In one of the applications that I'm involved in on my spare time, with a friend we decided to add a recommendation section. With such section, users who interact with one product will get a recommendation of other similar products. I already read a little about this problem and seems to be not trivial. And because I like challenges decided to write my simple recommendation engine.
To make sure I won't discard this project too soon I decided to participate in Redis Hackathon not for prizes or splendour but to have fun with limitations and maybe cooperate with others thru post series.
My initial thoughts are to use Redis for the stream of user events and storage of products to match (eventually QuestDB but hey, it's Redis Hackathon!). Algorithms? Still researching, the easy one to implement seems to be the Jaccard index extended by weights on individual events.
If you are interested in this journey, follow me and in the next post, we are starting an adventure!