Certainty-in-Science-Communication

Official Github Repo for EMNLP 2021 paper Measuring Sentence-Level and Aspect-Level (Un)certainty in Science Communications by Jiaxin Pei and David Jurgens.

Data

Annotated scientific findings with sentence-level and aspect-level certainty:

data/annotated_data

The URLs of science news and research papers used in our analysis

data/urls contains 128942 news mentions of 57807 papers. We release URLs for news and DOIs for papers.

Code

Python pacakge for certainty prediction

If pip is installed, certainty-estimator could be installed directly via pip:

pip3 install certainty-estimator

Please check this Github repo for code and examples.

Pre-trained model

We use Hugging Face to host both the sentence-level and aspect-level prediction models.

Code to train the model

If you only need to use the certainty prediction model, please checkout our python package. We use Hugging Face Transformers to train our models. Please email Jiaxin Pei ([email protected]) if you need our code to train the certainty prediction models.

Contact

Jiaxin Pei ([email protected])