ACQUA is looking at the discretised version of linguistic features of each candidate answer and predicts which answer is going to be marked as "accepted". Past knowledge such as user reputation or future knowledge, such as score of the answers is not taken into account. Hence, ACQUA can predict which answer is going to get accepted in real-time settings with minimum resources.
ACQUA makes use of the StackExchange API, fetches all answers and analyses them. Our web service is then highlighting one answer indicating it as the "accepted".
You can then visit any Stack Exchange website. A bordermark will highlight the best answer.
You can read more details about the methodology in the following paper:
G. Gkotsis, K. Stepanyan, C. Pedrinaci, J. Domingue, and M. Liakata. It's all in the Content: State of the art best answer prediction based on discretisation of shallow linguistic features. In Proceedings of the 2014 ACM Conference on Web Science, WebSci '14, pages 202–210, New York, NY, USA, 2014. ACM.