🌟Paper accepted! NeoQA: A benchmark for evidence-based question answering with generated news events. Unlike traditional datasets, NeoQA constructs fictional timelines, ensuring LLMs must reason exclusively over retrieved evidence. 🔍 📃 arxiv.org/abs/2505.05949 #ACL2025 #NLProc

Don't miss the CCSum presentation at #NAACL2024. We present a large-scale dataset (1.3M) for abstractive news summarization. More factual and informative than summaries in CNN/DM, XSum & Multi-News! Joint work w/ Xiang Jiang 💻 github.com/amazon-science/cc… 📃 aclanthology.org/2024.naacl-…

We define an abstractiveness metric called MINT, which measures abstractiveness as a percentage score. It incorporates contiguous and non-contiguous overlaps of the summary with the input. Implementation on github.com/amazon-science/ab…

We analyze the factuality of abstractive summaries -- at different levels of abstractiveness. We do this across multiple datasets and models. We visualize how factuality changes as we gradually incrementally abstractiveness.

#ACL2019 in Florence Italy

Next NAACL in Minneapolis. #NAACL2018 #NAACL2019

NAACL best paper award #NAACL2018

Mirella Lapata: Structure is coming back! A good summary of current trends. #acl2017nlp

A fun little NLP problem for Apple to solve.

Amazon Alexa Science Internships 2016 #nlproc #machinelearning linkedin.com/pulse/amazon-al…

You can now extend Amazon Echo and implement your own voice commands zdnet.com/article/amazon-thr… #amazonecho #alexa

Meme generation: "Such data, so big. Very text. Much helpful. Wow." #naacl2015

How to pronounce Wurstküche? youtube.com/watch?v=K5p_L0mb… @wurstkuche #language #German #GER #Germany

Fishy fun vine.co/v/blhh5qgFYaQ