This project determined the possible benefits of using an online personalized filtering system for paper greeting cards at Sparks.com. A test recommender system was developed in Java that implemented content-based and collaborative filtering algorithms. Additionally, a combination of these filtering algorithms was created to see if this approach gave more accurate results. Using information extracted from their purchase history database, the system was tested against other methods for displaying cards that do not utilize personalized filtering. We found that using content-based, collaborative or top 10 approach will result in highly accurate predictions compared to the random ordering of cards currently done at Sparks.com.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Zebrowski, Zachary Albert
• Sao Pedro, Michael A.
• Bhojnagarwala, Sharad Y.

Publisher

• Worcester Polytechnic Institute

Identifier

• 00D029M

Advisor

• Claypool, Mark L.

Year

• 2000

Center

• Silicon Valley, California Project Center - MQP

Sponsor

• Sparks.com

Date created

• 2000-01-01

Resource type

• Major Qualifying Project

Major

• Computer Science

Rights statement

• In Copyright