Monday’s news that Google has implemented a form of artificial intelligence as part of its page-rank algorithm should come as no surprise to sci-fi fans, who have long expected this development. For the rest of us, here’s what’s going on in the simplest of terms.
RankBrain is the name of Google’s newest addition to the page ranking system that inspires so much study and frustration. It is a machine-learning artificial intelligence system that helps the search giant sort data to process the results it offers users. The artificial intelligence/machine learning aspect means that unlike traditionally programmed machines, RankBrain continues to refine its activities based on previous “learning” and making connections between the information it processes.
That applies both to interpreting the queries users make and prioritizing the best answers from the information included in the results. Making sense of searches that use natural language isn’t easy for a machine, and one of the things RankBrain works to accomplish is sort out just what a human being might really be looking for with an ambiguously worded query. By finding patterns in searches that don’t appear to be related but have similar results, the program can learn to identify language usage patterns to deliver more accurate answers that correlate to the intent of the search.
While the news was only shared this week by Bloomberg and confirmed by Google to Search Engine Land, RankBrain was rolled out early in the year and has been making its cool or creepy contributions to global searches for several months. Google reports that the actual learning takes place offline, with batches of search data that are analyzed by the program so it can learn to make predictions. After testing, if the predictions hold up, RankBrain is assumed to have learned useful skills and the version that includes the new information goes live.
Somehow in all this, RankBrain is assessing an as yet unrevealed factor in the content it scans while live that becomes a contributing component of page rank. The third most important factor in page rank, specifically. Bloomberg’s article includes this information from Google:
“In the few months it has been deployed, RankBrain has become the third-most important signal contributing to the result of a search query.”
So what are the first and second most important elements? Google’s not telling. Search Engine Land has asked – twice – and gotten no answer. It’s all very exciting, and confusing – and infuriating for those who really want to know. There’s a ton to explore about RankBrain and Google’s machine learning research if you’ve got the time and the motivation. You can even use this fun word-to-vector tool to play around with natural language processing yourself. But as for page rank? The best advice is to keep a strong focus on creating high-quality content, because that’s the one thing we already know Google is very serious about.