Research Associate @ JHU HLTCOE
As a Research Associate at the Human Language Technology Center of Excellence at Johns Hopkins University, I am supervised by Nicholas Andrews and collaborate with Elizabeth Allyn Smith (Université du Québec `a Montréal). Our recent work created a new benchmark and established the state of the art for speaker attribution by applying authorship attribution methods to conversational speech transcripts (paper, code).
I am specifically interested in robust and explainable attribution models that work well in the challenging forensic linguistics setting, which involves minimal and often noisy data and requires higher standards for accuracy given the high-stakes consequences.
I received my Ph.D. in Linguistics from Harvard University, advised by Stuart Shieber. My dissertation created the first analysis of reciprocals in the formalism of synchronous tree adjoining grammar (STAG), unified this analysis with a parallel one of reflexives, and extended the formalism for the first time to encompass morphology.
I did an M.S. in Cognitive and Decision Sciences at University College London, advised by David Lagnado. My master’s dissertation combined script theory and causal Bayesian networks to create a formalized crime script for cash-in-transit robbery that could ultimately be used by the police to identify effective prevention strategies.
I also hold a B.A. in Applied and Computational Mathematics and a B.A. in Linguistics from the University of Southern California. Fight on!
Interested in a little bit of everything, I love traveling the world by sailboat/train/plane/bus/motorcycle/car, connecting with people, learning new skills, and working out. One of my favorite trips was sailing across the Pacific Ocean on a 34-foot sailboat with two other women (S/V Islander).
Computational Linguistics
Can authorship attribution models distinguish speakers in speech transcripts?
Cristina Aggazzotti, Nicholas Andrews, and Elizabeth Allyn Smith
Transactions of the Association for Computational Linguistics 12, pp. 875–891 (2024) [code]. Oral presentation at ACL 2024.
Can authorship representation learning capture stylistic features?
Andrew Wang*, Cristina Aggazzotti*, Rebecca Kotula, Rafael Rivera Soto, Marcus Bishop, and Nicholas Andrews
Transactions of the Association for Computational Linguistics 11, pp. 1416–1431 (2023). Oral presentation at EMNLP 2023.
Reflexives and reciprocals in synchronous tree adjoining grammar
Cristina Aggazzotti and Stuart M. Shieber
Proceedings of the 13th International Workshop on Tree Adjoining Grammars and Related Formalisms, Association for Computational Linguistics, pp. 31–42 (2017).
A unified analysis of reflexives and reciprocals in synchronous tree adjoining grammar
Cristina Aggazzotti
Doctoral dissertation, Harvard University (2019).
Forensic Linguistics
Testing authorship analysis on spoken language transcripts: Establishing a first benchmark
Cristina Aggazzotti, Nicholas Andrews, and Elizabeth Allyn Smith
4th Focus Conference on Forensic Linguistics of the International Language and Law Association (ILLA) (2023).
Authorship attribution across right-wing manifestos
Cristina Aggazzotti and Elizabeth Allyn Smith
16th Biennial Conference of the International Association for Forensic and Legal Linguistics (IAFLL) (2023).
Disambiguating ‘ground truth’ for forensic applications in artificial intelligence
Cristina Aggazzotti and Elizabeth Allyn Smith
Proceedings of the American Academy of Forensic Sciences Annual Conference, vol. 29, p. 380 (2023).
A comparison of two computational approaches to analyzing last words
Cristina Aggazzotti
Proceedings of the American Academy of Forensic Sciences Annual Conference, vol. 28, p. 483 (2022).
A quantitative comparison of last words to related text genres
Cristina Aggazzotti
Proceedings of the American Academy of Forensic Sciences Annual Conference, vol. 28, p. 486 (2022).
A quantitative assessment of last words using Suicide Note Assessment REsearch (SNARE)
Cristina Aggazzotti
Proceedings of the American Academy of Forensic Sciences Annual Conference, vol. 27, p. 370 (2021).
Communication strategies to minimize bias and strengthen scientific foundations in forensic science
Cristina Aggazzotti, Katherine Ramsland, Carole E. Chaski, Elizabeth Allyn Smith, and Roderick T. Kennedy
Proceedings of the American Academy of Forensic Sciences Annual Conference, vol. 23, p. 38 (2017).
Confirmation bias and metalinguistic awareness
Carole E. Chaski, Elizabeth Allyn Smith, and Cristina Aggazzotti
Proceedings of the American Academy of Forensic Sciences Annual Conference, vol. 22, p. 902 (2016).
Cognitive Science
The use of causal Bayesian networks to formalize crime scripts, with an application to cash-in-transit robbery
Cristina Aggazzotti
Master’s dissertation, University College London (2013).