About me
My name is Kun Xu. Since 2025, I have been with TikTok as the Head of Content Intelligence and Feed Quality, leading multimodal content understanding algorithms that power TikTok Recommendations. Previously, I was a Manager and Researcher at Kuaishou (Jan 2023–Feb 2025), and a Senior Researcher at Tencent AI Lab in Seattle, USA (September 2018–December 2021). Before that, I was a postdoctoral researcher at the IBM T.J. Watson Research Center in the United States.
I hold a PhD in Natural Language Processing and Machine Learning from Peking University, advised by Professors Yansong Feng and Dongyan Zhao. My research focuses on NLP and multimodal learning, with interests spanning semantic parsing, knowledge-based question answering, information extraction, and natural language interfaces to databases.
You can find me at: Google Scholar.
News
[Dec. 2025] We released our NExT-Vid encoder and our paper and code !
[Nov. 2025] We have one paper accepted by AAAI 2026.
[Oct. 2025] We have one paper accepted by EMNLP 2025.
[Jan. 2025] We have one paper accepted by ICLR 2025.
[Dec. 2024] We have one paper accepted by AAAI 2025.
[Sep. 2024] We have one paper accepted by NeruIPS 2024.
[May. 2024] We have one paper accepted by ACL 2024.
[May. 2024] Our Video-LaVIT is accepted by ICML 2024. [project]
[Feb. 2024] Our paper on Probing MultiModal LLMs is accepted by Coling 2024.
[Feb. 2024] We propose a new text-to-video model, Video-Lavit. [paper][project]
[Jan. 2024] Our Multi-Modal paper LaVIT is accepted by ICLR24.
[Nov. 2023] We released our multi-modal model LaVIT. [paper][code]
[Oct. 2023] We have one paper accepted by TASLP.
[Sep. 2023] My team released the first application of LLM in Micro-video scenario, KwaiTalk. [News]
[May. 2023] We have one paper accepted by PAMI 2023.
[Oct. 2022] We have one paper accepted by EMNLP 2022.
[Apirl. 2022] We have one paper accepted by NAACL 2022.
[Mar. 2022] We have one paper accepted by ACL 2022.
[Sep. 2021] We have four papers accepted by EMNLP 2021.
[Aug. 2021] Our paper on Commonsense Knowledge Graph is accepted by TACL 2021.
[May. 2021] Our papers on Dialogue Understanding and Dialogue Generation are accepted by ACL 2021.
[Apirl. 2021] Our paper on Grammar Induction was accepted by NAACL2021 and won the BEST Long Paper Award !
[Mar. 2021] Our paper on Conversation Semantic Role Labeling is accepted by IEEE Transactions on Audio, Speech and Language Processing.
[Mar. 2021] Our paper on image grounding is accepted by CVPR 2021.
[Jan. 2021] Our paper on conversation Coreference Resolution is accepted by EACL 2021.
[Jan. 2021] The techinical report of our Tencent NLP tool TexSmart is published (https://arxiv.org/pdf/2012.15639.pdf) !
[Oct. 2020] Our team (Kun Xu, Linfeng Song and Liwei Wang) won the first prize in the JD multi-modal dialogue challenge among almost 400 teams and we are invited to give an online presentation in CCL 2020. More details could be found in https://jddc.jd.com/notification/detail/38.
[Sep. 2020] My paper on Dialogue Rewriting got accepted by EMNLP 2020.
[Sep. 2020] Our paper on TTS got accepted by InterSpeech 2020.
[Apirl. 2020] Our two papers got accepted by ACL 2020.
[Nov. 2019] Our two papers got accepted by AAAI 2019.
[Aug. 2019] Our paper got accepted by EMNLP 2019.
[May. 2019] Our three papers got accepted by ACL 2019.
[Apr. 2019] Our two papers got accepted by KDD 2019.
[Feb. 2019] My paper on KB-QA got accepted by NAACL 2019.
[Nov. 2018] Our paper on QA got accepted by AAAI 2019.
[Nov. 2018] Our paper get accepted by NIPS18' workshop on Relational Representation Learning.
[Aug. 2018] Our 3 papers got accepted by EMNLP 2018.
[Dec. 2017] Our papers got accepted to present at LREC 2018.
[Nov. 2016] My paper on a novel hybrid QA model got accepted by COLING 2016
[May 2016] My paper on hybrid QA over Freebase and Wikipedia got accepted by ACL 2016
[May 2016] We won the First Place in the QALD-6 (Question Answering over Linked Data) Evaluation
[Jul. 2015] Our short paper on relation extraction got accepted by EMNLP 2015
[Jul. 2015] We are invited to present our QA system (Best of the CLEF'14 Labs) at CLEF 2015
[Jun. 2015] Our short paper on semantic interpretations for superlative expressions got accepted by ACL 2015
[May. 2015] We won the First Place in the CLEF-QALD5 (Question Answering over Linked Data) Evaluation