OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

Simgrag

Simgrag is a Python framework for knowledge-graph-driven retrieval-augmented generation. It embeds graph nodes and relations, retrieves similar subgraphs through vector search, and uses an LLM for tasks including question answering and fact verification.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Simgrag — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Simgrag when you need a codebase for graph-grounded retrieval and generation, and can prepare the model, embedding and vector-database services it depends on.

Who it is for

  • Researchers exploring knowledge-graph retrieval-augmented generation.
  • Developers building graph-based question-answering or fact-verification pipelines.

Documented uses

  • Index a supported knowledge graph for similarity-based subgraph retrieval.
  • Run question-answering queries over MetaQA using supplied one-, two- and three-hop scripts.
  • Run fact-verification queries over FactKG and inspect the configured result file.
Before you choose

Requirements & limits

The project is provided as a public GitHub repository with Python implementation code.

  • The pipeline requires an LLM, node-and-relation embedding model and vector database.
  • The example stack is Ollama with Llama 3 70B, Nomic embeddings and Milvus.
  • Alternative components are described as replaceable when their APIs are prepared, but the repository does not document a tested alternative integration.
  • The supplied data preparation and scripts are for MetaQA and FactKG; other datasets require changes.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Prepare the documented Simgrag components

Set up the example language-model, embedding and vector-database components that the pipeline requires.

Before you start

  • A local environment for installing Ollama and Milvus.
  • The Simgrag repository.
  • The Nomic embedding model from Hugging Face.

Steps

  1. Install Ollama, then run ollama run llama3:70b.

  2. Start the Simgrag-required Ollama service with the command below.

    Documented command
    bash ollama_server.sh

  3. Clone nomic-embed-text-v1 into data/raw using the README's directory and Git commands.

  4. Install Milvus and start the service needed by Simgrag by following Milvus documentation.

Expected outcome

The three documented component services and model files are prepared for use by the pipeline.

Guide limitations

  • This guide covers component preparation, not indexing or running queries.
  • The README also requires a dataset and configuration for the subsequent pipeline stages.
Documentation 1 ↗
Source-based practical guide

Run the documented MetaQA or FactKG pipeline

Index a documented dataset and run its supplied query scripts to produce result files.

Before you start

  • The Simgrag repository and a Python environment.
  • The language model, embedding model and vector database prepared for Simgrag.
  • MetaQA or FactKG downloaded into data/raw and organized as described in the README.
  • Configuration files set for the intended run.

Steps

  1. Prepare MetaQA or FactKG in data/raw and review the configs files.

  2. From pipeline, run the command below for MetaQA or the command below for FactKG.

    Documented command
    python metaQA_index.py
    Documented command
    python factKG_index.py

  3. For MetaQA, run the command below, the command below and the command below; for FactKG, run the command below.

    Documented command
    python metaQA_query1hop.py
    Documented command
    python metaQA_query2hop.py
    Documented command
    python metaQA_query3hop.py
    Documented command
    python factKG_query.py

  4. Read the result file assigned to output_filename in the configuration.

Expected outcome

The configured result file contains query results; FactKG result lines include a `correct` key for the final answer.

Guide limitations

  • The documented scripts cover MetaQA and FactKG; another dataset requires implementation of data mappings and pipeline adjustments.
  • The repository instructions do not establish results for other component integrations or datasets.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Retrieval-augmented applications

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
SimgragIndex a supported knowledge graph for similarity-based subgraph retrieval.CLI, Self-hostedThe project is provided as a public GitHub repository with Python implementation code.
LangsearchCrawl website or local data for a knowledge base.CLI, Self-hosted, Server
Hybrid Search RagCombine document retrieval and language-model generation in a configurable workflow.
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
YZ-Cai/SimGRAG · Official homepage title · retained in its source language
How we review sourcesSuggest a correction