Thoughts on Data Collection and My Positive Experience from it
Some Background and Concerns About Data Collection
Whether you’re shopping for clothes or browsing through your favorite social media platform, your actions are being monitored on nearly every website nowadays. Whenever a user signs up for websites such as Amazon or Facebook, people brush over the end-user license agreement without hesitation and agree to it. What’s usually hidden in the giant wall of text of an agreement is the permission to collect and use your information with machine learning to better market towards the user. What makes it worse is that not only does your information get tracked, such as your search history on the website, but sometimes these agreements also include access to your photos, camera, or microphone which can be used to listen in to your conversations. Companies such as Google and Amazon have their smart speakers as a way to provide service to justify the need to listen in, with the tradeoff of being able to connect and control your smart devices (lights, security cameras, thermostat, etc.) under one device for convenience. This has lead to the ethical question of whether companies should be allowed to be this invasive throughout the years, and as of recently, Apple is trying to respect and protect their user’s privacy by announcing their future feature of the App Tracking Transparency earlier this year.
How This Has Improved my Experience
Although I don’t ever click on the personalized ads whenever I’m scrolling through Instagram, one of the few instances I enjoyed the results of some form of data collection is getting my Discover weekly playlists from Spotify. In this instance, the most important data that Spotify collects from their users is the music they search for, frequently listen to. This information is used to build a profile for their user, and then songs are used about finding more songs similar to what the user enjoys. In order to determine what exactly the users like to listen to, convolutional neural networks are used on the song’s audio to categorize the songs by their genre or mood. They also use Natural Language Processing to tag common phrases or terms that frequently show up amongst artists and songs and match other artists and songs that share common terms. I believe the engine that Spotify uses to recommend songs to their users is a great use of data collection that uses machine learning to improve the user’s experience, as I feel like I’ve found songs and artists that have become my favorites throughout the years due to the music streaming services recommendation.
