This project aims to split clothing categories into more subtle groups which can capture the inner similarity of clothes and the fasion sensibility of customers, which can serve as product positioning suggestion. Some detailed results and data analysis can be seen in BS_analysis_2.1.pdf
Mainly algorithm steps:
- Matrix factorization to learn products feature representation
- Graph clustering to cluster similar products.
Note
This is a reproduced work of Deciphering Fashion Sensibility Using Community Detection