Bayesian active summarization

Bayesian Active Learning has had significant impact to various NLP problems, but nevertheless it’s application text summarization been explored very little. We introduce Summarization (BAS), as a method of combining active learning methods with state-of-the-art models. Our findings suggest that BAS achieves better and more robust performance, compared random selection, particularly for small data annotation budgets. More specifically, applying model like PEGASUS we managed reach 95% the performance fully trained model, using less than 150 training samples. Furthermore, have reduced standard deviation by 18% conventional selection strategy. Using showcase it is possible leverage large models effectively solve real-world problems limited annotated data.