3 ms·
A pure recommendation engine of the sort we learn about in class would simply aggregate user preferences based on whatever content-appropriate properties are av
by cratermoon 6y ago
A pure recommendation engine of the sort we learn about in class would simply aggregate user preferences based on whatever content-appropriate properties are available. A real-life recommendation engine is written and deployed as part of a business plan.
I think it's fair to say that all recommendation engines deployed by for-profit companies have some element of recommending things that drive more revenue for the company. The question is, how much does end user give up, in terms of getting actually interesting recommendations vs being recommended something that is more profitable for the company, in exchange for getting things of interest to their attention.
I've seen enough to be convinced, anecdotally, that YouTube's recommendation engine is heavily skewed towards revenue-driving. For example, how much of what YouTube recommends is influenced by their big-money media customers? What users get is as marginally interesting as necessary to addict viewers.