diff --git a/README.md b/README.md index 6b836df..347ff81 100644 --- a/README.md +++ b/README.md @@ -1,13 +1,34 @@ # NewsAskAI -šŸš§ **Work in Progress**: -NewsAskAI scrapes the latest news and generates a list of the most relevant articles. The project leverages Retrieval-Augmented Generation (RAG) to allow users to ask questions about the news and receive context-rich, real-time answers. It is powered by open-source Hugging Face embeddings and Chroma DB for efficient storage and retrieval. +
-## Future Steps + + +
+ +**NewsAskAI** is an open-source project designed to scrape the latest news on a user-specified topic and generate a curated list of the most relevant articles. It leverages Retrieval-Augmented Generation (RAG) to enable users to ask questions about the news and receive context-rich, real-time answers. The project is powered by open-source Hugging Face embeddings, the Phi-3.5 Large Language Model, and Chroma DB for efficient storage and retrieval of embeddings. + +Additionally, it features a user interface built with Textual, replicating the experience of a conversational chat. + +## Example + +
+ + + +
+ +## šŸš§ **Work in Progress: Future Steps** + +- **Download Full Articles**: Implement functionality to download the entire article found based on the specified topic. + - *Note: Currently, the system only retrieves the titles and abstracts of articles.* +- **Parametrization**: Parameterize the application, allowing users to switch between different configuration settings easily. +- **Multiple Topics**: Enable ingestion of multiple topics simultaneously or allow users to explore top news across various categories (e.g., sports, technology, politics) without requiring a new ingestion process for each category. +- **Optimize the inference speed:** Improve the system's performance to ensure faster and more efficient real-time responses to user queries. + - *Note: The current performance is too slow.* +- **Topic Modeling / Named Entity Recognition**: Add advanced features, such as a second processing stage, to label or tag articles for more efficient query filtering or to perform reranking of the embeddings. + - *Example: Allow users to filter results by entities (e.g., ā€œFilter by entity: ā€˜Doramas Baezā€™ā€) or specific topics.* -- **Download Full Articles**: Implement functionality to download the entire article found based on the specified topic. -- **Multiple Topics**: Enable ingestion of multiple topics or allow users to explore top news across categories (e.g., sports, technology, politics) without requiring a fresh ingestion process. -- **Topic Modeling / Named Entity Recognition**: Add advanced features, such as a second processing stage to label or tag articles for more efficient query filtering (e.g., ā€œFilter by entity: ā€˜Elon Muskā€™ā€). ## How to Use It diff --git a/src/news_ask_ai/static/assets/News_AI_Logo.png b/src/news_ask_ai/static/assets/News_AI_Logo.png new file mode 100644 index 0000000..ffa031b Binary files /dev/null and b/src/news_ask_ai/static/assets/News_AI_Logo.png differ diff --git a/src/news_ask_ai/static/assets/usage_example.gif b/src/news_ask_ai/static/assets/usage_example.gif new file mode 100644 index 0000000..1f95467 Binary files /dev/null and b/src/news_ask_ai/static/assets/usage_example.gif differ