Results show promising performance for structured fields such as title and author (up to 99% accuracy) but significantly lower accuracy for complex fields like subject and genre (below 50%). Overall, model performance does not yet meet the 95% quality threshold required for full automation.