NLP Assignments

This repository contains two university assignments focused on Natural Language Processing (NLP) tasks. Both assignments involve different models and methodologies for tackling specific NLP problems. The assignments are designed to provide insights into the capabilities of NLP models in solving various tasks.

Assignment 1: POS Tagging

The first assignment involves Part-of-Speech (POS) tagging using three different models:

• Baseline Model: Input layer, LSTM layer, Dense layer.

• Extended LSTM Model: Extends the baseline model by adding an extra LSTM layer.

• Extended Dense Model: Extends the baseline model by adding an additional Dense layer.

Assignment 2: Multi-Label Classification

The second assignment focuses on a multi-label classification problem to identify human values behind arguments using NLP models with BERT as the pre-trained model. Different configurations of textual inputs are concatenated for the experiments:

• Model 1: Only the conclusion feature as input

• Model 2: Stance and conclusion as textual input features.

• Model 3: Stance, conclusion, and premise as textual input features.

Experiments and Results

The experiments conducted in both assignments provided valuable insights into the ability of NLP models to solve these types of tasks. The results demonstrated how different model architectures and input configurations can impact the performance in POS tagging and multi-label classification tasks.

Contribution

Contributions are welcome! To contribute:

Fork the repository. Create a branch for your modifications. Make a pull request.

Authors

• Fabio Zanotti (fabio.zanotti@studio.unibo.it)
• Antonio Morelli (antonio.morelli@studio.unibo.it)
• Edoardo Conca (edoardo.conca@studio.unibo.it)
• Federico Rullo (federico.rullo@studio.unibo.it)