Here’s some news you may have missed about the latest Google algorithm update. In the coming months, Google will begin rolling out a new search algorithm based on a recently developed cutting-edge technology called Multitask Unified Model or MUM. Google estimates that this new format, which is based on the T5 text-to-text framework, will make search “1000 times more powerful than its current system BERT.”
The jury is still out on whether MUM will make SEO obsolete, but there are some experts who have already started to forecast so. They say MUM will not find pages directly, it will, instead, ask users questions, & MUM will answer them just like a human assistant would. The importance of keywords would be minimal. Others, though, beg to differ.
What Is Google’s Current Search Ambition
Google’s aim in delivering search results is to maximize access to information, which means that it needs to parse relevant images, videos & audio files in addition to text based files in order to serve up the best results. Originally, when Google was launched in 1997, its aim was more limited. Larry Page, Sergey Brin, & Scott Hassan wanted to search for text only, in publicly accessible documents offered by Web servers.
The latest Google algorithm update MUM, though, will give Google the “teeth” to come much closer to its evolved aim since it’s multimodal – meaning that it can, currently, understand information from both text & images. Google believes this capability can be extended to include video & audio in the near future.
Google said this on its official blog: “We’ll bring MUM-powered features and improvements to our products in the coming months and years. Though we’re in the early days of exploring MUM, it’s an important milestone toward a future where Google can understand all of the different ways people naturally communicate and interpret information.”
How Will MUM Be Different
Not only is MUM multimodal, it also “understands” language much better than previous search models. For instance, if you search for museums in the south of Toronto, Google might currently misunderstand the “in the” & the “of” giving you a host of results for museums x-distance “from” the south of Toronto. MUM will offer far more relevant results since it will search within the context of the entire natural language question – by better understanding the search “needs”, it will offer a far more correct result.
The latest Google algorithm update MUM is trained across 75 languages & can handle different tasks simultaneously, thus giving it greater comprehension than it had with previous models. It’ll be faster & will obtain results quicker – Google estimates that relevant search results that equate to complex queries are only obtained after at least 8 combined search attempts, MUM will answer questions, according to Google, as if it were a human expert.
The technology’s multimodal capability will mean that users might be able to use image search combined with either voice or text search to gain insightful results. For instance, a user could upload an image of a teddy bear & ask Google if the toy is safe for a 2-year-old, & get an informed expert answer.
Internationalization Of Search
The new search algorithm will concatenate data from multiple language based sources, which will effectively internationalize search. Currently, if you search in English you receive English-based information, but since MUM can search across 75 languages you will then be able to access specialized data that is, perhaps, only available in a foreign language. This facility will be especially helpful to tourists, but will go beyond just tourism. For instance, let’s take medical researchers wanting to compile all relevant data on an epidemic, find & assimilate all available information; obviously results that include all available research online in multiple languages would assist their efforts enormously – & it’s exactly the kind of thing that will be solved via MUM.
Google has said it is testing MUM thoroughly like it did BERT, before rolling it out the final version for the world to use.
Image credit: Google
