#rag
Rag: 18 AI articles covering rag news, analysis, and research
Articles
Tables in PDFs for RAG: Don’t Flatten the Grid⭐9
Stop losing critical data in PDFs: Learn how structured table extraction—not flattening—boosts RAG retrieval accuracy.
A RAG That Says 'Not in This Document' Has to Show Four Kinds of Evidence⭐9
A four-part RAG system must prove a "not found" answer with evidence from retrieval, generation, grounding, and verification to avoid misleading users.
FAQ as RAG: When You Get to Design the Corpus⭐10
FAQ as RAG rethinks retrieval when the corpus is Q&A pairs: parsing becomes trivial, retrieval works as a cache, and few-shot prompting turns into a lookup prob...
Noisy Text in RAG: Typos, OCR, and the Gap Classical Spell-Check Leaves⭐10
How typos, OCR errors, and transcription noise break RAG retrieval—and why classical spell-check falls short, leaving embeddings to bridge the gap.
RAG Is Not the Whole Toolkit: The NLP Techniques Real Problems Still Need⭐8
RAG is powerful but not the only answer; six cheaper NLP methods—exact match, spelling fixes, keyword search, and more—solve most problems faster and with expla...
One Document Type, a Million Files: Structured Extraction into the SQL Table RAG Queries⭐9
Learn how to structure five thousand identical contracts into SQL table RAG queries with repeatable extraction, smart indexing, and cost arithmetic.
Building an End-to-End Document Intelligence Pipeline with deepDoctection⭐9
Build an end-to-end document intelligence pipeline with deepDoctection: layout detection, table parsing, OCR, reading order, and structured export for RAG syste...
Retrieve One Row from a Table, Not the Whole Table: Row-Level Chunks for RAG⭐10
Retrieving one table row—not the whole table—as the chunk for RAG answers precise questions like policy caps without distracting the model with unrelated rows.
Three Kinds of RAG Corpus, and What It Costs to Build for the Wrong One⭐9
Discover the three RAG corpus types—unrelated reports, duplicate contracts, and claim folders—and what building for the wrong one costs.
Kimi K3's 1M Token Context Window vs. RAG: Cost, Latency and Answer Quality⭐9
Comparing Kimi K3’s 1M token context to RAG across cost, latency, and answer quality in a blind test.