RankGraph-2: Lifecycle Co-Design for Billion-Node Graph Learning in Recommendation
2026-06-18T08:52:20Z•f799da114d7afa471b30e19b0b2f19c273b10bd9352cde84d9ba555ea4d01c5b
AutoML for recommender systemsLensKit-AutoMLM-head rescalingMeta deploymentQuerit-RerankerRankGraph-2SAERecSHIFTSPLADEdense retrievalgraph learninginformation retrievalinterpretable intentslabel-efficient adaptation','model merging','Hugging Face modellanguage bias mitigationmultilingual IRpersonalized PageRankpopularity bias correctionrecommender systemsrerankerresidual quantizationsparse autoencodersparse retrievalsubsamplingtraining stability
What happened
Collection of recent IR/recSys papers (arXiv 18 Jun 2026) covering lifecycle co-design for billion-node graph retrieval (RankGraph-2, deployed at Meta), multilingual retrieval bias correction (SHIFT), stability fixes for learned sparse retrieval (MLM-head rescaling for SPLADE), automated recommender selection (LensKit-Auto), interpretable intent priors via sparse autoencoders (SAERec), compact multilingual rerankers and label-efficient adaptation (Querit-Reranker), hierarchical/metadata-guided RAG (SproutRAG, MCompassRAG), and theoretical/algorithmic work on compact reachability embeddings for
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_ir
- Record identifier
- f799da114d7afa471b30e19b0b2f19c273b10bd9352cde84d9ba555ea4d01c5b
- Enrichment time
- 2026-06-18T08:52:20Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.