LIFE-STAGE PREDICTION FOR PRODUCT RECOMMENDATION IN E-COMMERCE Speaker: Jim-An Tsai Advisor: Jia-ling Koh 1 Author: Peng Jiang, Yadong Zhu, Yi Zhang, Quan Yuan Date: 2015/9/24 Source: KDD’15 OUTLINE  Introduction  Method  Experiment  Conclusion 2 INTRODUCTION 3 INTRODUCTION 4 FRAMEWORK User Predict for User’s life-stage Generate Recommedation One-kid family MESMM Multi-kids family GMM 5 OUTLINE  Introduction  Method  Experiment  Conclusion 6 NOTATIONS 7 MAXIMUM ENTROPY SEMI MARKOV MODEL (MESMM) 8 MESMM 9 FEATURES FOR THE CLASSIFIER  1. Category features 10 FEATURES FOR THE CLASSIFIER 2. Query  A user behavior sequence may contain user product search activities. User search queries can directly reflect users’ requirements, which can indirectly reflect a baby’s age.  3. Product property features  Taobao is a distributed market space where sellers sell products. Many sellers provide meta data about the products.  11 FEATURES FOR THE CLASSIFIER 4. Product title features  Product titles are created by sellers who are not affiliated with Taobao. Sellers are very creative about product titles. As a result, many product titles can be very informative about the life stage of the consumer.   5. Temporal Effect of Features 12 FEATURES FOR THE CLASSIFIER 13 RECOMMENDATION BASED ON LIFE-STAGE  To make product recommendations, we propose a model to estimate the probability of a user purchasing a product at a specific age a. P(p_productj) be the probability of the user purchasing the product j P(a|p_productj) be the conditional probability of making x is a feature vector that represents a purchasing action, the w is purchase age a given that the userofwill product j. learnt by at maximizing the likelihood thepurchase training data. 14 MIXTURE MODEL FOR MULTI-KIDS  Gaussian Mixture Model (GMM): 15 MIXTURE MODEL FOR MULTI-KIDS (b) Gaussian mixture model with two components for user purchasing age. One component represents 11-month old, and the other one represents 40-month old. 16 OUTLINE  Introduction  Method  Experiment  Conclusion 17 EXPERIMENTAL SETUP 18 LIFE-STAGE PREDICTION AND SEGMENTATION 19 LIFE-STAGE PREDICTION AND SEGMENTATION 20 ANALYSIS OF TEMPORAL EFFECT 21 ANALYSIS OF MULTI-KIDS 22 ONLINE EXPERIMENTS 23 ONLINE EXPERIMENTS 24 OUTLINE  Introduction  Method  Experiment  Conclusion 25 CONCLUSION    1. We introduced the conception of life stages into Ecommerce recommendation systems 2. We proposed a new Maximum Entropy Semi Markov model for stochastic life stage segmentation and prediction 3. We developed a practical efficient large scale industry solution for life stage segmentation and labeling in mum-baby domain, where the life stage transition is deterministic 26 CONCLUSION    4. We proposed a probability model which can explicitly incorporate the life-stage information into an Ecommerce recommender system. 5. We proposed a solution for modeling multi-kids scenario via Gaussian mixture models. 6. We verified of the effectiveness of the life-stage based approach in both offline and online scenarios. 27 THANKS FOR LISTENING 28