ISI Directory

Ralph Weischedel

Senior Supervising Computer Scientist, Research Team Leader


Weischedel began his research career at the University of Delaware, where he became a tenured associate professor in the Computer and Information Sciences Department.

On leaving academia, he joined BBN Technologies in Cambridge, MA, where, over the years, he grew a group of roughly 25 researchers, many of whom went on to successful research careers focused on natural language processing (NLP). That group was a leader in the paradigm shift from manually written rules for NLP to statistical learning approaches with applications in information extraction from text and speech, information retrieval, machine translation, and open-domain question answering. Some accomplishments include the first machine learning algorithm for named entity recognition (NER) to achieve scores competitive with manually written rules, and the design and delivery of OntoNotes, a massive, integrated database of manually annotated named entities, parse trees, propositions, coreference, and verb/noun senses, which is available in English, Arabic, and Chinese in multiple genres from the Linguistic Data Consortium.

Weischedel rose to the rank of Principal Scientist at BBN. For a time, he served as head of the 175-person Speech, Language & Multi-Media Department.

Weischedel left BBN in 2017, returning to his academic roots when he joined USC ISI in 2017 in ISI’s Boston Office as a Senior Supervisory Computer Scientist and Research Team Lead.

According to Google Scholar at the end of summer, 2021, Weischedel had over 190 publications and over 11,800 citations of his work.

Research Summary

Though he has prior work in information extraction, information retrieval, and machine translation, his work since joining ISI has focused on forecasting of geopolitical events using machine learning, on open-domain question answering using pre-trained language models (PTLMs), on discovery of typical steps in a complex event, on hypothesizing what could happen next (or what might have happened though not observed) in a complex event, and on text generation for story generation and for dialogue.


Fellow of the Association for Computational Linguistics (ACL)
Past President of the ACL
Co-author of a Best Paper Award from ACL 2008: “A New String-to-Dependency Machine Translation Algorithm with a Target Dependency Language Model”