Website Ranking Algorithms
https://www.searchenginejournal.com/google-neural-matching/271125/
The article discusses a new algorithm that Google is implementing in their search function. Historically, Google has used the original and updated version of Alphabet/Google’s CEO Larry Page’s ranking method, PageRank. PageRank ranks article relevance by counting the number of links and quality of links to said webpage. The underlying assumption is that, generally, more relevant pages will have a higher number and better quality of links than less relevant ones. The article, however, discusses a new algorithm they call Neural Matching. The new algorithm uses actual content within the article to rank how relevant it may be to a user’s search. This means the algorithm must understand synonyms and the ideas in the article. Neural networks now have the ability to understand complex concepts and Google is trying to apply this to their search engine. The article states that pages will still be ranked with traditional information retrieval that depends on the network structure (the links in this case) to “vet” the websites, then adjust the rankings of the top sites based on what the algorithm understands the important concepts to be.
The article directly relates to our class because we have discusses networks overall and PageRank as a network. The article basically states that the new algorithm will no longer rely on network structure to refine searches, but rather the actual content that the pages have. If it seems like information is more relevant to the search on one page than others, the article will rank higher according to the neural network. While they will still use traditional information retrieval to primarily rank the webpages, the reranking will no longer rely on the network of webpages in the way that PageRank relies on them.