Why Machine Learning Is Revolutionizing Search

by Mark Traphagen
Here’s Why Monday Episode 57 Machine learning is the next great computer revolution, one that is already here. We don’t have to wait for the future; Google has been using machine learning to solve many complex search-related problems for years, and the applications keep growing, including last year’s announcement of the addition of RankBrain to the search algorithm.Read the full article

Why Machine Learning Is Revolutionizing Search

In this week's episode of Here's Why, Mark Traphagen asks Stone Temple Consulting CEO, Eric Enge, to tell us why machine learning is such a revolution, and what it means for the future of search! Links Referenced: The Machine Learning Revolution: How It Works and Its Impact on SEO https://moz.com/blog/machine-learning-revolution The Stone Temple Twitter Engagement Preditor: http://www.retweetpredict.com/ Complete Here's Why video archive: http://www.stonet.co/HeresWhyArchive Mark: So Eric, before we dig into what machine learning is and how it will affect search and SEO, you told me that you've gone extra deep into this topic. Tell our viewers about that. Eric: Sure. When I realized just how big machine learning is becoming for Google and others, I decided I needed to know it from the inside out literally. So I enrolled in a university level course in machine learning and went on to start writing my own machine learning programs. One of those is actually included in the article I wrote for the Moz Blog. Mark: Okay, yeah, that's sure going deep. Speaking of your Moz article, you do a really thorough job of explaining what machine learning is and how it will impact SEO. Now can you summarize that for those viewing today? Eric: Sure, happy to do that. At a simple level, machine learning is a class of algorithms that enable computers to learn from a feedback loop of information. The computers then, in a sense, alter their own programming in response to what they learned. Mark: So what's an example of how that could be put to use? Eric: Well, one of the first applications that I learned about came out in an interview I did with Google's Peter Norvig back in 2011. Peter told me that when Google wanted to build Google Translate, they found examples of material on the web already translated into various languages, fed that into a machine learning algorithm, let the machine figure out the principals of translations between the various languages. Mark: Wow, that sounds incredibly complex. Eric: It was, but now in 2016, Google and other companies such as Apple, Microsoft, and Facebook are using machine learning for all kinds of applications. Mark: And you mentioned before that you created your own machine learning based program. Tell us about that. Eric: Well yes, I used what I learned in the course by Andrew Ng, I hope I pronounced that right, of Stanford University. And I used data from our own study of what causes engagement on Twitter combined with social authority metrics from Moz's Followerwonk tool to build the Twitter Engagement Predictor. Given the Followerwonk social authority of the tweeter and how many of the various features she or he has used in the tweet, things such as images and the like, the program predicts the probability of the tweet getting a retweet. Mark: Now let's move on to how machine learning is affecting SEO. Eric: Well, I think a prime concern at Google is the quality of search results. By that I mean when a user clicks through to a site from a Google result, how good and how satisfying is that result? Does it quickly and effectively give the user what he or she was looking for? Does it anticipate other related needs the user might have? Does it deliver all of that in a fast and easy to understand manner? Of course, quality of results has always been important to Google. What's different now is that I think they are using machine learning in things like the Penguin and Panda updates as well as other search quality algorithms to be able to better than ever really tell if a human would think a web page is the best possible result for their query. Mark: And the implications of that for anyone doing site SEO should be obvious, I would hope. Eric: Indeed. Now more than ever, you've got to be focused on user intent, related intentions the user might have, and identifying gaps in your content, and how your quality compares to your competitors for the same query, and how you're measuring and iterating page improvements for the future. Mark: Well, thanks, Eric. There is a lot more and I mean a lot more about how machine learning works and the practical implications for our SEO at your Moz article.