ScaR: Real-Time Recommender Framework

Project Overview (March 2015 – December 2020, ongoing occasionally)

At Know Center Research, I have extensively contributed to ScaR (Scalable Recommender), our in-house recommender framework designed for real-time, scalable, and context-aware recommendation scenarios. ScaR follows a microservices architecture, integrating seamlessly with streaming data to provide immediate recommendations without costly recalculations. More information about ScaR is available on the official project website.

Key Projects Using ScaR

• Master’s thesis on social-based recommendation algorithms
• Student job matchmaking in collaboration with Studo
• Matchmaking on the VHDD platform
• Personalized learning content recommendations in the Cogsteps project
• Adaptive conference session recommendations with Conference Navigator

Technical Challenges

• Immediate processing of high-frequency streaming data
• Scalability in cloud-based and distributed environments
• Integration of diverse recommendation algorithms
• Real-time updates without performance degradation

Technologies & Methods

• Core Development: Java, Apache Solr, Spring Boot, Microservices Architecture
• Data Management & Processing: Real-time data ingestion and handling using Apache Solr’s near-real-time features
• Continuous Integration & Deployment: Jenkins, Maven, Docker, Apache ZooKeeper
• Recommendation Algorithms: Content-Based Filtering, Collaborative Filtering, Hybrid Approaches
• Testing & Maintenance: JUnit, comprehensive unit, and integration testing

Results & Impact

• Enabled multiple successful commercial and research projects
• Improved recommendation accuracy and speed across various use cases
• Featured in many academic publications, with the main reference:

@inproceedings{lacic2015scar,
  title={Scar: Towards a real-time recommender framework following the microservices architecture},
  author={Lacic, Emanuel and Traub, Matthias and Kowald, Dominik and Lex, Elisabeth},
  booktitle={Proceedings of the Workshop on Large Scale Recommender Systems (LSRS2015) at RecSys},
  pages={16--20},
  year={2015}
}

Personal Contribution

• Acted as a core contributor, significantly influencing the framework’s features and development
• Implemented and tested various recommendation algorithms within ScaR
• Integrated ScaR into multiple industry and academic projects
• Assisted with maintenance, deployment, and infrastructure improvements for enhanced scalability and reliability