PEPO: Petition Executing Processing Optimizer Based on Natural Language Processing

In this paper, we propose "Petition Executing Process Optimizer (PEPO)," an AI-based petition processing system that features three components, (a) Department Classification, (b) Importance Assessment, and (c) Response Generation for improving the Public Work Bureau (PWB) 1999 Hotline petitions handling process in Taiwan. Our Department Classification algorithm has been evaluated with NDCG, achieving an impressive score of 86.48%, while the Important Assessment function has an accuracy rate of 85%. Besides, Response Generation enhances communication efficiency between the government and citizens. The PEPO system has been deployed as an online web service for the Public Works Bureau of the Tainan City Government. With PEPO, the PWB benefits greatly from the effectiveness and efficiency of handling citizens' petitions.