Can User Behaviour Sequences Reflect Perceived Novelty

Serendipity is highly valued as a process for developing original solutions to problems and for innovation. However, it is difficult to capture and thus difficult to measure, but novelty is a key and critical indicator. In this work, we investigate the relationship between user behavioural actions and perceived novelty in the context of browsing. 180 participants completed an open-ended browsing task, while their behaviour actions were tracked. Each seven-action sequence was analysed with respect to the participant's perception of Novelty. Results showed that 6 of the 7 actions map to a sub-sequence that discriminates between high and low novelty. Notably, switching between exploration and immersion, and checking SERPs about the same request in-depth are indicative of highly perceived novelty. The results show that analysing behavioural action sequences leads to better prediction of novelty, and thus the potential for serendipity, than individual browsing actions.