(Source: The Marketing Technologist)
(Source: The Marketing Technologist)

 

Have you ever wondered how YouTube recommends videos that you may enjoy? How Facebook populates your newsfeeds with content that is relevant? How Netflix always seems to know which movies you want to watch? And how Amazon suggests products that you may be interested in purchasing?

This is the work of a recommendation engine, a very practical application of data science and machine learning. Recommendation engine are used by organizations such as Facebook, Amazon, Netfilx, and YouTube, to match products, services, and content to the relevant consumers.

In all of the cases, there is a common aspect; the requirement to provide users with an array of relevant possibilities to save time and effort involved in finding them. This ease of use experienced by the consumer is often a reason for increased sales and revenue.

Let’s look at some details of Recommendation Engines

There are 3 approaches for creating such Recommendation Engines

  • Content based filtering
  • Collaborative filtering
  • Hybrid filtering

Content Based Filtering:

In content based filtering, recommendations are made based on what users have liked in the past. Therefore, this approach requires a considerable knowledge engineering effort from the organization in profiling users and content. This method is used by multiple organizations such as Rotten Tomatoes, IMDB, and Jinni.

Collaborative Filtering:

In collaborative filtering, your recommendations are made based on what similar users have liked in the past. This involves studying and tracking the behavior of users but does not require much knowledge of the deliverable itself.

Social networking platforms such as Facebook and Twitter are known for using collaborative filtering when they suggest friends, connections, pages, posts, etc.

(Source: The Marketing Technologist)(Source: The Marketing Technologist)

 

Hybrid Filtering:

This approach is a combination of content based and collaborative filtering with a goal to capitalize on the advantages of both. There are multiple ways to combine content and collaborative filtering such as averaging, weighing, cascaded, etc. Netflix is an example of the hybrid filtering approach. The platform makes recommendations by comparing the search and viewing habits of similar users (collaborative filtering) and by offering movies that share characteristics of their predecessors (genre, cast, ratings etc.)

The Netflix Prize Story

A notable event in the history (https://en.wikipedia.org/wiki/Netflix_Prize) of recommendation systems is the Netflix Prize – an open competition sponsored by Netflix to find a team/individual who could take a dataset of over 100 million movie ratings and return recommendations that were 10% more accurate than Netflix’s own. The idea was to develop an algorithm, which would be able to predict the ratings of films solely based on historical data

However, the solution was not implemented due to several factors that made it unfeasible for real world use

 

How do Recommendation Engines add value to business?

In his book “The Paradox of Choice”, Barry Schwartz argues that eliminating consumer choices can greatly reduce anxiety for shoppers (https://www.ted.com/talks/barry_schwartz_on_the_paradox_of_choice). With many shoppers moving online, this concept has grown in scale because now they are exposed to millions of potential options. But it is important understand that users will not buy everything, hence it becomes extremely important for the business’ to make recommendations on the most appropriate items in order to maximze the potential of the sale. One of the best uses of recommendation systems is of Amazon, which uses complex algorithms for optimal product placement.

How BizofIT is using Recommendation Engines?

BizofIT (https://www.bizofit.com/) is the de-facto platform for linking the Enterprise clients with the appropriate IT service providers based on their needs. This mapping is enabled by the sophisticated recommendation engines working under the hood to enable IT service provider and Enterprise clients to find one another using cutting edge machine learning and artificial intelligence techniques scaled to work in real time with no compromise on performance. This makes BizofIT the perfect choice for the Enterprise clients as well as the IT service providers.