You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This repository presents and compares HeterSUMGraph and variants doing extractive summarization, named entity recognition or both. HeterSUMGraph and variants use GATv2Conv (from torch_geometric).
Ce fut mon prémier projet NLP où j'ai réalisé la détection de spam en utilisant les algorithmes d'embedding pour encorder mes textes. J'ai utilisé Random Forest et Milti-Layres Perceptrons pour la phase de classification. Ce qui a pemit l'obtension des précisions respective de 97% et 98%. J'ai aussi appris à documenter mes codes via sphinx
The project focuses on developing medical word embeddings using Word2vec and FastText in Python to create a search engine and Streamlit UI. The use of embeddings helps overcome the challenges of extracting context from text data, making it easier to represent words as semantically meaningful dense vectors.
Classification of acronyms and their long forms using an RNN (LSTM), CNN, and FFNN model. The experiments focused on the RNN and used different vectorisation methods and hyperparameters. Models were built with Keras and the notebook code runs on Google Colab.