GraphRAG: 8 способов укротить расширенный контекст у LLM

В 22% случаев онкологи не могут назначить лечение из-за рисков, связанных с хроническими заболеваниями. Сегодня разбираем кейс, в котором ИИ помогает врачам решать сложнейшие когнитивные задачи, связанные с лечением различных видов рака. Рассказываем про методологию GraphRAG, разбираем, как устроены и работают решения из кейса и проводим сравнительный анализ всех способов решить эту сложнейшую задачу. Привет, Хабр! Это Андрей Носов, AI-архитектор из Raft. Я проектирую и создаю системы, которые должны стоять годами — сегодня речь пойдёт именно о них. В этой статье по мотивам моего доклада на

https://habr.com/ru/companies/oleg-bunin/articles/984340/

#rag #graphrag #llm #medtech #медицина #Machine_Learning #Оптимизация_производительности #хранилища #PostgreSQL #базы_данных

Vault Intelligence 4.0: True #GraphRAG comes to #Obsidian #PKM

If you’ve ever felt like #AI plugins just "keyword search" your notes but miss the connections you’ve spent hours building, this update is for you.

This update to Vault Intelligence is moving it from a standard research assistant to a full Graph-Augmented engine. 🧵 1/2

This was fun to build!

Release notes at https://cybaea.github.io/obsidian-vault-intelligence/docs/Version%204.0%20release%20notes.html

Technical notes at https://cybaea.github.io/obsidian-vault-intelligence/devs/shadow-graph-technical.html

[MIT licence]

Vault Intelligence 4.0: True #GraphRAG comes to #Obsidian #PKM

If you’ve ever felt like #AI plugins just "keyword search" your notes but miss the connections you’ve spent hours building, this update is for you.

This update to Vault Intelligence is moving it from a standard research assistant to a full Graph-Augmented engine. 🧵 1/2

This was fun to build!

Release notes at https://cybaea.github.io/obsidian-vault-intelligence/docs/Version%204.0%20release%20notes.html

Technical notes at https://cybaea.github.io/obsidian-vault-intelligence/devs/shadow-graph-technical.html

[MIT licence]

Are you using @obsidian for #PKM? Ever wished you could ask questions like, "What do I know about X?" instead of hunting for keywords? Now you can. Vault Intelligence #plugin comes with #ontology management, #GraphRAG, search grounding to validate facts, and much more.

Try it at https://cybaea.github.io/obsidian-vault-intelligence/

(Yes, it's another AI tool. I like AI for "better search".)

This was also an interesting learning experience with #Obsidian, Typescript, Google #Gemini, #RAG and GraphRAG, and more.

Are you using @obsidian for #PKM? Ever wished you could ask questions like, "What do I know about X?" instead of hunting for keywords? Now you can. Vault Intelligence #plugin comes with #ontology management, #GraphRAG, search grounding to validate facts, and much more.

Try it at https://cybaea.github.io/obsidian-vault-intelligence/

(Yes, it's another AI tool. I like AI for "better search".)

This was also an interesting learning experience with #Obsidian, Typescript, Google #Gemini, #RAG and GraphRAG, and more.

GibRAM an in-memory ephemeral GraphRAG runtime for retrieval

https://github.com/gibram-io/gibram

#HackerNews #GibRAM #GraphRAG #in-memory #runtime #retrieval #ephemeral #technology

Гибридная нейросимвольная архитектура для превращения вероятностных ответов LLM в детерминированный код

Нам обещали, что ИИ заменит инженеров, а дали вежливых чат-ботов, которые галлюцинируют и путаются в зависимостях. Мы потеряли cтруктуру. В этой статье я реанимирую идеи Символистов 80-х и объединяю их с мощью современных LLM. Это история создания ZervGen - Фреймворка, который превращает ваш Obsidian в живой граф знаний и пишет работающий код для RPG через бесплатные API. P.S. Ровно два месяца назад, 11 ноября, я сделал первый коммит. Сегодня, 11 января, я показываю, к чему это привело. Совпадение? Не думаю. Вскрыть архитектуру

https://habr.com/ru/articles/984126/

#сезон_ии_в_разработке #ииагенты #llm #python #graphrag #нейросимвольный_ии #obsidian #автоматизация #zervgen #mcp

Yet another ontology and #GraphRAG project, specifically targeting local LLMs:
- Extracts entities and relations from engineering PDFs
- Builds a materialized knowledge graph (NetworkX)
- Prevents semantic drift using community-aware retrieval
- Supports hybrid retrieval (Graph + Vector / FAISS)
- Runs targeted LLM improve cycles for factual gaps
- Evaluates multiple RAG strategies
- Generates DOCX and PDF technical documentation automatically
https://github.com/tvlnsiva/graphrag-engineering-pdfs

Semantica is a semantic layer for knowledge engineering which promises to transform chaotic data into intelligent knowledge. A comprehensive OSS framework powering Knowledge Graph-Powered RAG (GraphRAG), AI Agents, Multi-Agent Systems, and AI applications with structured semantic knowledge.

