How nishantk-iisc/recomsystem_cf works
This repository documents a content-based apparel recommendation project, with title-based text methods and CNN-based image analysis as its main recommendation branches. Its intended workflow runs from product-data acquisition through cleaning and text preprocessing to similarity-based recommendations, with A/B testing for comparison. The tree contains a notebook and data artifacts but no source excerpts, so implementation details and most internal call relationships cannot be verified; the graph uses only relationships described in the README.
Main entry points: Rs_CF.ipynb Text recommendations · 17k_apperal_data Product data · 16k_apperal_data_preprocessed Refined data · images Apparel images
Components 17
Data preparation
- Data acquisition
Rs_CF.ipynb - Product data
pickels/17k_apperal_data - Data cleaning
Rs_CF.ipynb - Title deduplication
dedupe - Refined data
pickels/16k_apperal_data_preprocessed - Text preprocessing
Rs_CF.ipynb
Text recommendations
- Bag of Words
Rs_CF.ipynb - TF-IDF
Rs_CF.ipynb - Average Word2Vec
Rs_CF.ipynb - Weighted Word2Vec
Rs_CF.ipynb - Text recommendations
Rs_CF.ipynb
Image recommendations
- Apparel images
images - CNN image analysis
Rs_CF.ipynb - Image recommendations
Rs_CF.ipynb
Evaluation
- A/B testing
Rs_CF.ipynb
Other components
- Product API
- E-commerce user
Connections 21
- Product API → Data acquisition: provides data
README.md - Data acquisition → Product data: gathers products
README.md - Product data → Data cleaning: is cleaned
README.md - Data cleaning → Title deduplication: removes duplicates
README.md - Data cleaning → Refined data: refines data
README.md - Refined data → Text preprocessing: supplies titles
README.md - Data cleaning → Apparel images: saves images
README.md - Text preprocessing → Bag of Words: prepares text
README.md - Text preprocessing → TF-IDF: prepares text
README.md - Text preprocessing → Average Word2Vec: prepares text
README.md - Text preprocessing → Weighted Word2Vec: prepares text
README.md - Bag of Words → Text recommendations: represents titles
README.md - TF-IDF → Text recommendations: represents titles
README.md - Average Word2Vec → Text recommendations: represents titles
README.md - Weighted Word2Vec → Text recommendations: represents titles
README.md - Apparel images → CNN image analysis: analyzes images
README.md - CNN image analysis → Image recommendations: supports similarity
README.md - Text recommendations → A/B testing: compared by
README.md - Image recommendations → A/B testing: compared by
README.md - E-commerce user → Text recommendations: receives suggestions
- E-commerce user → Image recommendations: receives suggestions