the algorithm knows you - and might even be shaping you
intro to a new series on understanding and developing recommendation engines
hello!!
i love algorithms because not only do they try to understand who you are, they end up revealing aspects of you - that you never thought you’d come across. Now, with great power should come great responsibility. We willingly or unwillingly give our data to these recommendation engines - often treated as black boxes - continuously predicting, and even determining our next moves. There might be helplessness. Loss of autonomy. But if we unlock our algorithms, we might just be able to turn the tables on who influences whom. And perhaps, in the process, understand why we behave the way we do.
Here’s a quick rundown of this series.
what i want to talk about:
recommendation logics - what goes on behind the curtain of your feed? we’ll be diving into content based, collaborative filtering, knowledge based systems, hybrid logics and much more.
how can we implement these logics in various mediums? how does the level and intricacy of logic change with each?
case study based deep dives: how spotify, twitter, youtube, instagram and other big names shape what you see. what we can learn from them and the differences that come with each?
thought experiments: how I might go about building recommendation logics for different use cases?
might touch upon consumer psychology, its quantitative analysis and how contexts might define it.
why am i doing this?
i work with these logics almost every day at work. I am currently only using excel and rules based logics - so here, I think understanding the fundamentals becomes important > rote learning of models to train. the specifics of the logic become more vague as we progress towards bigger data and more robust applications like python.
i want to see through the eyes of the recommender. how do they perceive me behaving? how do they expect me to behave? how do they predict what i might do/like next?
i lovee data. but one should know how how to use it - for each use case one comes across. recommendation logics also involve data inferencing - how do you connect those numbers on your screen to consumer behaviour? what does it tell you about them? what does it tell you about yourself?
who might find this relevant:
recommendation engine connoisseurs
data lovers
excel/python enthusiasts
curious minds
intentional doomscrollers
If you have anything, you’d like to see me cover, click the button below!
additional thoughts:
make your recommendations so good, it feels like #serendipity
as always, I leave you with a song: