Spectral - User-friendly recommendation system (Round 4)
Summary
Spectral is a news article recommender system that reduces the effects of a filter bubble. Through machine learning it adapts and learns the limits of each user's reading preferences. The aim is to overcome inherent political bias while improving reader comprehension in a user friendly way.
The solution
Conventional recommender systems are primarily serving the hotspot of preferences, reinforcing existing beliefs and promoting filter bubbles. Readers grow accustomed to certain text standards and forms, making them less receptive to different types of narratives. Spectral will recommend articles that might be slightly more challenging or demand a new perspective. Through machine learning the aim is to find inspiring texts, personalised for each reader, to help them escape their personal filter bubbles, and shape a better-informed society.