Publications

  • Bjarke Felbo, Alan Mislove, Anders Søgaard, Iyad Rahwan and Sune Lehmann. Using millions of emoji occurrences to learn any-domain representations for detecting emotion, sentiment and sarcasm. Empirical Methods in Natural Language Processing (EMNLP) 2017 long paper (oral).
    [PDF] [Code] [Model] [Demo]
     
  • Andres Campero, Bjarke Felbo, Joshua B. Tenenbaum and Rebecca Saxe. A First Step in Combining Cognitive Event Features and Natural Language Representations to Predict Emotions. Cognitive Computational Neuroscience (CCN) 2017 short paper.
    [PDF - will be added soon]
     
  • Bjarke Felbo, Pål Sundsøy, Alex 'Sandy' Pentland, Sune Lehmann and Yves-Alexandre de Montjoye. Modeling the Temporal Nature of Human Behavior for Demographics Prediction.  European Conference on Machine Learning (ECML-PKDD) 2017 long paper (oral).
    [PDF - will be added soon] [Code] [Model]

 

Non-archival Publications

  • Nick Obradovich, Bjarke Felbo, Pinar Yanardag, Adrian Borsa, Jennifer Anne Burney, Jason J. Jones and Iyad Rahwan. The effects of climatic events on expressed climate change belief. Annual Meeting of American Political Science Association (ASPA) 2017.
     
  • Bjarke Felbo, Alan Mislove, Anders Søgaard, Iyad Rahwan and Sune Lehmann. Using millions of emoji occurrences to learn any-domain representations for detecting emotion, sentiment and sarcasm. Association for Computational Linguistics (ACL) 2017 Repl4NLP workshop.
     
  • Richard Kim*, Bjarke Felbo*, Nick Obradovich and Iyad Rahwan. Quantifying structural factors affecting racial tensions during 2016 US presidential election. International Conference on Computational Social Science (IC2S2) 2017.
     
  • Bjarke Felbo, Pål Sundsøy, Alex 'Sandy' Pentland, Sune Lehmann and Yves-Alexandre de Montjoye. Using Deep Learning to Predict Demographics from Mobile Phone Metadata.  International Conference on Representation Learning (ICLR) 2016 workshop.

     

Code

  • DeepMoji, Python library for transfer learning wrt. emotion-related NLP tasks. Author.
    [Code] [Website]
     
  • bandicoot, a Python library for mobile phone metadata analysis. Part of initial development.
    [Code] [Website]