This repository is a curated collection of resources on search technology, primarily focusing on e-commerce search. It serves as a comprehensive knowledge base for engineers, researchers, and product managers interested in understanding and improving search systems, covering everything from classic lexical search to advanced semantic and multimodal approaches.

How It Works

The project is structured as a vast, categorized list of links to articles, papers, tools, and case studies. It covers the entire search lifecycle, including query understanding, retrieval, ranking, relevance, user experience, and evaluation metrics. The content is organized thematically, allowing users to dive deep into specific areas like vector search, hybrid search, or search quality assurance.

Quick Start & Requirements

This is a curated list of resources, not a software project. No installation or specific requirements are needed to access the information.

Highlighted Details

• Extensive coverage of semantic search, including embeddings, vector retrieval architectures (bi-encoders, cross-encoders, ColBERT), and techniques like Matryoshka embeddings and SPLADE.
• Detailed sections on search quality assurance, metrics (NDCG, MRR), offline and online evaluation methods, and the use of LLMs as judges.
• Comprehensive exploration of search UX, including best practices from Baymard Institute and Nielsen Norman Group, faceted search, and handling "no results" scenarios.
• In-depth discussion of query understanding, including intent mapping, segmentation, expansion, and the role of context.

Maintenance & Community

This is a static collection of links, with the last update indicated by sandbox issues from June 2021. The primary contributor is frutik.

Licensing & Compatibility

The repository itself is licensed under the MIT License, but the linked resources have their own respective licenses.

Limitations & Caveats

The content is a snapshot of resources and may not reflect the absolute latest advancements in the rapidly evolving field of search technology. The "sandbox" issues suggest a lack of active maintenance or updates beyond mid-2021.