I understand that Google is on a never-ending quest to give users the most relevant answers to their search questions. Because after all, a search query is basically a question. People land on your website looking for “the answer”. Recently, when I Googled “cult movies” I was presented with a timeline cover art preview of what Google deems cult films. I will not disagree that among the greated cult movies of all time include, “Erasure Head”, “The Big Lebowski”, and “Rocky Horror Picture Show” can and should be included the cult movie canon it seems a bit presumptuous for Google to use the knowledge graph to answer queries that could be considered ambiguous in nature. Some people may have a different opinion of what the classic cult movie is.
How Is Google Determining This?
Incredibly, I believe Google is presenting these movies using latent semantic analysis, where they look for articles/pages related to “cult movies” and aggregate the words (movie titles” that exist within these articles. While this may seem like a Herculean task, it’s actually quite feasible if Google is looking at some of the top movie sites (Rotten Tomatoes, Metacritic, Collider etc.) for the data. The other possibility could be that they’re using meta data (markup) from these movie sites, taking the number of occurrences under the sub category “cult films” and publishing in order of appearances within these lists.
Either Way, Isn’t This Scary?
Useful or not, I believe this presentation of the knowledge graph crosses the line a bit. What’s next? Working for an eCommerce site, they have definitely applied this to retail products by aggregating user reviews and price meta information. But when they delve into the territory of pop culture things can get scary. We’re no longer talking about cold hard facts like review numbers and prices but opinions. Shouldn’t it be left up to individuals and websites?