Google and Its New Machine-Learning Project Keen

Today, a new project called Keen is launching from an in-house incubator for new ideas of Google, area 120. They intend for it to help users tracking their interests. The application is like a modern redesign of the Google Alerts service. It allows users to monitor the web for specific content. Keen leverages a combination of human collaboration and machine learning techniques to help users curate content around a topic. It won’t send emails concerning new Google Search results, as was the previous model.

Each area of interest of the individual is called a “keen.” That word references someone with an intellectual quickness. C.J. Adams is the co-founder of the project. The idea for the project came about after he realized he spends too much time on his phone. Adams was mindlessly browsing images and feeds to fill his downtime. He realized that time might be better spent learning more concerning a topic he was interested in. Perhaps it could be something he always wanted to research more, or a skill Adams wanted to learn.

He and four colleagues at Google worked in collaboration with the company’s PAIR (People and AI Research) team. Thus, they wanted to explore the idea together. The partnership focused on human-centered machine learning. They intended to create what has now become Keen.

Keen is available both on Android and on the web. First, you must sign in with your Google account. Then you must enter a topic that you want to research. It might be something like learning about typography or learning to bake bread and bird watching. These are just some of the topics Adams suggested in an announcement about the new project.

Keen

Moreover, Keen can suggest additional topics related to the user’s interests. For example, if you type in “dog training,” Keen can recommend “dog training videos,” “dog training books,” “dog training classes,” and so on. You can click on the suggestions you want to track, and a “keen” becomes available.

When you return to the keen, you will find a pinboard of images that link to web content that is close to your interests. For the dog training example, Keen found blog posts, YouTube videos, and articles featuring curated lists of resources, and Amazon links to dog training treats and more.

For every collection, the service uses machine learning and Google Search to help discover more content related to the given interest. The better these recommendations become, the more you add to a keen and organize it.

Factually, it is like an automated version of Pinterest.

When a keen is created, you may then optionally add to the collection. Moreover, you can remove items you do not want. You can also share the keen with others, to allow them to add content. When new content is available, Keen can also email you alerts.

In powering its news feed in the Google app, Google, to some extent, already uses similar techniques. The feed uses a combination of items from your Google Search topics and history you explicitly follow to find information and news it can deliver to you.