Algorithm for “recommend items” for a user who has some preferences tracked [closed]

Question

Closed. This question needs to be more focused. It is not currently accepting answers.

---

Want to improve this question? Update the question so it focuses on one problem only by editing this post.

Closed 4 years ago.

Improve this question

Are there any well known algorithms for figuring out the "Recommended Items" that websites often use? On a new project, I'm currently tracking things they've viewed, purchased, searched on, favourited, etc. I am curious what the best way would be to utilize all of this information to intelligently give meaningful recommendations.

Answer 1

I recommend you two books:

1. Programming Collective Intelligence

   Toby Segaran (ISBN: 978-0-596-52932-1)
   http://shop.oreilly.com/product/9780596529321.do

   　In this book, "Chapter 2. Making Recommendations" covers:

   • Collaborative Filtering
   • Collecting Preferences
   • Finding Similar Users
   • Recommending Items etc..
     
     

2. Collective Intelligence in Action

   Satnam Alag (ISBN: 1933988312)
   http://www.manning.com/alag/

   In this book, "Capter 12. Building a recommendation engine" covers:

   • Recommendation engine fundamentals
   • Content-based analysis
   • Collaborative filtering
     
     

I hope it helps you.

Answer 2

I thing easyrec would be a good starting point for you to track such user interactions.YOu just need to include some javascript code to your page and it does the rest.

http://easyrec.org/recommendation-engine

You can also check the other recommendation engines listed here https://stackoverflow.com/questions/4469281/recommendation-engine# but problem is most of these frameworks target user rating predicting over collaborative filtering.

Answer 3

1. There are mechanisms like 'people who bought this items also bought'.
2. Category tops: if someone is viewing 'computers/storage devices', show the top selling products in that category.
3. Previous searches: use the (non offensive) items that the user has browsed in previous sessions. (there is a way to track via cookie for non logged in users)
4. Promoted items: show items that are promoted and make big business benefit. However, do not get swayed away by this, show one or two promoted items to keep viewers interest.

There are recommendation engines, but as a developer business needs drives the recommendation creation. So, many times I find it's better to hack some SQL scripts to get them.