PageRank in Online Shopping
When you’re online shopping in stores such as Target or Amazon, there’s usually a section labelled “Recommended” filled with items that perhaps you realized you do have an interest in purchasing. But how do these sites know? Maybe we can relate this to PageRank. PageRank is used by Google to figure out which sites are relevant and trustworthy so that they can return these to you when you input something into the search bar. Authority points are given to sites depending on how many sites have been linked to them and in return, hub points are given to the sites that linked that certain site.
Maybe it works similarly when Target or Amazon is trying to find the most relevant items to recommend to you. Say you have an item you have purchased. Perhaps other people who have bought the same item as you are also buying another item, giving this new item authority points based on how many people have bought that previous item. Maybe there are more items that you have purchased, and other people are buying those items as well this new item, that would contribute a higher authority score to this new item and most likely in the end, this new item is the one that pops up in your recommended.
The article linked shows an example of this. The article describes a situation where Target ended up sending coupons to items linked to pregnancy to this teen despite the teen never outwardly saying anything about being pregnant. The Target statistician runs through the purchasing history of certain items and how they are linked to each other and eventually comes up with a pregnancy score for each shopper. This could be similar to having an authority score linked to a specific item related to pregnancy, or maybe authority scores are assigned to certain shopping themes that are then linked to other related products.
Relevant Article: https://www.forbes.com/sites/kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-was-pregnant-before-her-father-did/#2c16906c6668