PeopleofSuperintelligence

Researcher

Shay Cohen

Also known as Shay B. Cohen

Shay Cohen is a researcher in artificial intelligence and natural language processing at the University of Edinburgh, with work on parsing, summarization and language generation. After a PhD at Carnegie Mellon University in 2011 under Noah Smith, he spent 2011 to 2013 as a Computing Innovation Fellow at Columbia University and was a Chancellor's Fellow at Edinburgh by 2014. His papers include a 2014 study of spectral learning for latent-variable PCFGs, a 2017 study with Lea Frermann and Mirella Lapata on guessing the culprit in CSI episodes, and the 2018 extreme summarization paper with Shashi Narayan and Mirella Lapata. A second edition of his book Bayesian Analysis in Natural Language Processing came out in 2019 with a new chapter on representation learning and neural networks. He is a Reader in the School of Informatics and a Fellow of ELLIS, and his work draws on large language models and includes collaborations with biologists.

Works
251
Citations
4,865
h-index
28

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2000–2011

    PhD at Carnegie Mellon

    He earns a BSc in 2000 and an MSc in 2004 at Tel Aviv University. In 2011, he completes a PhD at Carnegie Mellon University under Noah Smith, with a dissertation on computational learning of probabilistic grammars in the unsupervised setting.

  2. 2011–2013

    Postdoc at Columbia

    From 2011 to 2013, he is a Computing Innovation Fellow at Columbia University, where he works with Michael Collins on spectral learning algorithms for latent-variable models of language.

  3. By 2014

    Chancellor's Fellow at Edinburgh

    After the Columbia fellowship, he moves to the University of Edinburgh, and by January 2014 he is a Chancellor's Fellow in the School of Informatics.

  4. 2014

    Spectral learning for L-PCFGs

    In 2014, he coauthors "Spectral Learning of Latent-Variable PCFGs: Algorithms and Sample Complexity" with Karl Stratos, Michael Collins, Dean P. Foster and Lyle Ungar. The paper introduces a spectral learning algorithm for latent-variable PCFGs and proves statistically consistent parameter estimates under a separability condition.

  5. 2017

    Predicting culprits in CSI

    In 2017, he joins Lea Frermann and Mirella Lapata in turning the scripts of 39 CSI: Crime Scene Investigation episodes into inputs for an LSTM model, in the paper "Whodunnit? Crime Drama as a Case for Natural Language Understanding".

  6. 2018

    The extreme summarization task

    In 2018, he coauthors "Don't Give Me the Details, Just the Summary!" with Shashi Narayan and Mirella Lapata at EMNLP. The paper introduces extreme summarization, the task of writing a one-sentence news summary, builds a dataset from BBC articles, and proposes a topic-aware convolutional neural network.

  7. 2019

    Bayesian Analysis in NLP, second edition

    In 2019, Morgan & Claypool publishes the second edition of his book Bayesian Analysis in Natural Language Processing, in the Synthesis Lectures on Human Language Technologies series. The edition adds a chapter on representation learning and neural networks in the Bayesian context.

  8. As of October 2026

    Reader at Edinburgh, ELLIS Fellow

    He is a Reader in the School of Informatics at the University of Edinburgh and a Fellow of ELLIS in its Edinburgh unit, working on artificial intelligence and natural language processing. His work draws on large language models and includes collaborations with biologists, and in Spring 2026 he teaches Advanced Topics in Natural Language Processing.

Career

  1. Since 2011University of EdinburghChancellor's Fellow
  2. 2018–2021Israel Institute for Biological Research
  3. 2011–2013Columbia UniversityComputing Innovation Fellow
  4. 2005–2007Tel Aviv University

Education

  1. 2011Carnegie Mellon UniversityPhD
  2. 2004Tel Aviv UniversityMSc
  3. 2000Tel Aviv UniversityBSc

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Don’t Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization2018 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 1,202 citationsWith Mirella Lapata
  2. Stock Movement Prediction from Tweets and Historical Prices2018 · Annual Meeting of the Association for Computational Linguistics (ACL) · 428 citations
  3. The 2017 Conference on Empirical Methods on Natural Language Processing (EMNLP)2015 · 276 citationsWith Claire Gardent
  4. An Incremental Parser for Abstract Meaning Representation2017 · 156 citations
  5. Feature Selection via Coalitional Game Theory2007 · Neural Computation · 144 citations

From OpenAlex, as of 21 October 2026.

  1. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes University of Edinburgh. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  2. Universities had a net outflow of 2,068 moves in AI careersIncludes University of Edinburgh. Our estimates of career flows and median h-index across universities and companies in AI, including the role of joint appointments and earlier academic jobs.
  3. People shaping AI work at a median of 3 organizations over a career: 3 for university starters and institute and lab starters, versus 2 for company startersIncludes University of Edinburgh. Across 6,853 people shaping AI with a dated work role, we estimate a median of 3 parent organizations over a career (mean 3.6), comparing role groups, starting sectors and starting decades from the 1950s to 2020–2026.
  4. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes University of Edinburgh. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Shay Cohen yet. Send us a free-licensed photo.

  1. Shay Cohen - University of Edinburgh Research Explorer research.ed.ac.uk. Checked 20 October 2026.
  2. Shay Cohen - The Mathematics Genealogy Project genealogy.math.ndsu.nodak.edu. Checked 20 October 2026.
  3. My Experiences as a CIFellow - CCC Blog cccblog.org. Checked 20 October 2026.
  4. Spectral Learning of Latent-Variable PCFGs: Algorithms and Sample Complexity jmlr.org. Checked 20 October 2026.
  5. 39 episodes of 'CSI' used to build AI's natural language model theregister.co.uk. Checked 20 October 2026.
  6. Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization - ACL Anthology aclanthology.org. Checked 20 October 2026.
  7. LINGUIST List 30.1843: Books: Bayesian Analysis in Natural Language Processing: Cohen linguistlist.org. Checked 20 October 2026.
  8. Shay Cohen's homepage homepages.inf.ed.ac.uk. Checked 20 October 2026.
  9. Shay Cohen | School of Informatics inf.ed.ac.uk. Checked 20 October 2026.
  10. Shay Cohen | European Laboratory for Learning and Intelligent Systems ellis.eu. Checked 20 October 2026.