Contributors
- Sarah Alqaysi
- Hanmaro Song
- Sean Torres
- ANalyze and Preprocess IMDB data as well as Netflix
- Use pre-trained Bert model from Hugging Face to analyze the semantics of synopsis for each title
- Given an input, recommend a list of titles that are similar to the gicen based on genres and plots
In the Modeling.ipynb, do the following.
get_recommendation('Fast & Furious', top_k=top_k, use_genre=False)
which will output list of recommendation like
sorted by similarity score in descending order.
if use_genre is set to True, then the model will use both the plot and genre to compute the similarity score.
There are total of 768 semantics columns from HuggingFace Bert model ('bert-base-uncased')