[TBD]Understanding item based collaborative approach for recommendation

Introduction

Laymen explanation

If you are a shop-owner, you might wanted to recommend your customer about products which they can try out. In this way, you will get more business and customer will get happiness only if recommendation is good for him/her (otherwise face risk of loosing regular customer). So, this is a tempting but risky. Moreover, most of the time, you might not have used the product. What you will do? Simply avoid recommendation? Not really. You can observe feedback of other users who is purchasing same product from your shop. In this way, you will get confident about the product quality and user experience.

Now next question arises- what kind of product this customer will like so that you can recommend. For this, you have some knowledge of this customer which includes his lifestyle, tastes. For example, if customer has small kids, then you can recommend items which are purchased by other customers who has small kids. So, you are going to recommend what similar customers purchases. Need a care here. Similar customer might be purchasing items for himself or friends or his kids (GOD knows!!). Now, if you recommend something which is not applicable for small kids, it will be not be rightly taken by customer. what to do? You need to find similarities on items. You can do it by observing what similar customers are purchasing is relevant to what. If you do this, you will be able to filter kids related items purchased by similar customers. This article helps in this approach.

Technical explanation

Reference

https://en.wikipedia.org/wiki/Item-item_collaborative_filtering

https://engineering.linkedin.com/recommender-systems/browsemap-collaborative-filtering-linkedin

http://www.cs.carleton.edu/cs_comps/0607/recommend/recommender/itembased.html

https://ashokharnal.wordpress.com/2014/12/18/worked-out-example-item-based-collaborative-filtering-for-recommenmder-engine/