Taohua Social Distribution Platform Recommendation System: Practice from 0 to 1
Taohua is Alibaba's social e-commerce platform, and I am responsible for the recommendation system for building the platform from 0 to 1.This project gave me an in-depth understanding of the characteristics and challenges of the social e-commerce recommendation system.
# Project Background
Taohua is a social distribution platform where users can earn commissions by sharing items.This model puts new demands on the recommendation system: not only to recommend products that users like, but also to consider distribution effects and social relationships.
# System Architecture
We have built a complete recommendation system architecture:
# # 1. Burial Point Data
First, a perfect burial site system was established to collect user behavior data.This includes browsing, clicking, sharing, purchasing, and social relationship data.
# # 2. Real-time behavioral features engineering
A real-time feature engineering system has been built, which can calculate user characteristics, commodity characteristics, scene characteristics, etc. in real time.This is especially important for social commerce, as user behavior changes rapidly.
# # 3. Product modeling
A product portrait system was established, including product attributes, price, sales volume, evaluation, etc.At the same time, the distribution properties and social communication capabilities of the goods are considered.
# # 4. User Modeling
A user portrait system has been established, including user attributes, purchase preferences, social relationships, etc.Particular attention is paid to the distribution capabilities and social influence of users.