In this technology-driven era, recommendation engines today are playing a vital role in our day to day life. These technologically advanced engines are changing the way how we used to make a decision and decide on something. From buying t-shirts online to deciding on which movie to watch to which song to listen to, recommendation engines are everywhere.

For example, have you ever noticed your phone’s music player that suggests you songs every now and then, and if not completely, at least to an extent, the suggestions are just what we actually want to listen. Another example is the online shopping app you use. It also keeps suggesting you clothes and accessories you might want to check or even want to buy. It even suggests your product that actually fits perfectly to the budget you have.

But, how does recommendation engines do that? How can it know and understand the user? How much time does it actually take to deliver that accuracy?  In this article, we will have a look at a use case that would give you all a view of how a recommendation engine delivers results.

Before jumping right into how a recommendation engine reads and understands its user, let us first have a look at what a basically a recommendation engine is. So, recommendation engines are intelligent information filtering system. They are developed in such a way that they narrow the decision-making process and predicts and shows the results that a user would like to have a look. Recommendation engines are not at all 100% accurate, however, with all the recent advancements they have reached a significant level of accuracy.

Talking about how this intelligent system works, basically, a recommendation engine works based on data collection, data storage, data analyses and data filtration. So, it takes a little bit of time for the system to understand its user. However, there are few recommendation engines that are pretty advance and doesn’t take much time to deliver results that are significantly accurate.

How Viu Uses Machine Learning To Advance Its Recommendation Engine

Viu is one of the most emerging platforms in the OTT space in India at present. And it is not just the content that has done its job in making popular, but its recommendations engine is also playing a vital role. Let us have a look at how Viu clip is using AI and ML to make its recommendation engine top-notch.

The recommendations of the platform are based on data collected from the end user — such as what kind of content the user is seeing, how much time it is sticking to it etc. The more data keeps on getting collected, the better the engine becomes for the end user as the platform keep changing the recommendation algorithm and tailor as much as possible. The accuracy of the recommendation engine is also depending on the same parameters. However, there are few recommendations that are available to users all the time because the hit rate on that is very high.

Furthermore, Viu has also created experimental ribbons where the engine keeps checking the data and it depends on the segment of users on more than 200 attributes. Across all these attributes, multiple permutations and combinations get created. And depending on that, the recommendations get created.

Talking about how much time does it take for a recommendation engine to understand its user, Viu uses a hypothesis, which is based on data collected from first party and third party sources. Using that data, the platform trying to learn from the other market and does profile matching. For example, they try to match and see if a content that is popular in one region can also be popular in some other region.  In this entire process, machine learning plays a vital role.

“Once we launch in a region, we start collecting the data and when we reach certain critical mass, the ML engine to become more accurate,” said Abhijit Bhide, SVP – Engineering at Vuclip in an interaction with Analytics India Magazine.

Viu is actively using machine learning to better the results of its recommendation engines. And today, the platform has reached a level of accuracy that in a brand new place, within a month,  the platform’s is able to get a fairly good set of recommendations for the end users.

Outlook

Recommendation engines today are a vital part of every online platform — whether is media ore-commerce. While big players like Amazon, Flipkart, Netflix, are extensively advancing the technologies behind their recommendation engine, emerging players are also not much lagging behind. And behind all these advancements, machine learning is proving to be the prime catalyst.