About Thinkwee

About Me

  • Acquired Bachelor of Communication Engineering at BUPT in 2018
  • Acquired Master of Computer Engineering at CIST@BUPT(Center of Intelligence Science and Technology) in 2021
  • Now works at Tencent as Application Researcher, focusing on NLP for recommendation, advertising.


  • CIST(Student)
    • Tasks
      • Text Classification
      • Summarization
      • Sequence Tagging
    • Methods
      • Seq2seq
      • Topic Modeling
      • Determinantal Point Process
      • Graph Neural Networks
      • Pretrained Language Models
  • Sina Weibo(Intern)
    • Text Classification(Bert Finetune & Data Augment)
    • Document Keywords Scoring(DUET improvement)
  • Tencent(Intern)
    • KeyPhrase Ranking and Generation
    • MRC
    • Text Generation
  • Tencent MLPD(Application Researcher)
    • Summarization
    • Graph Embedding
    • CB Recall, improve the diversity and Real-time recommendation sense
    • User Profile Modeling, Recall&Prerank Model Improvements
  • (NEW!) Co-authored paper Fast and Constrained Keyphrase Generation get accepted on AAAI 2022


  • My Google Scholar
    • [paper] (AAAI 2022 Long Paper) Fast and Constrained Absent Keyphrase Generation by Prompt-Based Learning
    • [code][paper] (ACL 2021 Findings Long Paper) UniKeyphrase: A Unified Extraction and Generation Framework for Keyphrase Prediction
    • [paper] (EMNLP 2020 Workshop) CIST@CL-SciSumm 2020, LongSumm 2020: Automatic Scientific Document Summarization
    • [code][paper] (CoNLL 2019 Long Paper) In Conclusion Not Repetition: Comprehensive Abstractive Summarization with Diversified Attention Based on Determinantal Point Processes
    • [paper] (RANLP 2019 workshop) Multi-lingual Wikipedia Summarization and Title Generation On Low Resource Corpus
    • [paper] (SIGIR 2019 workshop) CIST@CLSciSumm-19: Automatic Scientific Paper Summarization with Citances and Facets
    • [code][paper] (Arxiv Preprint) Subjective Bias in Abstractive Summarization
    • [paper] (Arxiv Preprint) CO2Sum:Contrastive Learning for Factual-Consistent Abstractive Summarization
    • [paper] (CCIS 2018 Long Paper) A Multi-View Abstractive Summarization Model Jointly Considering Semantics and Sentiment