2019B - Final Project in Introuction to NLP @OpenU

This is the work of @ira-vitenzon(אירה ויטנזון) and @shaharlinial (שחר ליניאל)

This project aims to create a sequence-to-sequence model that can detect metaphors in the English language, and potentially in other languages as well, given a large dataset of metaphors.

The model is implemented using deep learning techniques, specifically Long Short-Term Memory (LSTM) networks, and is built using the PyTorch deep learning library. The ability to detect metaphors is a crucial task in natural language processing, as it can be used to improve the accuracy and relevance of a wide range of applications, such as sentiment analysis, text summarization, and language translation.

The model will be trained on a large dataset of annotated metaphors and will be evaluated using standard evaluation metrics This project is made with PyTorch.

Installation Notes:

Automatic Installation (Ubuntu Only):

1. Clone this repository with git clone
2. Get into repo directory cd deep-meatphor-detection
3. Run sudo bash installation.sh
4. Run python3 run.py

Manual Installation:

1. Clone or Download this repository.
2. Download Glove by NLP Stanford Common Crawl
3. Create a directory named glove inside the repository directory
4. Extract glove.840B.300d.zip to glove.840B.300d.txt
5. Rename glove.840B.300d.txt into glove840B300d.txt
6. Run python3 run.py

deep-metaphor-detection
│   README.md
│   model.py
|   run.py    
|   requirements.txt
|   util.py
|   seq2seq_with_attention_model.py
│
└───glove
│   │   glove840B300d.txt