As a society of low birthrates and aging populations, capitals and large cities around the world are becoming more concentrated and complex problems are occurring accordingly. As more people flock to the city, the amount of complaints has increased, but there is not enough manpower to respond to them.

Kim Jang-hyun, a professor of interaction science at our university proposed an algorithm to automatically classify 160,000 civil texts through machine learning from 2006 to 2017 and published in CITIES (SSCI, JCR 2019 IF= 4.802, Top 2). Based on Word2vec and Random forest, artificial intelligence can automatically classify complaints, including transportation, environment, and culture, with an accuracy of about 70%. Inefficient administrative procedures, which had to be classified by existing civil servants in charge of civil complaints, can be quickly and accurately transformed into efficient civil complaints through machine learning.

In addition, we propose a method to analyze automatically complaints using dynamic topic modeling to predict complaints in the future.

Finally, the data analysis process was disclosed in github and a book (Urban Data Standards Analysis Model: Civil Petitions Analysis) so that each local government can help build a smart city system in the future.

You can find the book that manualized the paper and analysis process through the website below.

https://www.sciencedirect.com/science/article/pii/S0264275120312890#f0005

https://github.com/SeoulDigitalFoundation/VoiceOfSeoul_AnalysisGuide

https://sdf.seoul.kr/research-report/1241

Image 1. Subway seat care for elderly and preganant women

Image 2. Emphasis on fine dust and energy saving issues