100% Open Source • MIT Licensed • Production Ready • Community Driven https://github.com/Hawksight-AI/semantica
#GraphRAG #KnowledgeGraphs

🔧 GraphQLite – mở rộng SQLite hỗ trợ Cypher, cho phép lưu trữ và truy vấn đồ thị ngay trong một file DB. Kết hợp sqlite-vec, tạo stack RAG tích hợp: lưu thực thể, quan hệ, dùng PageRank, community detection. Thích hợp cho các dự án GraphRAG mà không cần Neo4j. Cài đặt: pip install graphqlite. #GraphQLite #SQLite #GraphRAG #AI #CSDL #MachineLearning #CôngNghệ #Vietnam

https://www.reddit.com/r/LocalLLaMA/comments/1q0si1a/graphqlite_embedded_graph_database_for_building/

Ok I DID end up building a minimum viable GraphRAG implementation...To be finalised (and a CLI tool 🤓). But here's how to do it in and how it works.

https://www.mostlylucid.net/blog/graphrag-minimum-viable-implementation

Uses a different approach with ONNX / BERT, BM25, TF-IDF THEN the LLM. Not as good but ALSO doesn't do HUNDREDS of LLM calls for each document.

So instead of paying TONS for OpenAI credits, do it on your laptop with Ollama.

#graphrag #rag #localllm #ollama

More because I was interested last night (someone put the entire life sciences corpus in GraphRAG) I decided to take a look an play with code so here ya go. Probably WON'T implement it on the live site (unless I figure out how to make it perform!) but it's a fascinating approach. A MASSIVE cross-internet GraphRAG seems like the end goal with this technology. THAT would be stunning! All LLM models beign able to access a whole internet corpus INSTANTLY!!
Hopefully like MANY MSR projects this won't just be used to hoover up patents and stifle competition...but it likely WILL.

GraphRAG: Why Vector Search Breaks Down at the Corpus Level
https://www.mostlylucid.net/blog/graphrag-knowledge-graphs-for-rag

#llm #rag #graphrag #csharp

Как спроектировать AI ассистента для поддержки и не слить бюджет. Часть 1: От идеи до выбора архитектуры

Каждая команда сейчас хочет заменить людей на AI. Но есть другой подход - усилить текущие возможности с помощью AI. Это может быть крутым драйвером для масштабирования и роста, а не поводом для увольнения. Покажу как спроектировать агента который решает проблемы и можно внедрять в продакшен. Буду рассказывать на примере юридической поддержки, но подход универсальный. Поговорим про RAG и GraphRAG, про развёртывание и выбор модели. Статья будет полезна как для больших компаний так и для маленьких - разница только в том где вы будете хранить и обрабатывать данные с LLM.

https://habr.com/ru/articles/975230/

#RAG #graphrag #ai #llmагент #ииассистент #автоматизация_поддержки

📣 3x 6-month #internships at Wimmics for M2 students! Work on #NeuroSymbolicAI at the intersection of #MachineLearning & #KnowledgeGraphs: analogical reasoning, KG tasks, structured memory, #RAG, #GraphRAG, #LLM. Co-supervision:
@fabien_gandon
& Ndeye-Emilie Mbengue.
📍 Where: Sophia Antipolis (France)
⏳ Duration: 6 months
Details below!

Advanced RAG: LongRAG, Self-RAG and GraphRAG Explained:
https://www.glukhov.org/post/2025/11/advanced-rag-variants-longrag-self-rag-graphrag/
#RAG #longrag #selfrag #graphrag #python #ai #llm

FalkorDB’s GraphRAG just outperformed traditional vector retrieval by 3.4× on structured benchmarks. By blending LLM reasoning with schema‑aware graph constraints, it delivers faster, more accurate results for hybrid data pipelines. Curious how hybrid approaches and metadata handling give this edge? Dive into the details. #FalkorDB #GraphRAG #LLM #VectorRetrieval

🔗 https://aidailypost.com/news/falkordb-graphrag-beats-vector-retrieval-34-structured-benchmarks

Graph Retrieval-Augmented Generation: A Survey. A bit heavy on the property graph and less on the RDF side of things, but fun reading for the connaisseurs out there. https://github.com/pengboci/GraphRAG-Survey #GraphRAG #KnowledgeGraphs

Last week we hosted #PyDataVenice #23 🎉

🕸️ We kicked off with a hands-on session on the basics of #GraphRAG
📈 Then we dived into exploring LLMs in #Finance

Huge thanks to everyone who made this evening possible! 🙌

🎦 If you missed the #meetup, no worries: you can watch the #livestream on #YouTube:
👉 https://www.youtube.com/watch?v=_1ApTJuY-68

💾 See you all on Thursday, December 18 #SaveTheDate! 💪

#PyDataVE #23 #Meetup #PyData - #AI #LLM #RAG

온톨로지 전문가가 사라진다: LLM이 바꾸는 지식 그래프의 미래

https://aisparkup.com/posts/5918

SuperMemory is just one example of a growing ecosystem of knowledge graph systems (Graphiti- Zep, Fast Graph RAG, TrustGraph...) Some in Python, some TypeScript with the added benefit of having graph visualization. Even in Rust and Go there is a growing list of open source graph-RAG.
https://github.com/supermemoryai #KnowledgeGraph #GraphRAG