Title:	USING NATURAL LANGUAGE PROCESSING TO DETECT URBAN PROBLEMS FOR GOVERNANCE
Author(s):	Jing Li, Chunhua Cao, Yue Luo, Xiaobo He, Xian Liu, Yahui Jia, Xianfeng Jin, Shu Wang, Jiying Run, Xin Su, Yuxuan Zhong, Qiuhan Zhao, Zifu Wang and Megan Rice
ISBN:	978-989-8704-24-5
Editors:	Hans Weghorn
Year:	2020
Edition:	Single
Keywords:	Urban Informatics, Deep Learning, Machine Learning, Big Data
Type:	Full
First Page:	37
Last Page:	45
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	Urban informatics has become an increasingly important domain in the past few years. At the same time, the volume and variety of urban data resources has become increasingly diversified and larger. These sources include the millions of emails that serve as an invaluable communication mechanism between citizens and municipal managers. The responses to these emails could benefit from urban analyses, such as through location and tracking. By analyzing key information such as the subject content and spatial location, we can discover whether any major problems in a certain city or any long-term problems have been resolved. However, processing such large number of emails manually is tedious — for example, a secretary can only process approximately 100 emails per day. This paper uses deep learning, machine learning, and natural language processing to mine and analyze government-related emails in a semi-automated process. In this method, the email content is first cut and digitized by natural language processing technology. Then, the digitized text content is modeled by different algorithms through machine learning and deep learning technology—the accuracy of each algorithm is compared with one another. Finally, the emails can be automatically classified into a variety of topics, automatically analyzing the problems in the city. This proposed process greatly improves the efficiency of analyzing city information which could ultimately help governments optimize their resources and support their decision making.
	Go Back