WEAVESS
WEAVESS is an open-source benchmarking and evaluation framework for graph-based approximate nearest neighbor (ANN) search methods, providing code and experiments for large-scale vector similarity search. It is useful for researchers and practitioners comparing vector indexing algorithms for vector databases and AI search applications.
About this tool
title: WEAVESS slug: weavess category: benchmarks-evaluation brand: lsyhprum source_url: https://github.com/Lsyhprum/WEAVESS featured: false images:
- https://opengraph.githubassets.com/1/Lsyhprum/WEAVESS tags:
- ann
- benchmark
- similarity-search
Description
WEAVESS is an open-source benchmarking and evaluation framework for graph-based approximate nearest neighbor (ANN) search methods. It provides the code, datasets, optimal parameters, and experimental setup used in the paper “A Comprehensive Survey and Experimental Comparison of Graph-based Approximate Nearest Neighbor Search”. The framework is aimed at researchers and practitioners who need fair, reproducible comparisons of vector indexing algorithms for large-scale vector similarity search, vector databases, and AI search applications.
Features
-
Comprehensive experimental framework
- Implements the full experimental pipeline used in the referenced survey paper.
- Provides code, datasets, and detailed experimental configurations.
- Ensures fair comparison by reimplementing all evaluated algorithms under a unified design pattern, programming language (with a noted exception for the IEH hash table), and common experimental setup.
-
Implemented / evaluated algorithms
- Supports benchmarking of thirteen representative graph-based ANNS algorithms:
- KGraph
- FANNG
- NSG (Navigating Spreading-out Graph)
- NSSG
- DPG
- Vamana
- EFANNA
- IEH
- NSW (Navigable Small World)
- HNSW (Hierarchical NSW)
- NGT-panng
- NGT-onng
- SPTAG-KDT
- SPTAG-BKT
- HCNNG
- KDRG
- Includes links or references to original papers and, where available, original code repositories for each algorithm.
- Supports benchmarking of thirteen representative graph-based ANNS algorithms:
-
Standardized implementation details
- Algorithms reimplemented in a consistent way to reduce implementation bias in comparisons.
- Uses the same set of implementation “tricks” and optimizations across methods where applicable.
-
Real-world datasets
- Includes eight real-world ANN datasets commonly used in prior work.
- Each dataset is pre-split into:
- Base data (indexing corpus)
- Query data (evaluation queries)
- Provides ground truth neighbor data for evaluation, in the form of top-20 or top-100 nearest neighbors.
-
Synthetic datasets for scalability studies
- Provides twelve synthetic datasets to analyze:
- Scalability of each algorithm with respect to dataset size.
- Performance under varying data distributions and characteristics.
- Provides twelve synthetic datasets to analyze:
-
Evaluation focus
- Designed for large-scale vector similarity search and ANN benchmarking.
- Suitable for evaluating indexing methods used in vector databases and AI search systems.
Use Cases
- Research on graph-based ANN algorithms and their comparative performance.
- Reproducing or extending the experiments from the WEAVESS survey paper.
- Benchmarking new ANN algorithms against a standardized suite of baselines and datasets.
- Evaluating vector indexing strategies for production vector databases or search applications.
Pricing
- Open-source: No pricing information is provided; the project is available as open-source software via GitHub.
Loading more......
Information
Categories
Tags
Similar Products
6 result(s)An annual competition focused on similarity search and indexing algorithms, including approximate nearest neighbor methods and high-dimensional vector indexing, providing benchmarks and results relevant to vector database research.
ANN-Benchmarks is a benchmarking platform specifically for evaluating the performance of approximate nearest neighbor (ANN) search algorithms, which are foundational to vector database evaluation and comparison.
A 2024 paper introducing CANDY, a benchmark for continuous ANN search with a focus on dynamic data ingestion, crucial for next-generation vector databases.
Ruby gem for approximate nearest neighbor search that can integrate with pgvector and other backends to power vector similarity search in Ruby applications.
AiSAQ is an all-in-storage approximate nearest neighbor search system that uses product quantization to enable DRAM-free vector similarity search, serving as a specialized vector search/indexing approach for large-scale information retrieval.
This work by Jingfan Meng is a comprehensive research thesis on efficient locality-sensitive hashing (LSH), covering algorithmic solutions, core primitives, and applications for approximate nearest neighbor search. It is relevant to vector databases because LSH-based indexing is a foundational technique for scalable similarity search over high-dimensional vectors, informing the design of vector indexes, retrieval engines, and similarity search modules in modern vector database systems.