Understanding Google’s Latest Search Engine Algorithm
Understanding Google’s Latest Search Engine Algorithm
  • Reporter Lee Ji-hwan
  • 승인 2022.12.10 01:28
  • 댓글 0
이 기사를 공유합니다

▲Google MUM / Link Building
▲Google MUM / Link Building

 During Google’s path to becoming one of the most influential Tech Giants, Google has started to impact our daily lives in nearly every aspect. Google started in 1996 as a search engine and is continuously researching to create a more accurate search machine to this day. 
 One of the most praised features in its search engine results page is the snippet, which was introduced in the fall of 2014. A snippet, a small piece or brief extract, can be found at the top of Google’s search results to answer simple queries quickly. Google’s search engine algorithm makes use of the most reliable and accurate sources as the answer to the question, but it often fails to successfully answer the question. Not only this, but the Google snippet is unable to comprehend a question asking for multiple pieces of information at once. For instance, if a user wants to compare how hard it is to climb Mount Everest to Mount Kilimanjaro, they need to google the height of each mountain, then the weather, and much other required information. However, Google started to employ a new search engine algorithm called Multitask Unified Model (MUM), which can answer the same question as if it is a professional climber by providing all relevant information.
 MUM was first introduced in May 2021 and was first applied in August 2022. Unfortunately, it is yet unable to solely rank the results or improve the quality of snippets. Through continuous updates to further develop their latest search engine algorithm, Google aims to enable their search engine to understand advanced language with only subtle information available. Along with the overall improvement, Google intends to add three new features to its search results. The search results will provide information in relevant areas by analyzing the results for a better insight on the topic personalized for the user. Also, the information provided will be categorized into subtopics allowing users to select an area to explore more. Lastly, visual exploration will be made possible by allowing users to search by images, videos, and articles with relevant questions.
 Before the introduction of MUM, the first deep learning system applied to the Google search engine was RankBrain in 2015. It was the first AI system to link different words with its concept allowing the search engine to find more relevant results. To this day, RankBrain plays a crucial role in providing the most suited results. In 2019, Google employed Bidirectional Encoder Representations from Transformers (BERT), which not only matched words but could understand the meaning and intention connoted inside the combination of words. 
 With MUM, Google is attempting to search for multiple information provided by sources written in different languages or formats. It has learned more than 75 languages, and research states that it has the potential to break down language boundaries when transferring knowledge. It can learn from sources that weren’t written in the language used to search and can provide relevant information translated with high quality. Also, MUM is multimodal, allowing them to analyze and understand the information in different formats. Google is attempting to allow users to ask a question related to an image or article uploaded and provide personalized answers.
 Google has successfully taken one more step in its plan in arranging a better environment for users to search. Throughout the next few years, they are planning to improve MUM and provide multiple features that allow users to ask more diverse questions in multiple formats and answer by utilizing sources unused before its introduction.