Tag
NVIDIA released Auto Ontology, an open-source (Apache 2.0) library that builds a trusted context/ontology layer from enterprise metadata, reconciles metric definitions across systems, and exposes it to agent harnesses via MCP, CLI, and API for grounded, validated agentic analytics and NL-to-SQL querying powered by NVIDIA NIM.
Lin Yuxiang (GitHub: Lum1104), a graduate student, saw his side project Understand-Anything jump from 655 stars to 56,400, and it was taken over by Egonex and officially released. It's becoming the default code-understanding plugin for mainstream AI coding tools like Claude Code, Cursor, and Copilot, turning any codebase into an interactive knowledge graph using Tree-sitter and LLM agents.
SciGraphSelect introduces a reference-grounded framework for auditable scientific data selection, organizing domain knowledge and reasoning operations into typed knowledge–reasoning graphs to score candidate documents with traceable evidence.
A personal Android notes app built with Hark Pro using Kotlin and Jetpack Compose, featuring an AI thinking partner with a connection engine, idea development modes, knowledge graph, and meaning-based search over notes.
The author compiled and shared 10 open-source AI education projects they have been following recently, covering interactive labs for primary and secondary schools in China, secondary school science curricula abroad, and AI visualization learning tools, arguing that AI will reinvent education.
A viral post highlights that Andrej Karpathy has been at Anthropic since May, pointing to two public resources — Anthropic's knowledge-graph cookbook and Karpathy's autonomous autoresearch loop that ran 700 experiments in 2 days — that can be combined to build a self-directing agent system.
Shared a ready-to-copy installation prompt for installing Graphify (safishamsi/graphify) in the Codex environment. It generates a knowledge graph for the project to reduce token usage from repeatedly reading files and tracing dependencies, claiming to save approximately 80% of tokens.
An AI-powered open-source "smart mistake notebook" project featuring photo-based problem recognition, automatic answers, knowledge-point tags and difficulty ratings, mistake management, practice drills with related problems, and one-click test paper printing — designed to pair with the K-12 knowledge graph open-source project.
The article introduces a toolchain for design workflows centered on Claude Code, with a focus on Graphify—which can turn any folder into a knowledge graph. Combined with tools such as Figma MCP, Playwright MCP, and impeccable, it helps AI-generated web pages break free from cookie-cutter designs.
LangChain highlights an experiment pairing openwiki with a lightweight evidence graph, allowing AI agents to trace generated knowledge back to its supporting evidence for more reliable, domain-specific responses.
The author updated the open-source K-12 knowledge graph project cn-math-taxonomy, adding three difficulty tiers (Basic/Advanced/Extension) of questions for each knowledge point, and improving the adaptive diagnostic order and prerequisite backtracking logic. The project covers math, Chinese, science, English, and physics, and provides an online diagnostic tool, though the author notes it is an AI-generated draft that still awaits teacher review.
FinLLM-KG 是一个模型无关的框架，将因果 LLM 与金融知识图谱结合，从 SEC 10-K 文件中注入公司级、时间对齐的知识以改进金融情绪分析与算法交易，同时提供可解释的预测依据。
The paper introduces InsightAgents, a framework that distills insights from graphs of related papers to equip LLM debate agents with literature-grounded expertise for collaborative research idea generation, outperforming three baselines across 100 NLP topics.
GEMS-RAG 是一个面向工程标准文档的图驱动多模态检索增强生成框架，通过层级解析、视觉实体抽取、多模态知识图谱构建与图驱动混合检索，在 MUTCD 数据集上相较文本、多模态和图基 RAG 基线提升了检索与问答效果。
该论文提出一种知识引导的LLM框架，结合语法约束语义解析与知识图谱驱动的澄清循环，用于集成地面与非地面6G网络中的意图驱动服务编排，在200条意图数据集上达到89%分类准确率与95%策略有效性。
A developer is asking the community how they actually solve agent memory in production, whether through vector DBs with RAG, knowledge graphs, or human-in-the-loop approval for a shared knowledge base.
This paper proposes a multi-granularity interaction network for relational triple extraction that jointly leverages word-word, entity-relation, and head-tail interactions to better handle unseen entities, achieving F1 improvements of 3.1%/20.5% on the NYT-R and WebNLG-R datasets in the Unseen metric.
该论文提出了研究进度图（Research Progress Graph, RPG），一种将研究构想的迭代演化记录为带时序标注的认知状态与推理转移的表示方法，并发现现有35个研究智能体均只是RPG的不完整实现，主张研究智能体应从生成静态构想转向显式建模构想演化轨迹。
Marble has fully open-sourced its internal elementary school knowledge graph, os-taxonomy, containing 1,591 fine-grained knowledge points and 3,225 prerequisite dependencies forming a directed acyclic graph, covering 8 subjects and aligned with curriculum standards such as NGSS and Common Core. The license permits both research and commercial use. The project has quickly garnered 5,881 stars and 986 forks since launch, providing education products with a ready-to-reuse foundation of prerequisite relationships.
Netflix 介绍了其在超大规模基础设施上构建端到端知识图谱（本体驱动的可观测性）的实践，通过统一 MELT（指标、事件、日志、追踪）数据层并显式记录系统实体间关系，将典型故障根因排查时间从约 4 小时显著缩短。