“Deep Learning” Promises to Change Search Ranking Process

“Deep Learning” Promises to Change Search Ranking Process

A recent article in wired.com provided a fascinating – some might say alarming – analysis of the future of Google search and potentially much, much more.

“Deep learning” is a process whereby neural nets (masses of interconnected hardware and software) can learn… on their own.

Given enough examples to sift through, these nets “learn” how to identify other examples of the same things. And, they learn how to respond to search queries, sometimes better than the human-coded algorithms that Google now employs to generate page results.

How this happens is something of a mystery, which is both the fascinating and potentially alarming point. No one really knows right now. But artificial intelligence (AI), which is what these neural nets are exhibiting, appears to be the direction that Google is headed. The company recently moved its head of Artificial Intelligence, John Giannandrea, to oversee search engine development.

According to the article, not everyone, even at Google, is convinced that this is the best way forward. Since no one really knows how these neural nets create “rules” for themselves, it’s more difficult for engineers to identify specific factors to weight in order to optimize search results and ensure the process is fair.

Google introduced a form of AI, RankBrain, early in 2015. That program helps handle the 15 percent of all queries that Google hasn’t seen before. RankBrain is apparently better than the algorithms at figuring out what specific phrases and words in a search query mean and what other references might apply.

Deep learning isn’t limited to browser-based search queries. Digital assistants like Siri and Google Now use the same process to learn how to better recognize and respond to voice queries. Social media like Twitter and Facebook are also being affected.

As these nets continue to learn and demonstrate their effectiveness, the odds are good that we’ll be seeing more applications of this mysterious process across the Internet and potentially into wearable technology such as smartwatches and personal health monitors.

Deep Learning: An SEO Game-Changer?

What does this mean for dental practices? There’s no way to know for certain, but it seems very likely that it will become harder than ever to figure out the “rules” for search engine optimization (SEO). That’s because these neural nets generate their own rules by processes that aren’t fully understood. And, these nets keep evolving. While engineers can analyze the data to see how the nets treated certain queries, understanding why the results were generated is something else.

SEO is already constantly evolving as Google, Bing, Yahoo and other companies revise their search algorithms. That means that at any one point, SEO best practices are something of a guessing game even though there are known guidelines. The rule is test, revise, and test again for optimum search engine results placement.

As deep learning by neural nets pushes the evolution ever faster, the “rules” for SEO may become even more murky. However, Google and others will almost certainly have to figure out ways to keep advertisers and everyone else abreast of most of the changes. Advertising, after all, is where these companies make their money.

Whether you find the phenomenon of deep learning to be fascinating, alarming or both, it seems to be here to stay. I’ll keep you abreast of further developments as information becomes available.




Leave a comment

About the author
Colin Receveur is a nationally recognized speaker, author, and dental web marketing expert who has pioneered the way dentists market themselves online for the past decade. Since incorporating in 2001, Colin has established a rock solid track record with his dentist clients and turned SmartBox into a stalwart of proven results for hundreds of dental practices.