This repository showcases a proof-of-concept AI-assisted and rule-based validation tool for DITA (Darwin Information Typing Architecture) files. The goal is to demonstrate how machine learning can classify .dita files as compliant or non-compliant, while rule-based checks catch structural and syntactic issues.

When the FastAPI server is running on 127.0.0.1:8000, you should see something like this:

Uploading a compliant .dita file yields a higher compliance probability and no structural errors:

A non-compliant .dita file (missing id, invalid root element, etc.) shows a lower compliance score and a list of errors:

A typical structure for this repository might look like:

AI-Assisted-DITA/ ├── .venv/ # Python virtual environment ├── LICENSE # License file ├── README.md # This readme ├── backend/ │ ├── pycache/ │ ├── app.py # FastAPI routes, ML & rule-based checks │ ├── model.py # Model definitions │ ├── models/ # Model directory │ └── train_model.py # Script to train ML model ├── data/ │ ├── compliant/ # Example good DITA files │ └── non_compliant/ # Example bad DITA files ├── images/ │ ├── backend_running.png # Backend server screenshot │ ├── compliant_file.png # Compliant validation screenshot │ └── non_compliant_file.png # Non-compliant validation screenshot ├── index.html # HTML/CSS frontend ├── models/ │ ├── model.pkl # Trained classifier │ └── vectorizer.pkl # Text vectorizer ├── requirements.txt # Python dependencies ├── populate_data.py # Script to create DITA files └── test_script.py # Test script

Clone this repo:

git clone https://github.com/carlosrod723/AI-Assisted-DITA.git

Create and activate a Python virtual environment (recommended):

Install dependencies:

(Optional) Populate your DITA dataset if you want to see various examples

This script creates .dita files under data/compliant/ and data/non_compliant/.

If you need to retrain or update the logistic regression model:

Ensure .dita files exist in data/compliant/ and data/non_compliant/. Run:

This creates model.pkl and vectorizer.pkl in the models/ folder.

From the project root (or wherever app.py is accessible):

Start the FastAPI server: bash

Check the console output. You should see something like:

Open a browser at http://127.0.0.1:8000. You should see:

That means the backend is ready for file uploads at /validate.

Locate index.html in the project.

Open it in your browser (double-click or drag-drop).

If you see a CORS error, you may need to serve index.html via a local server or allow null origin in your CORS config. 4, Choose a .dita file.

Click "Validate DITA."

A compliance probability and any structural errors appear on screen.

Ensure your FastAPI server runs at 127.0.0.1:8000 to receive the file.