Unified and Scalable Incremental Recommenders with Consumed Item Packs

Abstract : Recommenders personalize the web content using collaborative filtering to relate users (or items). This work proposes to unify user-based, item-based and neural word embeddings types of recommenders under a single abstraction for their input, we name Consumed Item Packs (CIPs). In addition to genericity, we show this abstraction to be compatible with incremental processing, which is at the core of low latency recommendation to users. We propose three such algorithms using CIPs, analyze them, and describe their implementation and scalability for the Spark platform. We demonstrate that all three provide a recommendation quality that is competitive with three algorithms from the state-of-the-art.
Complete list of metadatas

Contributor : Erwan Le Merrer <>
Submitted on : Wednesday, June 12, 2019 - 10:58:06 AM
Last modification on : Friday, June 14, 2019 - 1:49:24 AM


Files produced by the author(s)


  • HAL Id : hal-02153388, version 1


Rachid Guerraoui, Erwan Le Merrer, Rhicheek Patra, Jean-Ronan Vigouroux. Unified and Scalable Incremental Recommenders with Consumed Item Packs. EURO-PAR 2019 - European Conference on Parallel Processing, Aug 2019, Gottingen, Germany. pp.1-14. ⟨hal-02153388⟩



Record views


Files downloads