SAP Community Call: Elevating User Experiences with AI-Powered Personalized Recommendation Services

Introduction

• Host: Menina Chow, SAP Community Team
• Guests:
  • Benjamin Tan, Full-stack Machine Learning Developer
  • Stephen Fu, Product Owner in SAP Artificial Intelligence
• Purpose: Discuss AI-powered personalized recommendation services

Topics Covered

Overview of SAP AI and Intelligent Enterprise

• Intelligent Enterprise: Utilizes AI, ML, and robotic automation for real-time business actions.
• SAP Business Technology Platform (BTP):
  • Combines best practices from 30+ industries
  • Offers flexibility and rapid time to value
  • Hosts AI business services

Personalized Recommendation Services

• Objective: Provide recommendations from a long list of items based on user context.
• Challenges:
  • Handling new users and items (costar scenarios)
  • Adapting to shifting user habits and trends
  • Delivering timely and accurate recommendations

Machine Learning Models

• Integration: Models are integrated with SAP Commerce Cloud and available publicly.
• Scenarios:
  • Next click/view/purchase predictions
  • Affinity recommendations
  • Smart search using NLP
• General Use Cases: Retail, HR learning platforms, content delivery

Features of Personalized Recommendation Services

• Personalized and Alternative Recommendations: Quick and relevant suggestions.
• Explainability: Provides confidence scores for predictions and attribute contributions.
• Customizable Strategy:
  • Supports coastal scenarios with minimal data
  • Allows attribute boosting for marketing strategies
• Deployment: Easy integration with SAP Cloud Platform through APIs

Demos

E-commerce Scenario

• Purpose: Demonstrate personalized recommendations in action.
• Features:
  • Trending products for new users
  • Similar item recommendations
  • Contextual data enhancement

Merchandiser Perspective

• Tools: Interface to manage search results and recommendation strategies.
• Features:
  • Boosting items based on tags (e.g., Star Wars week)
  • Explainability of recommendations

Additional UI Demo

• Data Set: Movie catalog
• Demonstrated Features:
  • Boosting strategies
  • ML explainability with confidence scores
  • Handling of costar items and users

Questions and Answers

• Public Availability: Service available with a free tier since May.
• Getting Started: Blog series and tutorials available for setup.
• Minimum Data Threshold: 100 items with 100 valid clickstream interactions each

Closing

• Feedback and Interaction Encouraged: Community questions and discussions
• Resources: Links to blog posts and community platform
• Next Steps: Try the service via SAP BTP, engage with community content.

Use these notes as a reference for understanding how SAP's personalized recommendation services work, their features, and how they can be applied across different industries.