BERTNN

BERTNN: Affective Meaning Estimation for Everyone

Introduction to BERTNN

BERTNN is a tool developed by Moeen Mostafavi, Michael D. Porter, and Dawn T. Robinson for estimating affective meanings in language. BERTNN harnesses advanced natural language processing techniques to explore the emotional dimensions of words and phrases, making it a valuable resource for researchers, educators, and anyone interested in the intersection of language and emotion.

Abstract

Do you need to find EPA profiles for words that are not present in the existing dictionaries? BERTNN will estimate those for you. This Jupyter-notebook file briefly reviews how the BERTNN model can estimate affective meanings. This is an ongoing project, and many parts may be changed in the final draft. This draft is for personal use. Please get in touch with the authors if you need to share it with your collaborators.

You can access all of the date, code, and technical files on this GitHub site.

No Programming Experience? No Problem!

You don’t need any programming background to use BERTNN. We’ve set up a user-friendly platform on Google Colab, allowing you to experiment with BERTNN without any coding.

You can watch this introductory video about how to run the tool on Google Colab/Cloud.

You can also run BERTNN in the cloud using one of the options below.

Google Colab: Access BERTNN on Google Colab

  • Note: We recommend logging in with your personal Google account, as university or work accounts may not support all features.
  • Google Cloud 1: Access BERTNN on Google Cloud 1
  • Note: Use this token: 675c374fda98208e45b440418f25d5e99b3aae021f91193d
  • Google Cloud 2: Access BERTNN on Google Cloud 2
  • Note: Use this token: df81efe5969a58d43a79a0345aeb8001a2c772bf4802ef29

Both platforms come with pre-installed libraries and environments, making it easy for you to start exploring affective meanings immediately.

Getting Started

  1. Run the Notebook: Simply click “Runtime” on the top menu and select “Run all.”
  2. Explore Functions: Use functions like EPA_sents() and get_output_new() for affective meanings estimation.
  3. Examples and Guidance: The notebook includes examples and detailed explanations.

Dive Deeper

For technical details, check out our Technical.

Authors

  • Moeen Mostafavi [moeen [at] virginia.edu]
  • Michael D. Porter
  • Dawn T. Robinson

Citation

Please cite our work if you use BERTNN in your research:

Mostafavi, M., Porter, M. D., & Robinson, D. T. (2024). Contextual Embeddings in Sociological Research: Expanding the Analysis of Sentiment and Social Dynamics. [To be published in]Sociological Methodology Journal.

Your Journey Begins Here

Discover the fascinating world of affective meanings with BERTNN!

Note: This project is an ongoing endeavor, and methods or data might evolve. Please contact the authors for collaborations or further sharing.