Dave in 🏴You might have heard me claim that most #NLG is #LowResource (not just #NaturalLanguageGeneration for #LowResourceLanguages). If you want to hear me explain a bit more, my talk from last year's #GEM workshop at #EMNLP2022 is now up online: https://underline.io/lecture/66771-most-nlg-is-low-resource-here-s-what-we-can-do-about-it
Daniel WhitenackGreat to talk with @JayAlammar at #EMNLP2022, and I'm honored that he featured my team's research in his latest video! Check it out: https://youtu.be/plCvF_7qrmY
EMNLP 2022 Conference Special Edition - Talking Language AI #4
Dave 🦔Quite enjoyed Mona Diab's (https://twitter.com/MonaDiab77/) #EMNLP2022 keynote today.
Lots of interesting work on #ResponsibleAI and #ResponsibleNLP. (Even if a lot of it is coming from Meta/Facebook 😬)
One of the earlier mentions was TCT (https://github.com/facebookresearch/text_characterization_toolkit) which re-implements a lot of #CohMetrix text statistics to help researchers characterise the corpora they're training their models on
What a great way to kick new 2023 episodes of @practicalai! This week we are sharing a conversation I had at #EMNLP2022 with Just Zwennicker, Andiswa Bukula, Rooweither Mabuya & Bonaventure Dossou
Amazing researchers building SOTA #AI #NLProc for their communities. Listen here: https://changelog.com/practicalai/205
NLP research by & for local communities with Just Zwennicker, Andiswa Bukula, Rooweither Mabuya & Bonaventure Dossou (Practical AI #205)
While at EMNLP 2022, Daniel got a chance to sit down with an amazing group of researchers creating NLP technology that actually works for their local language communities. Just Zwennicker (Universiteit van Amsterdam) discusses his work on a machine translation system for Sranan Tongo, a creole language that is spoken i...
Practical AI 🤖🦾 New episode of Practical AI! 🦾
While at #EMNLP2022, @dwhitena got a chance to sit down with an amazing group of researchers creating #NLP technology that actually works for their local language communities.
The group emphasized the need for more linguistically diverse NLP systems that work in scenarios of data scarcity, non-Latin scripts, rich morphology, etc. You don’t want to miss this one!
NLP research by & for local communities with Just Zwennicker, Andiswa Bukula, Rooweither Mabuya & Bonaventure Dossou (Practical AI #205)
While at EMNLP 2022, Daniel got a chance to sit down with an amazing group of researchers creating NLP technology that actually works for their local language communities. Just Zwennicker (Universiteit van Amsterdam) discusses his work on a machine translation system for Sranan Tongo, a creole language that is spoken i...
Gerard de MeloPaper Title:
Curriculum Prompt Learning with Self-Training for Abstractive Dialogue Summarization
Authors:
Changqun Li, Linlin Wang, Xin Lin, Gerard de Melo, Liang He
This is joint work with researchers from East China Normal University in Shanghai, including my former PhD student Linlin Wang, who is now a faculty member there, and, most importantly, her student collaborator Changqun Li.
Check out the #EMNLP2022 Paper for further details:
http://gerard.demelo.org/papers/dialogue-summarization.pdf
http://gerard.demelo.org/papers/dialogue-summarization.bib
2/2
Overwhelming amounts of dialogue are being recorded in social media, instant messaging apps, Slack channels, customer service interactions, and so on.
Wouldn't it be great if automated tools could provide us short, succinct summaries?
Here's talk I just gave at #EMNLP2022 about a new #LLM prompt learning method that works fairly well even when the training data is extremely limited.
1/2
Curriculum Prompt Learning with Self-Training for Abstractive Dialogue Summarization
Overwhelming amounts of dialogue are being recorded in instant messaging apps, Slack channels, customer service interactions, and so on. Wouldn't it be great if automated tools could provide us short, succinct summaries? However, challenges such as insufficient training data and low information density impede our ability to train abstractive summarization models. In this work, we propose a novel curriculum-based prompt learning method with self-training to address these problems. Specifically, prompts are learned using a curriculum learning strategy that gradually increases the degree of prompt perturbation, thereby improving the dialogue understanding and modeling capabilities of our model. Unlabeled dialogue is incorporated by means of self-training so as to reduce the dependency on labeled data. We further investigate topic-aware prompts to better plan for the generation of summaries. Experiments confirm that our model substantially outperforms strong baselines and achieves new state-of-the-art results on the AMI and ICSI datasets. Published at EMNLP 2022 Paper: http://gerard.demelo.org/papers/dialogue-summarization.pdf
David BauHad a great time meeting people at #BlackboxNLP and #EMNLP2022 last week. You can see my talk as well as great talks by Lena Voita and Catherine Olsson on youtube here:
StanfordNLPWe had lots of papers at #emnlp2022 … and lots of us were there in Abu Dhabi in person to present them!
The Data Therapist@thelonious @trochee here, let me toot you the antidote: the #BenderRule winner of #emnlp2022 : Søgaard — chapeau!
https://preview.aclanthology.org/emnlp-22-ingestion/2022.emnlp-main.351.pdf