Subjective bias in abstractive summarization
Published in arXiv preprint, 2021
This paper investigates the presence and impact of subjective bias in abstractive summarization systems. We analyze how different summarization models introduce subjective perspectives and biases when generating summaries, and propose methods to identify and mitigate these biases.
The research contributes to understanding the challenges of maintaining objectivity in automatic summarization and provides insights into developing more balanced summarization systems.
Recommended citation: L Li, W Liu, M Litvak, N Vanetik, J Pei, Y Liu, S Qi. (2021). "Subjective bias in abstractive summarization." arXiv preprint arXiv:2106.10084.
Download Paper