A self-contained tool for creating local SQLite databases of Wikipedia articles with efficient lexical and semantic search capabilities.
$ curl wikilite.eja.it|sh
Launch CLI search? (Y/n):
1. lexical/es
2. lexical/it
3. lexical/sc
4. semantic/es.minilm
5. semantic/it.all-minilm
6. semantic/it.qwen3
7. semantic/sc.minilm
Choose a file by number or enter to exit: 7
Core functionalities that make Wikilite powerful and versatile.
Combines FTS5 lexical search for exact matching with semantic search using embeddings for conceptual similarity.
Full llama.cpp integration with GGUF models embedded directly in database files for standalone operation.
Compatible with Termux environment using provided packages for mobile usage.
Pre-built databases available for multiple languages including Italian, Spanish, German, and more.
Command-line interface for terminal usage and web interface for browser-based access.
Requires only the Wikilite executable and database file for complete offline operation.
Two powerful ways to find the information you need.
Utilizes SQLite FTS5 for efficient keyword-based searching with microsecond response times.
Implements ANN quantization and MRL with text embeddings to find semantically similar content.
Who can benefit from offline Wikipedia access?
Offline access for students and researchers in areas with limited internet.
Access content while traveling or working in areas with unreliable internet.
Search without sending queries to external servers, ensuring complete privacy.
Use semantic search to find related content when learning new languages.
Create specialized knowledge bases for institutions requiring offline access.
Utilize as a test dataset for NLP applications and search algorithm development.
Explore live demos or download and build from source.
Clone the repository and build from source.
$ git clone --recursive https://github.com/eja/wikilite.git
$ make
View on GitHub
Install the Wikilite app directly on your device via the latest APK release.
Download APKDownload ready-to-use databases for multiple languages from Hugging Face.
Databases with "lexical" support full-text search, while others include both lexical and semantic search.
Browse Databases