# Evaluating Automatic Personality Recognition (APR) Approaches: 
The Role of Deep Learning in Enhancing Recommender System Performance

## Datasets

1. **PANDORA Dataset**  
   - Source: https://psy.takelab.fer.hr/datasets/all/pandora/  
   - Access: Not publicly available

2. **Amazon Reviews 2023**  
   - Source: https://amazon-reviews-2023.github.io/  
   - Category: Movies & TV  

## Files:

Data Preprocessing Amazon.ipynb
- Preprocessing Amazon Dataset for BERT and LIWC
- Running LIWC on Amazon Dataset
- Running LIWC on Reddit Dataset

Data Preprocessing Reddit.ipynb
- Preprocessing and examining Reddit Dataset

BERT APR model-Classification.ipynb
- Fine-tuning BERT classification model
- Contains instructions for converting between each trait and between 10-class and binary
- Also contains comments on converting to regression

BERT APR Amazon Data.ipynb
- Running trained BERT models on Amazon data

Baseline Recommender.ipynb
- Recommender for Amazon data with no personality included

Personality_Recommender_LIWC_Classification.ipynb
- Recommender for Amazon data with LIWC personality
- Contains instructions for both 10-class and binary


Personality_Recommender_BERT_Classification.ipynb
- Recommender for Amazon data with BERT personality
- Contains instructions for both 10-class and binary


The BERT model weights themselves were too large to include but can be provided if requested