Recommender-System
This is highly involved project on recommendation system with classical and advance method.
Project timeline
- At Ist stage, we will discuss basics of recommender system and run algorithms on Movie Lens Dataset. Time period of this notebook will be
5
days - At second stage we will extend the same idea to text based recommendation system where we will explore feature extraction for text such as
tf-idf, word-embedding, sent2vec
etc. This will be a heavy deep learning based system. It will be of7
days time period. - At third stage, we will explore some of the practical aspect of recommender system such as evaluation metrics, cross-validation etc. Then we will make a hybrid system. This type of models are used in real world application such as netflix, instagram, facebook, amazon etc. It won't be of the same level , but it will provide their essence at brief. It will take around
2
weeks. - At final stage, we will explore
graph neural network
for this domain. We will run experiments on building a good and advance features for recommender system. Time period for this stage will be decided later.