/EccentricRecommender

Using a new Metric called Eccentricity this Recommender System aims on improving the variations in modern Recommendation Systems. e.g. recommend more niche items to users

Primary LanguagePython

Eccentric Recommender

This is the source code for an Eccentric Recommender System. This work is mainly based on the theoretical work of, https://arxiv.org/abs/1709.10060

Install

To run this Python 3.x.x is needed. Various packages migth have to be installed,

  • numpy
  • sk-learn
  • contextlib
  • pickle
  • pathlib
  • scipy
  • tqdm

Run

Run file CreateRecommendations.py with python,

python CreateRecommendations.py

Most other files are just pre processing for the data sets. To create this for a new data set, follow the steps in the document. All functions are present the input-files might have to be renamed.

Good to know

On the first run, the system will created the dictionaries, which will speed up computation on the following runs. If someone is changing the system, it is important to delete the related dictonarie. If this is not done strange results might be created.