STOR-i Forum: James Grant
Personalised recommendations have become a staple feature of many online services and stores. State-of-the-art methods consider a history of user interactions with products and select recommendations based on the preferences of similar users. In an online betting context a number of unique challenges appear due to finite product lifespans (i.e. once a match has been played any bets relating to that will expire) that prevent the application of the most popular recommender systems approaches. This talk will cover these issues in more depth and detail a proposed solution which works by considering a broader definition of a “product” as a class of bet from which many may be selected.