Team Halloween - Reference Recommender System

This was the submission for Global Data Science Challenge IV by Team Halloween

Member:

Arka Prava Bandyopadhyay

Motto:

"Simplicity is the ultimate sophistication." - Leonardo da Vinci

Problem :

From a set of ppts, recommend top 10 relevant ppts for a given query.

Solution:

The core of the model was TF-IDF and cosine similarity and other engineered features used were:

• Sector
• Country
• Technology
• Year
• Name of the ppt

Based of these features my model computed a score for each ppt and sorted the ppts based on that score for a given query.

Yammer Group:

https://www.yammer.com/capgemini.com/#/threads/inGroup?type=in_group&feedId=13438430&view=all

Winner Announcement :

Capgemini Internal Link: https://builders.capgemini.com/readcommunicationonline?content_id=A5395BD6-8CAB-8480-212A-FD5C25AE1512