Combining Retrieval-Augmented Text Generation with LLMs for Reading Content Recommendations

Collection of recent 2026 arXiv papers on retrieval, LLM-augmented generation, web data extraction, recommender systems, and federated graph methods. Highlights include: a RAG+LLM pipeline for generating personalized reading content with improved groundedness; Co-Scraper, a query-aware DOM pruning system that synthesizes reusable scrapers (fine-tuned Qwen3-8B) achieving high reuse and extraction F1; a systems survey of Retrieval-as-a-Service (RaaS) in industrial pipelines; methods that incorporate implicit negative behaviors for sequential user modeling; federated graph recommendation guidedby