A Method and Algorithm to Dynamically Learn Heterogeneous Preferences with Clustering Algorithms

Technology Description

UCSD inventors from the Rady School of Management have come up with a system and method to dynamically learn consumer preferences and recommend products by sampling from clusters of existing consumer choice data. The invention exploits knowledge from large amount of existing data generated by consumer’s interactions with websites to enhance the learning process, which could greatly improve the effectiveness of prediction accuracy. The invention:

• can be utilized in ecommerce product recommendations, website content placement, and app optimization on mobile devices.
• can more accurately predict the type of the consumer and the probability of purchase.
• is currently at the computer model simulation stage using experimental data.

Patent Status