Synthesis” by Jacob Andreas, John Bufe, David Burkett, Charles Chen, Josh Clausman, Jean Crawford, Kate Crim, Jordan DeLoach, Leah Dorner, Jason Eisner, Hao Fang, Alan Guo, David Hall, Kristin Hayes, Kellie Hill, Diana Ho, Wendy Iwaszuk, Smriti Jha, Dan Klein, Jayant Krishnamurthy, Theo Lanman, Percy Liang, Christopher H. Lin, Ilya Lintsbakh, Andy McGovern, Aleksandr Nisnevich, Adam Pauls, Dmitrij Petters, Brent Read, Dan Roth, Subhro Roy, Jesse Rusak, Beth Short, Div Slomin, Benjamin Snyder, Stephon Striplin, Yu Su, Zachary Tellman, Sam Thomson, Andrei Vorobev, Izabela Witoszko, Jason Wolfe, Abby Wray, Yuchen Zhang, and Alexander Zotov
- “Time-and-Space-Efficient Weighted Deduction” by Jason Eisner
- “Toward Interactive Dictation” by Belinda Li, Jason Eisner, Adam Pauls, and Sam Thomson
- “Why Does Zero-Shot Cross-Lingual Generation Fail? An Explanation and a Solution” by Tianjian Li and Kenton Murray, also covered here