DeepMind powering the Google Play Store’s app recommendation system is the News of the Week

It was not long before when Artificial Intelligence (AI) came into existence to make out lives much easier than before. It is due to the technological advancements that we witness such drastic changes in our daily chore. Where technology has made our lives easier, it has brought about some social, physical and relevant implications as well.

Since machine learning has become famous, many labs have been developed that research on it. DeepMind is one of such labs that is regarded for its extensive works in AI and machine learning. Some time ago, Google purchased this lab, which is now working with a team under Alphabet at Google.

DeepMind looks for the technology used behind Google’s AI. Currently, DeepMind is used by Google for its Play Store app recommendations. Yes, you heard it right. Google is leveraging from DeepMind in full swing and we can see the improvement.

According to DeepMind team, Google has one of the largest recommendation systems due to millions of users on its platform. As we all know, Discovery is one of the most important features of any recommendation system and Google utilizes both editorial and algorithmic curation. The Play Store apps recommendation system is not like any other system. Such a huge amount of data cannot be found in many systems other than Google.

Google utilizes the previous behaviors of users. For instance, Google keeps a track of the “past user experiences”, downloads and installs and much more to offer much polished and personalized experiences.

DeepMind suggests that they started working with the Play Store team to further improve the recommendation system of the app. DeepMind is working on checking the relevance of the app with the user to generate more personalized recommendations.


Google’s recommendation system does not work in one step, rather it is divided into three models of recommendation system. The first one to be the candidate generator, the second one to be a re-ranker, and the third to be multi-objectives optimizer.



In the first step, when the user enters their preference the retrieval model finds the most suitable suggestions for them. After this, the reranker set the suggestions according to user’s preference and then the multi-objective optimizer presents the most suitable and preferred suggestions on the screen.

Hence, while we wait for the apps suggestion to show up, Google runs an extensive work at the back to come up with the best suggestions.

As the collaboration between Play Store and DeepMind teams brought them together and increased communication between them, they were able to make improvements in products as well. Both the teams are considering final testing phases, algorithm designs, and implementation to improve the product that the user is looking for.

The collaboration between Google and DeepMind is expected to give fruitful results in the future. Let’s see what Google is planning to achieve with the help of DeepMind.

Read next: This Page Reveals Extent of Google’s Knowledge About You
Previous Post Next Post