Preference Networks

Acknowledgements

César Hidalgo for being such a great leader and helping me learn to focus on a task until it's completed. Also, the text above was almost copied verbatim from his description until I have more time to write my own account of the project.

Related Links

MIT Media Lab - Macro Connections

“Preference Networks aim to improve our understanding of the world by putting together the pieces that our scientific disciplines have helped to pull apart.”

- César Hidalgo
In recent years, collaborative filtering has become ubiquitous. Amazon and Apple’s iTunes have popularized recommendation services based on the idea that users that bought Snow White will also bought a copy of Cinderella. Based mainly on co-purchasing, these links do not carry information on people’s preferences. This is because co-purchasing links are symmetric and undirected, while preferences are asymmetric and directed.

This lack of preference information limits both the power of our current recommendation systems and our ability to use them for more forward-looking predictions. Preferences do exist and they vary among people and affect their decisions.

This project aims to create tools for extracting preferences out of populations by combining crowdsourcing techniques and large datasets that can be used to quantify the value of hard-to-measure features, such as the aesthetic value of buildings or preferences that emerge from changes in the design of industrial products.