Unveiling Polarization Trends in Parliamentary Proceedings

This repository contains the code for our research on polarization trends in parliamentary proceedings.

---

Project Resources

Below are the links to the resources we created for this project:

1. Hebrew VAD Lexicon

The Hebrew VAD lexicons are available here:
Hebrew VAD Lexicon

Two versions are provided:

• Version A: Contains the original English words along with their translations and our additional annotations in Hebrew.
• Version B: Contains only the Hebrew words (formatted like the original English lexicon), where the V, A, D value for each word is the average of its values across all occurrences.

---

2. Knesset Corpus Labeled for VAD

The Knesset Corpus labeled for VAD is available here:
VAD Knesset Corpus

This dataset includes:

• Automatic Annotations: Sentences from all Knesset committees labeled with V, A, and D using the automatic models we developed.
• Manual Annotations: A file containing 120 sentences manually annotated with V, A, and D values.

---

3. Knesset-Multi Model

Our knesset-multi model is available here:
Knesset-multi-e5-large

---

4. VAD Regression Models

Our three VAD regression models are available here:
VAD Binomial Regression Models

---

Code Overview

All the code for data processing, model training, and analysis is included in this repository.