Yu Wu · 吴隅
A large collaborative benchmark for evaluating AI agents on long-horizon professional tasks, has been accepted to the Evaluations and Datasets track at NeurIPS 2026. I am glad to contribute to testing agent capability limits in using digital interfaces for historical knowledge work.
Presented Beyond Lexical Reuse: Evaluating Semantic Search for 18th-Century Intellectual History through the Case of Locke in the Digital Humanities Congress 2026 programme.
Joined CASCADE Camp 4 on Questions & Methods, with sessions on career pathways, communicating research to non-specialists, and applying research skills in an industry setting.
The paper combines a historian-built semantic taxonomy with expert annotation to evaluate dense retrieval through the case of Locke's reception.
Participated in the Helsinki event, which addressed digital language sovereignty, low-resource languages, language models, datasets, and research infrastructures.
Presented Detecting Latin in Historical Books with Large Language Models: A Multimodal Benchmark at the 19th Conference of the European Chapter of the Association for Computational Linguistics.
Our multimodal benchmark for extracting Latin fragments from noisy, mixed-language historical books was accepted to the main conference.
Joined the first CASCADE Convention, focused on presenting doctoral research, discussing data and methods, exchanging feedback, and connecting with the MECANO training network.
Presented the ongoing work Detecting Latin in Historical Books with Large Language Models: A Multimodal Benchmark in Norrköping, Sweden.
Presented the ongoing work Detecting Latin in Historical Books with Large Language Models: A Multimodal Benchmark in Espoo, Finland.
Joined CASCADE Camp 3 on the interplay between data, questions, and methods, with colleagues from King's College London, the University of Turku, and the University of Helsinki Computational History Group.
Joined CASCADE Camp 2 on research design, with contributions from Varieng Helsinki and the University of Glasgow.
Joined the first CASCADE Training Camp to meet the doctoral network, learn about each other's research, and discuss the requirements of the programme.