Project Description

Objective: Develop an experimental system for analyzing personality traits by comparing a rule-based approach in Prolog with a machine learning model (XGBoost).

Dataset: Collection and creation of a dataset combining Eye Tracking and Handwriting measurements, labeled via psychometric questionnaires.

Methodology: Data preprocessing (normalization and feature extraction), implementation of a Prolog rule-based system, and training of an XGBoost classifier to evaluate performance.

Evaluation: Quantitative comparison using accuracy, precision, recall, and F1-score, with particular attention to rule interpretability versus ensemble models.

Results: XGBoost achieved higher performance, but the Prolog system proved more interpretable and useful for explaining identified traits.

Skills developed: Prolog, Python, XGBoost, creation of multimodal experimental datasets, statistical analysis, and model interpretability.