A simple Text Predictor which uses RNNs to predict the next word of the input
Developed a Text Predictor Application using deep learning, trained on 5,000 movie titles with a 68% prediction accuracy. Collaborated for optimization, created user-friendly GUI, showcasing NLP and deep learning application in sequence modeling.
Install the following modules: Pandas, tensorflow, tkinter, numpy ,fuzzywuzzy ,Levenshtein (optional), functools and threading using pip install [ModuleName]
Before you finished typing a prediction is to be made (sometimes incorrect), can be alleviated with pressing anything on the keyboardFixed!- Sometimes produces the wrong prediction e.g. 'Spiderman' returns 'Superman', But 'Spider-Man' or 'Spider Ma' works fine
- There is small chance that if you type too quicklky there may not be a prediction produced
- Expand onto a bigger dataset and different dataset when i get a more powerful setup.
- Explore ideas such as pressing Tab to auto complete the suggestion. (The suggestion will be in grey and the user can be press tab to autocomplete)