Upload
others
View
12
Download
0
Embed Size (px)
Citation preview
Utilizing Marginal Net Utility for Recommendation in
E-commerce Jian Wang & Yi Zhang
Informa(on Retrieval and Knowledge Management Lab University of California, Santa Cruz
1
• Will you like it? • Tradi(onal recommender systems recommend item with the highest predicted ra(ng
2 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Recommenda(on based on Predicted Ra(ng is Not Good for Consumers
• Will you purchase it? • Diminishing of return
• A ra(onal user will purchase the product with the highest marginal net u(lity
Total Utility, Marginal Utility
3 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Total U(lity
Marginal U(lity
Purchase Quan(ty
Total Utility, Marginal Utility
4 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Total U(lity
Marginal U(lity
Purchase Quan(ty
Total Utility, Marginal Utility
5 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Total U(lity
Marginal U(lity
Purchase Quan(ty
Total Utility, Marginal Utility
6 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Total U(lity
Marginal U(lity
Purchase Quan(ty
Marginal Net Utility
7 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Marginal Net Utility
= Marginal Utility - Price
1. Model the consumer behavior based on marginal net u6lity
2. Make recommenda6ons to maximize the marginal net u6lity for each user
8 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Making Consumers Happier
Outline
• Mo(va(on • Algorithm
– Basic economics: two exis(ng u(lity func(ons studied by economists
– Problem seOng
– Propose a new u(lity func(on tailored for recommender systems
• Experimental Results
• Conclusion 9 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Basic Economics: Linear Utility Function
• does not capture diminishing of return characteristic • most existing algorithms are implicitly based on this utility function
10
• Total utility • purchase history • product ’s basic utility • purchase quantity of product in • marginal utility for the addition purchase of
Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Basic Economics: Cobb-Douglas Utility
• diminishing of return rate is fixed • basic utility is not personalized
11 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
• Total utility • purchase history • product ’s basic utility • purchase quantity of product in • marginal utility for the addition purchase of
Making Consumers Happier: Recommendation based on Marginal Net Utility
12 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Design • Design the u6lity func6onal form
Learn • Learn the u6lity func6on parameters from user history
Predict • Predict the marginal net u6lity of a product for a user using the func6on learned and user history
Rank • Rank products based on predicted marginal net u6lity
Problem SeOng
13 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
day 1 day 3 day 7 day 10
Time t
Design the U(lity Func(onal Form
14 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
• Total utility • purchase history • product ’s basic utility • purchase quantity of product in • marginal utility for the addition purchase of
Cobb-Douglas
New
15 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Design the U(lity Func(onal Form
• only same product will influence the marginal utility
New framework to revamp exis6ng recommenda6on algorithms
Revamp Exis(ng Algorithms
Marginal net utility
16 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Take Singular Value Decomposition (SVD) as an example
Learn the U(lity Func(on Parameters
Maximum A Posteriori Estimation
17 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Joint Likelihood
Purchase Likelihood Conditional on the marginal net utility
Outline
• Mo(va(on • Algorithm
• Experimental Results
• Conclusion
18 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Experiment Dataset
• More than 5-years purchase history from – 2004-01-01 to 2009-03-08
• 10,399 users, 65,551 products, and 102,915 unique (user, product) pairs
• 80% training, 10% validation, and 10% testing
19 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Experiment Design
• Methods to compare - - - same product - similar product
• Evaluation Metric
20 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
General Analysis
21 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
1 2 3 4 5
Conversion
Rate
K
conversion rate@K
87.39%
Further Analysis
• Consider all products – 13.79% repurchase – 90.64% new purchase
• Evaluate orders with the specific purchase type (repurchase or new purchase)
22 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Further Analysis
• The new u(lity func(on achieves significantly beber performance in both tasks
• Examples of Repurchases – diaper, pet food, etc. – Tend to be consumable products
• Examples of New purchases – computer, cell phone, bed frame, etc – Tend to be durable product – law of diminishing marginal u(lity
23 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
Outline
• Mo(va(on • Algorithm
• Experimental Results
• Conclusion
24 Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
25
Conclusion
Jian Wang & Yi Zhang Utilizing Marginal Net Utility for Recommendation in E-commerce
• Introduce a new framework for recommender system in e-commerce sites
• Recommend products with the highest marginal net utility
• Take SVD as an example to revamp
• Achieve significant improvement in the conversion rate
Utilizing Marginal Net Utility for Recommendation in
E-commerce Jian Wang & Yi Zhang
Informa(on Retrieval and Knowledge Management Lab University of California, Santa Cruz
26
Ques(ons?