CTR-Sink: New Framework Improves Language Models for Click-Through Rate Prediction in Recommendation Systems
Researchers have proposed CTR-Sink, a framework that introduces behavior-level attention sinks into language models to improve click-through rate (CTR) prediction in recommendation systems. The work addresses a structural mismatch between user behavior sequences—discrete actions separated by semantically empty tokens—and the coherent natural language on which language models are pre-trained. Better CTR prediction directly affects the relevance of content surfaced by recommendation platforms used by millions of people.
CTR-Sink is a novel framework designed to close a structural gap that arises when language models (LMs) are applied to click-through rate prediction tasks. User behavior sequences in recommendation systems consist of discrete actions joined by semantically empty separators, which differ fundamentally from the coherent natural language LMs are trained on; this mismatch causes attention to scatter across irrelevant tokens, a phenomenon the authors call semantic fragmentation. To address this, CTR-Sink inserts dedicated sink tokens between consecutive user behaviors and incorporates recommendation-specific signals—such as temporal distance between actions—to serve as stable attention anchors. A two-stage training strategy explicitly guides LM attention toward these sink tokens, while an attention sink mechanism amplifies inter-sink dependencies to better capture behavioral correlations. The approach was validated on one industrial dataset and two open-source benchmarks, MovieLens and Kuairec, with visualization results supporting the method's effectiveness. The paper has been submitted to and linked with an ACM publication (DOI: 10.1145/3770855.3817646), suggesting it has undergone additional peer review beyond the arXiv preprint stage.
What's missing
Computational cost and latency implications of inserting sink tokens and the two-stage training strategy relative to standard LM-based CTR approaches are not discussed in the available summary. Generalizability beyond the three tested datasets—particularly to domains outside video and movie recommendation—remains an open question.
What different sources said
- arXiv cs.CLCenter
CTR-Sink: Attention Sink for Language Models in Click-Through Rate Prediction
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.