WikidataChat

Retrieval Augmented Generation (RAG) with for answering question with the Wikidata REST API

Introduction

WikidataChat is an innovative tool designed to leverage the comprehensive knowledge base of Wikidata, transforming it into a user-friendly question-and-answer chat interface. It aims to provide the general public with validated and cited knowledge directly from Wikidata, reducing the chances of misinformation or "hallucinations" often associated with large language models (LLMs).

Features

WikidataChat boasts a unique textification pipeline with the following capabilities:

• Search and Download: Utilizes Google's Serapi search pipeline and Wikidata's REST API to fetch relevant JSON data.
• Textification: Converts Wikidata statements into string statements, preparing them for processing.
• RAG Pipeline: Merges Wikidata string statements with user-provided questions to generate informed and accurate answers through an LLM.
• User Interface: Offers a friendly UI that not only presents answers but also provides linked lists of Wikidata and Wikipedia URLs for further exploration.

Getting Started

Prerequisites

• Docker installed on your system or an active Python environment.

Docker Installation (Recommended)

1. Follow these instructions to install Docker on your system.

2. Deploy WikidataChat using Docker with the following commands:

DOCKER_BUILDKIT=1 docker build . -t wdchat

docker run  \
  --env HUGGING_FACE_HUB_TOKEN=$HUGGING_FACE_HUB_TOKEN \
  --env SERAPI_API_KEY=$SERAPI_API_KEY \
  --volume "$(pwd)/wikidatachat":/workspace/wikidatachat \
  --volume wdchat_cache:/root/.cache \
  --publish 8000:8000 \
  --rm \
  --interactive \
  --tty \
  --name wdchat \
  wdchat

This will deploy the UI to localhost:8000, allowing local access to the WikidataChat interface.

Usage

After installation, access the WikidataChat through your web browser by navigating to localhost:8000. If deployed on an internet-accessible server, the interface can be accessed from the respective web address, providing a seamless experience for asking questions and receiving answers.

The UI and Haystack functionality were developed with colleagues from Wikimedia Deutschland

• Haystack Pipeline: rti
• Vue3 UI: andrewtavis

Contributing

We welcome contributions from the community. Whether it's features, bug fixes, or documentation, here's how you can help:

1. Fork the repository and create a branch for your contribution. Use descriptive names like feature/streamlined_rag_pipeline for features or bug/localhost_is_blank for bug fixes.
2. Submit pull requests with your changes. Make sure your contributions are narrowly defined and clearly described.
3. Report issues or suggest features using clear and concise titles like feature_request/include_download_option.

Please adhere to the Wikimedia Community Universal Code of Conduct when contributing.

License

WikidataChat is open-source software licensed under the MIT License. You are free to use, modify, and distribute the software as you wish. We kindly ask for a citation to this repository if you use WikidataChat in your projects.

Contact

For questions, comments, or discussions, please open an issue on this GitHub repository. We are committed to fostering a welcoming and collaborative community.