Recommendation systems are systems that recommend products or services to people online. YouTube, Amazon, and other companies advertise these products on their websites and try to pinpoint a user’s taste from what they choose. These companies do not get a chance to verify whether their recommendation systems work, however, because a user will rarely rank all options; there are also potentially a large number of options that may change over time.This project will attempt to fill this lack of knowledge by gathering opinions and testing recommendation systems.
Question to be answered: "How are recommendation systems designed and which is the most accurate (for simple/small datasets)"?
The project is part research and part experiment. The project will involve researching and learning about recommendation systems. The second part will show differences in recommendation systems and lead to a conclusion on the most accurate recommendation system for certain data. The final product will be a two-part report on (1) the research and current methods surrounding recommendation systems and (2) how different models compare on complete data.
So far, I have created the basic outline of current topics surrounding recommendation systems, a list of theoretical topics that the systems are modeled after, and a list of a few resources or relevant papers.
Aside from ultimately benefiting the companies that utilize recommendation systems, the theory and mathematics behind these systems advance our understanding of other topics. The same theory might be used for traffic monitoring and computer vision, and other applications may be revealed in the future.
Project Team:
- Leah W.