NEW PAPER ALERT! 🌿🤖 Our essay on how #AI and #NLP can transform conservation evidence synthesis just dropped in #ConservationBiology. We explore how machine learning can help scientists tackle the explosion of research across disciplines to better inform conservation policy. 🧵: 1/6

(Late to the party but ...) #BirdersOfBluesky: who were your First Birds Of the Year (FBOYs)?

Yard/window FBOY: A dapper + tuxedoed Black Phoebe
Binocular FBOY: A very confiding White-breasted Nuthatch
Digiscoped FBOY: An unusually out-and-about Sora

Wishing everyone a very birdy 🦜🦉 2025!

Kali - "Three is technically a starter pack but we're more of a power ✨ trio" 🐕 🐾

#academicdogs

what #starterpack more like a #barkerpack

#dogsofbluesky

The end result? After expert review, we identified 364 study groups. Each was mapped to one or more NCS pathways and their impacts on people/nature. This became the foundation for our evidence map and downstream analyses: a map of what we know (and what we don’t) about NCS pathways & impacts. 🧵 7/8

Finding NCS groups: Density-based clustering revealed groupings among the studies. With help from @nature.org experts, we reviewed these clusters, creating a *manageable* yet *comprehensive* dataset that spanned the input’s breadth. 🧵 6/8

Filtering: We used #BERT embeddings, which reflect patterns & relationships in human language. This grouped studies into meaningful themes. We measured how closely each study aligned with its group. It’s like having a machine-assisted lens to organize millions of abstracts by meaning. 🧵 5/8

Step 1: Automation. We used Python scripts and APIs to query academic databases like Scopus and Web of Science in 2022. Exciting update: Platforms like #OpenAlex now provide free, open access to this kind of data--no subscriptions required. #OpenScience FTW! 🧵 3/8

Hey #AI & #NLP methods nerds--ever wondered how to sift through 2 million scientific studies on natural climate solutions (#ncs, #NbS)? 🌍✨ Here’s how we built a machine learning pipeline 🤖 🐍 to make sense of it all--with humans very much in the loop! 🧪 🌐 🧵 1/8

Congrats on the report. On gaps: we just created a systematic map applying #ML and #NLP to millions of #NCS #NbS studies (rdcu.be/d10Nh). Pathways such as wetlands protection/restoration & some high-mitigation countries (e.g. #COG) lack evidence relative to other ecosystems (e.g. forests) & regions

Our evidence map also shows that landscapes serve multiple functions. Different types of human and environmental well-being outcomes and NCS pathways are often studied together in the same locations. 🧵 5/n

Ultimately, NCS implementation will happen in specific locations. We saw that there are action countries--high mitigation potential and larger bodies of evidence (greener colors), and countries we should prioritize for additional research (more yellow/gold colors). 🧵 4/n

In contrast, some important pathways with high mitigation potential, such as #biochar or #conservationag, lacked evidence entirely or received comparably scanty research. 🧵 3/n

Using transformer language models, we categorized studies to combinations of NCS pathways and impacts to humans and the environment. We found that some pathways (especially forests) had relatively large amounts of evidence 🧵 2/n

The impacts of natural climate solutions: we analyzed millions of scientific studies for a comprehensive analysis of *what* evidence exists for NCS and impacts to people and nature. New work out in @naturesustain.bsky.social doi.org/10.1038/s418...

#ncs #naturebasedsolutions 🧪 #NLP #ConservationAI

Who else thinks Black Skimmer is like a tern x toucan? So nice to see them🪶 #birds #birdphotography

IMHO, #AcademicDogs is the real starter "pack" #Dogsofbluesky

totally agree about Northern Flicker 🦉

Say's Phoebe looking very soft! 🦉

Hey I'm Charlotte!

I use natural language processing to study the impacts of ecosystem-based climate solutions on people and nature.

I post about AI and ML methods for computational sustainability, and also cute photos of my dog