A recommendation system is an information filtering system used to predict users' ratings or preferences for items. It can find out the connections that will eventually be made between the user and the item. Everything in the world is connected. The social connection between people has promoted the birth of social products. The consumption connection between people and commodities has created countless e-commerce products. More and more reading connections between people and information promote the birth of social products. This led to the emergence of various information flow products.
The course catalog of the recommendation system is:
The first stage: Understanding the recommendation system
- week1: machine learning basics
- week2: Recommendation system basics
- Project exercise: Implement a text recommendation system based on user collaboration
second stage:
- week 3: Content portrait and user portrait
- week 4: User portrait
The third stage: build a multi-channel recall + ranking recommendation system
- week 5; traditional match method
- week 6: Deep match method
- week 7: Classic ranking method
The fourth stage:
- Week 8: Graph Embedding Family and User Behavior Construction Diagram
- Week 9: Graph recommendation with the introduction of sideinfo information, graph recommendation based on reasoning
- Week 10: Deep Ranking Model
- Exercise: DeepFM implementation
- Project actual combat: implement text recommendation system based on graph + depth ranking
The fifth stage:
- Week 11: Real-time interaction and cold start in the industry news recommendation system
- Week 12: Hot Articles Real-time Recall Strategy
- Week 13: ESMM with multiple goals and multiple interests of users
- Week 14: Reinforcement Learning and Recommendation System
Four major projects:
- Portrait-based recommendation system
- News recommendation items
- News recommendation items combined with graph structure
- Build a real-time recall recommendation system
The first phase of the recommendation system of Greedy Academy has ended. It feels very good and in-depth. You need +VX (daydayit) and note: Greedy recommendation system
