A Machine Learning Story

Machine Learning is a subset of a larger category of machine based learning and decision making known as Artificial Intelligence or acronymically known as ML and AI, respectively.

The recent emergence of ML/AI has largely been powered by convergence of compute costs and data storage costs that have lowered to the point of making these resource intensive processes available to a widening circle of business applications. A once expensive and therefore elite approach to data mining has become a relatively affordable commodity.

Over the past 10 years the ML/AI developer space has evolved from relative scarcity to an abundance of technologies, techniques, tooling including the launch of software-as-a-service offerings targeted at novice operators on platforms such as Google Cloud Platform and Amazon Web Services.

Another key factor to the rising trend in ML/AI popularity can firmly be attributed to the value and promise-of-future-value that ML/AI holds. A quote from a Forbes Insights article from December of 2018 shows the current state of confidence in ML/AI value – 

“The closest thing to a universal belief in business today is that artificial intelligence (AI) has the power to deliver decisive competitive advantage. In fact, 91% of the 700 global C-suite executives surveyed by Forbes Insights agree that AI adoption will help them outpace their industry rivals in the years ahead.”

https://www.forbes.com/sites/insights-delltechnologies/2018/12/04/behind-every-ai-strategy-is-a-data-strategy/

The Brainly Machine Learning Story

At Brainly we are focused on providing students with personalized, real-time learning help and access to the world’s best educational products.

Like many other companies today we recognize Machine Learning as a key ingredient to creating better products. When used in combination with our scale of traffic and data, we see this as a competitive advantage and differentiator for us. As we look to a fast approaching future, we see highly competent and creative applications of Machine Learning as core to our offerings and simply as table stakes for being competitive in tomorrow’s ed-tech market.

To help explain the possibilities and power of Machine Learning at Brainly I like to use a narrative that puts the focus on the user’s point of view.

Imagine a user named Liam comes to Brainly for the first time. This is Liam’s first interaction with Brainly and therefore Liam is viewed by Brainly as an anonymous user with no history.

Liam begins to browse Brainly content, and within a handful of interactions he has begun to tell Brainly a story about his educational needs and interests. Before the end of an average session, Brainly has begun to generate a model of Liam’s interests and has stored his interaction history and learner profile against a temporary identifier. While Brainly doesn’t know Liam’s name or personal data this temporary id helps to customize his experience the next time Liam returns to Brainly.

After a couple of sessions, a rich history of interactions has allowed Brainly to become predictive about Liam’s needs. Brainly is able to correlate Liam’s content browsing and searches with his curriculum needs and school system and Brainly has begun suggesting the appropriate subject matter and grade level content that might be relevant to Liam’s needs and interests.

After a few more sessions the results have become highly tailored to Liam and the content being offered matches content from his classroom textbook and classroom learning resources whenever available and appropriate.

In areas where Brainly recognizes Liam may need more help to understand the subject matter, he is matched with, and offered, follow-on opportunities for practice and enhanced learning. 

In areas where Brainly recognizes competence in a subject Liam is offered the opportunity to help other learners who are struggling with that topic he has mastered. 

When Liam eventually decides to register, his rich history and learning profile are transferred from an anonymous temporary profile to his registered identity which allows Liam the same level of personalized experience across any computer or mobile device he uses to log into Brainly. The more Liam uses Brainly, the richer and more meaningful the predictions and his subsequent interactions with Brainly become. 

Brainly’s work to understand Liam’s needs doesn’t stop when Liam leaves Brainly. With over 150 million monthly users, Brainly’s intelligence is powered by Machine Learning algorithms assessing patterns and behaviors across a global community of learners to create a better understanding of how to improve Liam’s learning experience.

The next time Liam logs in he will be guaranteed an improved experience that has anticipated his learning needs and he will be offered suggestions for content before he makes any searches or indication of what he’s looking for. Every click and search Liam sends to Brainly will refine his highly personal and tailored experience in his future interactions with Brainly.

This highly personalized and tailored approach to learning creates a positive feedback loop, resulting in better educational outcomes, more confident learners, and a community of learners working together. Nobody knows everything, but everybody knows something, and Brainly powered by Machine Learning helps our community of learners learn better.

Afterword

Some of the capabilities described in the story above are either under-development or goals to be accomplished in the future as we expand our team and capabilities in ML/AI at Brainly. If you have a passion for working on big and challenging ideas that help millions of people every month AND you have skills and experience in one of the areas of Machine Learning for personalization, OCR, Natural Language Processing, or large scale data analytics, we want to talk to you. Email me at (bill.salak [at] brainly.com]) to start a conversation about how we might work together.