AUTOMATIC TEXT GENERATION IN WIKICREOLE

Project description

This project aims to automatically generate text in WikiCreole syntax using a RNN with TensorFlow. The primary goal is to produce a Wikipedia page for a new French region, inspired by existing pages of French regions. The objectives include preprocessing data by converting WikiCreole text into sequences of characters, training an RNN model to predict the next character in a sequence, and generating a new Wikipedia page in WikiCreole for a fictional region.

The methodology involves several key steps. First, the corpus, consisting of WikiCreole files from the pages of French regions, is prepared. This includes converting characters into identifiers and generating sequences of characters, which are crucial for the model's learning process. The RNN model, which includes Embedding, GRU (Gated Recurrent Unit), and Dense layers, is then trained over 100 epochs to optimize its predictive capabilities.

Once trained, the model is used to generate new text in WikiCreole. Starting with a seed text, the model predicts subsequent characters to create a coherent and syntactically correct text.

Discover more about this project and click on the button below to access the GitHub Repository.

• Find out more on this project

Explore More Projects

If you're interested in exploring more projects, please select another project from the dropdown menu.