Tackling Climate Change with Machine Learning NeurIPS 2020 Virtual - PowerPoint PPT Presentation

Tackling Climate Change with Machine Learning NeurIPS 2020 Virtual Workshop December 11-12, 2020

Tackling Climate Change with Machine Learning Why topic detection for climate change ? Sentiment analysis Output Question-answering Documents Fact-checking 2

Tackling Climate Change with Machine Learning Not as easy as it seems… • Some examples:

Tackling Climate Change with Machine Learning CLIMATEXT 4

Tackling Climate Change with Machine Learning Wikipedia data (Wiki-Doc-Train/Dev/Test) 1) Land surfaces are Negatives heating faster than the ocean surface, leading to heatwaves, wildfires, and the expansion of deserts. Weakly .csv Documents 2) As the temperature labeling difference between the Arctic and the equator decreases, ocean Positives currents that are driven by that temperature difference, like the Gulf Wikipedia Stream. 3) …. Links Graph 5

Tackling Climate Change with Machine Learning AL-Wiki and AL-10Ks train data NB model AL-WIKI AL - WIKI Wikipedia 10-Ks 10-Ks AL - 10-Ks (unlabeled) NB model AL-10-Ks 6

Tackling Climate Change with Machine Learning Wikipedia, 10K, and Claims evaluation data Document label sampling scheme (labeled by raters) BERT-predictions sampling scheme (labeled by raters) Websites of claim collections and other (labeled by raters) 7

Tackling Climate Change with Machine Learning Main Result Our contributions: We introduce CLIMATEXT, a dataset for sentence-based climate change topic 1. detection, which we make publicly available. We analyze keyword-based, naïve-Bayesian, and a BERT-based approach to explore 2. their performance in identifying climate-change relevant text in Wikipedia, 10-K filings, and in climate-related claims database. Going forward: Make the annotated data public. - Improve algorithms to detect climate-change topic in a wide range of text sources. - 8