Google’s Slow Takeover of Apple Products
Apple has decided to integrate Google Web Search into Siri’s capabilities to replace Bing. Apple prefaced this change by saying they were striving for more consistency in the search function across different platforms. A significant complaint users had of Siri was that Siri was “dumb” or “useless,” since she wouldn’t return relevant answers back to the user. In cases of particularly long vocal queries, Siri wasn’t able to provide results that accurately responded to the request. Google’s Web Search is more powerful in terms of long-tail questions; not only is the engine better at data extraction, but it also has built in a stronger ranking system and results format. The quality of results is better due to: 1) filtering of web content; and 2) presentation in a list format, which works great for Siri on iOS devices. What’s important and related to this class is the ranking system.
One way to rate websites is to use a link analysis algorithm of hubs and authorities – the idea that the ratings of hub and authority pages determine relevance and accuracy of webpages. Authority pages are pages that have to do with a certain search topic. Hub pages are tangential pages that can link or reference authority pages (but aren’t pages of the topic itself). The idea is, the more hub pages point to an authority, the more relevance and credibility an authority page has. The more higher-ranked authority pages a hub page points to, the higher quality a hub page has. This keeps iterating back and forth to eventually reach an equilibrium set of values that can then be “normalized” based on the total value of authority or hub pages collected, respectively. Google basically uses this sort of paradigm to create a comprehensive ranking system in its search engine to provide users with only the top-quality pages. Using this idea to check both authority and hub pages, Google’s search engine filters out the best results.
Google’s PageRank, an algorithm named after Google’s Larry Page, is the most popular/well-known algorithm used to rank webpages. The basic idea is that in a network of nodes and edges, each node (web page) holds a certain value that represents their ranking and that the most important web pages will collect / pool a large amount of ranking, and less important web pages will collect little. The way that the values are circulated and equilibrium is reached is basically that each node splits its values among its outgoing edges simultaneously. A network at equilibrium means that with each cycle or iteration, each node will always hold the same value. Nodes receive values based on “votes” that are determined by hyperlinks from other pages that point to these pages. Or by clicks. These rankings are determined by how high of a value these web pages hold, and the higher the values are, the better and earlier these results will be displayed.
Paired with Google’s other strong algorithms, this PageRank algorithm finds and excavates only the most important results and returns them. For a service like Siri, the expectations of which are incredibly high because users demand quick and accurate answers, a strong search engine is imperative.
Link: https://www.fastcompany.com/40475434/siri-may-be-the-big-winner-in-apples-switch-to-google-for-web-search