Data 607 - Discussion 11: Recommender Systems

Task Assignment

Your task is to analyze an existing recommender system that you find interesting.

You should:

• Perform a Scenario Design analysis as described below. Consider whether it makes sense for your selected recommender system to perform scenario design twice, once for the organization (e.g. Amazon.com) and once for the organization’s customers.
• Attempt to reverse engineer what you can about the site, from the site interface and any available information that you can find on the Internet or elsewhere.
• Include specific recommendations about how to improve the site’s recommendation capabilities going forward.
• Create your report using an R Markdown file, and create a discussion thread with a link to the GitHub repo where your Markdown file notebook resides. You are not expected to need to write code for this discussion assignment. Here are two examples of the kinds of papers that might be helpful backgrounders for your research in #2 above (if you had chosen amazon.com or nytimes.com as your web site):
  • Greg Linden, Brent Smith, and Jeremy York, “Amazon.com Recommendations: Item-to-Item Collaborative Filtering,” IEEE Internet Computing, 2003(!). https://datajobs.com/data-science-repo/Recommender-Systems-[Amazon].pdf
  • Alex Spangher, “Building the Next New York Times Recommendation Engine”, Aug 11, 2015. http://open.blogs.nytimes.com/2015/08/11/building-the-next-new-york-times-recommendation-engine/

Spotify’s Reccomendation Systems

We are all familiar with recommendation systems from Amazon, Netflix and Facebook but one system I really love is Spotify’s Discover Weekly playlist. I often find myself listening to it thinking “How on earth did they figure this out - are the hard wired into my brain?”. It turns out what goes into creating this list each week is really very interesting and summarized in this Quartz article http://qz.com/571007/the-magic-that-makes-spotifys-discover-weekly-playlists-so-damn-good/ (many other articles - including many that are super technical can also be found).

Scenario Design

For this case we will perform the scenario design once for the company (Spotify)

1. Who are your target users? Subscribers and non-subscribers to Spotify’s music services who have listened to a song recently (predictions don’t work well if you rarely use Spotify)
2. What are their key goals? Finding and listening to new music that they might not have found on their own.
3. How can you help them accomplish those goals? Creating a weekly playlist of songs that utilize previous listening history to customize which songs are included in the playlist.

Reverse Engineer

A lot of information has been published about Spotify’s recommendation system so I don’t have to reverse engineer it.

An overview of the technical process can be found here - http://www.slideshare.net/MrChrisJohnson/from-idea-to-execution-spotifys-discover-weekly.

A flowchart of how this process works at a basic level from the Quartz article is below.

How to improve going forward

A few suggestions from an active user:

1. Allow more interaction with my ‘clouds’ of musical interest. Eg. That one time I listened to disco music might not have been a fluke.
   
2. Create multiple kinds of discover weekly playlist. Rather than only having one create a active, concentrate, mellow, etc playlist.
   
3. Integrate with other systems (i.e. iTunes and Apple Music) to further enhance recommendations. Many users discover songs on Spotify but play them in other services.