CEAM-A-Model-to-Deal-with-Long-Tail-Problem-in-Recommender-Systems

The items are recommended to customers after using an expenditure appetite filter on top of the collaborative filtering.

It is hard to make recommendations of new items with no rating or low ratings; we call this difficulty as the long tail problem. Also, items recommended to customers are either excessively costly or are of too low cost then what he/she typically spends. The paper proposes a process of resolving the long tail problem. It partitions the whole available items set into head-tail items and clusters of tail items and predicts ratings based on that of the clusters. The items are recommended to customers after using an expenditure appetite filter on top of the collaborative filtering. We show that the proposed model incurs low recommendation error rate while maintaining user preferable items in a recommendation list.

We have done this project in 2 part.

1. Cleaning and preprocessing
2. Top N item recommendation

Database used:

Synthetic Movielence (https://www.kaggle.com/rounakbanik/the-movies-dataset/data)

Software needed:

• Anaconda 3.6+
• Jupyter notebook (included in Anaconda)

Run Instruction:

1. Install anaconda
2. Go to command line/terminal and navigate to the folder
3. Run command : jupyter notebook

Paper :

CEAM: A Model to Deal with Long Tail Problem in Recommender Systems