Thai-News-Vector-Dialogue

This project integrates news scraping, title translation, vector store knowledge base creation, and multi-turn dialogue using OpenAI's API. It allows you to:

• Scrape news from multiple RSS feeds (within the last 24 hours).
• Translate news titles into Traditional Chinese.
• Save the news data into a file (news_titles.txt).
• Upload the file to OpenAI's File API and build a vector store knowledge base.
• Engage in multi-turn conversations using the OpenAI responses API with context memory, leveraging the uploaded knowledge.

Features

• News Scraping & Translation:
  Fetches news from various RSS feeds and translates the titles into Traditional Chinese.

• Knowledge Base Creation:
  Saves the news data into news_titles.txt, then uploads it to create a vector store for semantic search.

• Multi-turn Dialogue with Context Memory:
  Uses the OpenAI responses API to search the knowledge base (via file_search tool) and supports context retention over multiple interactions.

Prerequisites

• Python 3.7 or later.
• Required Python packages:
  • feedparser
  • requests
  • openai
• A valid OpenAI API key with access to File API and vector store features.

Installation

1. Clone the repository:

   git clone https://github.com/your-username/News-Vector-Dialogue.git

2. Navigate to the project directory:

   cd News-Vector-Dialogue

3. Create a virtual environment and install dependencies:

   python -m venv env
   source env/bin/activate  # On Windows use: env\Scripts\activate
   pip install -r requirements.txt

Usage

1. Set Your API Key:
   In news_vector_chat.py, replace the placeholder API key with your actual OpenAI API key:

   client = OpenAI(api_key="sk-xxxxxxxxxxxxxxxx")

2. Run the script:

   python news_vector_chat.py

3. Interact with the Dialogue System:

   • The script will scrape and translate news, build the knowledge base, and then enter a multi-turn dialogue mode.
   • Type your questions at the prompt.
   • Type quit to exit the dialogue.

Project Structure

News-Vector-Dialogue/
├── news_vector_chat.py       # Main script integrating news scraping, translation, vector store upload, and dialogue.
├── news_titles.txt           # Generated file containing the scraped and translated news.
├── requirements.txt          # List of required Python packages.
└── README.md                 # Project documentation.

License

This project is licensed under the MIT License.

Contributing

Contributions are welcome! Please open an issue or submit a pull request for any improvements or bug fixes.