Computational Surprise, Perceptual Surprise, and Personal Background in Text Understanding

The concept of surprise has special significance in information retrieval in attracting user attention and arousing curiosity. In this paper, we introduced two computational measures of calculating the amount of surprise contained in a piece of text, and validated with the perceived surprise by users with different background knowledge expertise. We utilized a crowdsourcing approach and a lab-based user study to reach a large amount of users. The implication could be used to propose or refine future computational approaches to better predict human feeling of surprise triggered by reading a body of text.