Hotstar is using machine learning in order to provide a personalized experience to its users. It calculates the total watch time per user per month. This helps to derive algorithms which are specific to every user and their preferences. Machine Learning models are also used to deal with the diversity of users across regions.
Star India is betting extensively on Hotstar with IPL matches as the largest online streaming platform is gaining momentum with movies, television shows, original series etc.
In the beginning, Hotstar intended to stream Cricket World Cup on its mobile application. Later as it expanded it increased its content to 35,000 hours with options of eight different languages in entertainment, movie, and sports.
In the ICC Cricket World Cup 2015, it garnered 340 million views across all the matches and over 200 million views during the Indian Premier League Season 8. It has over 100 advertisers and more than 100 million downloads. And a bulk of Hotstar’s revenue is generated through advertisements.
In the inaugural week of IPL 2018, Hotstar reached a record of 82.4 million viewers, with over 22 percent of the tournament’s overall viewership. Star told that IPL viewership on Hotstar had a growth of 76 percent over the last year.
Hotstar leveraged Akamai Technologies, the world’s largest and most trusted cloud delivery platform for their global streaming.
This cloud delivery platform delivers ninety-five exabytes of data, ingest 2.5 exabytes of data per annum, and interact with over 1.3 billion client devices per day.
This helps the machine learning engines to automatically improve reliability, performance, and security.
It also integrates with web and mobile performance, cloud security, enterprise access, video delivery, analytics and reporting solutions which helps in delivering superior experiences.
Hotstar relies on Akamai’s solutions to cater to the vertical growth in demand, especially during IPL. Auto-scaling is one such method that the company believes may not always prove to be a useful strategy.
If the legacy backend fails to scale beyond a point, it makes them difficult to accept newer customers on the platform.
Machine Learning is applied by Hotstar to achieve greater customer engagement like Personalised experience. It calculates the total watch time per user per month.
This helps to derive algorithms which are specific to every user and their preferences. Machine Learning models are also used to deal with the diversity of users across regions.
Local collaborative filtering models to deal with regional users need to be deployed on the top of global recommendation models.
Modeling strategies such as latent semantic index, clustering or deep learning generate personalized ads and content feeds.
These also include identifying correct data, translation between business and analytics, keep pace with the rapidly evolving industry, customer centricity and expanding analytics skill sets.
Hotstar also uses real-time stream data platform, Knol for data exchange. The app collects huge amounts of data (around 10 TB) every day including ad impressions, behavioral clickstream data, customer support data etc. for solving business problems.
The Machine Learning team of Hotstar is working on various POCs to better understand the future demand and needs.
Some of the POCs they are currently engaged in is machine translation for subtitles, audio-to-text conversion, Video processing-compression, objects detection, scene classification and a way to define similarity between movies using Word2vec.
The research team is finding algorithms to automate feature generation. For this, they are exploring Autoencoders and Deep Learning algorithms which will help in the future expansion.
IPL 2018 might just turn out to be the breakthrough India was looking for in live sports streaming!
Interested in Machine Learning? Learn Machine Learning from Basics to Advanced concepts from the Best faculties with industry relevant case studies. To know more Click here: Machine Learning Course
What are you waiting for? Register Now!