Journal Vetting Assistant

This Python application helps research labs process and evaluate academic journals. It:

• Reads PDF files from a resources/ folder
• Summarizes each journal to reduce verbosity
• Splits text into chunks for embedding
• Prompts a language model (via LangChain) to recommend journals based on user-defined criteria

Note: Only PDF files are supported. Place your PDFs into the project/resources/ directory before running.

---

Features

• Extract raw text from PDFs using PyMuPDF
• Interactive metadata template creation via console prompts
• Text splitting into ~50-token chunks using LangChain's TokenTextSplitter
• Summarization of each journal to keep prompts concise
• Querying an OpenAI Chat model (GPT-4.1) through LangChain to recommend journals

---

Prerequisites

• Python 3.8 or newer
• A valid OpenAI API key

---

Installation

1. Clone this repository

   git clone <repo-url>
   cd <repo-folder>

2. Create a virtual environment (optional but recommended)

   python3 -m venv venv
   source venv/bin/activate   # on Windows: venv\\Scripts\\activate

3. Install dependencies

   pip install langchain openai python-dotenv pymupdf

---

Configuration

1. Environment Variables

   Create a .env file at the project root (adjacent to your script) with:

   OPENAI_API_KEY=your_openai_api_key_here
   

2. Resource Folder

   Place all your journal PDF files in:

project/project/resources/

   The script will automatically load every `.pdf` found there.

---

## Usage

Run the main script:

```bash
python run_journal_vetting.py

1. The app will read all PDFs in resources/.
2. It will summarize each journal.
3. You’ll be prompted to define metadata keys and values (e.g., field: computational modeling, impact_factor: >5).
4. Finally, the app will query GPT-4.1 to recommend the best-fit journal(s) and print the result.

---

Project Structure

project/
├── resources/       # Place your .pdf journal files here
├── run_journal_vetting.py  # Main application script
├── requirements.txt # (Optional) dependencies list
├── .env             # OpenAI API key
└── README.md        # This file

---

Customization

• Chunk size: Adjust chunk_size and chunk_overlap in createembeddings().
• Model settings: Modify model_name, temperature, and max_tokens in the ChatOpenAI constructor.
• Prompt templates: Tweak the wording in buildquery() or compressjournals() to fit your vetting criteria.

---

Contributing

Feel free to open issues or pull requests for additional features, bug fixes, or enhancements.

---