Tara Safavi honored with Kuck Dissertation Prize for impactful thesis on advanced graph learning

CSE alum Tara Safavi (PhD CSE 2022) has been recognized with the 2022 Kuck Dissertation Prize for her outstanding doctoral thesis on advanced graph learning. Her dissertation, “Augmenting Structure with Text for Improved Graph Learning,” explores how graphs and text are complementary knowledge representations.

Her award-winning dissertation investigates how pretrained contextual language models can add renewed depth and richness to graph-structured knowledge bases. The thesis additionally aims to improve graph learning tasks that arise in information retrieval and recommender systems by jointly modeling document interactions and content. The proposed methodologies “consistently improve accuracy over both single-modality and cross-modality baselines, suggesting that, with appropriately chosen inductive biases and careful model design, [one] can exploit the unique complementary aspects of structure and text to great effect.”

In her research, Safavi primarily focuses on relational learning and reasoning, knowledge representations, and graph-based learning. Her work “aims to leverage their strengths in order to advance graph learning and explores applications in information retrieval, recommender systems, and natural language processing, three domains in which there are an abundance of relations between entities.” Safavi graduated in 2022 and now is a full-time researcher at Microsoft Research exploring emerging applications and capabilities of large pretrained language models for information retrieval. She was advised by Prof. Danai Koutra.

She previously received the Rackham Predoctoral Fellowship, the Rackham Dean’s and Named Fellowship, an Outstanding Research Award, and a Marian Sarah Parker Prize from U-M. In 2017, she received a Women Techmakers Scholarship from Google and a Best Paper nomination from IEEE ICDM. She also received an NSF Graduate Research Fellowship in 2018. In 2019, she earned the Best Student Paper Award from IEEE ICDM, then won the Outstanding Reviewer Award from EMNLP in 2020.