Use of a text mining method for classifying citizen report data and analyzing the occurrence trend of local problems

Abstract

An important task of any municipality is the maintenance and improvement of the street-related living environment and traffic safety for citizens.  For this, their department of street maintenance is expected to efficiently perform the maintenance and inspection of streets according to priority with limited human and budgetary resources.  Recently, municipalities in various countries are adopting “the citizen report system,” which is a system of reporting problems of streets, such as damaged streets, by citizens to their municipality, for citizens to perform part of street maintenance and inspection.  It is possible that the data obtained by municipalities through the citizen report system can be utilized not only for early problem detection but also for prioritizing administrative measures by using it for analyzing the occurrence trend of problems.  Problems reported by citizens, however, are classified by different methods from municipality to municipality, and thus the collection and comparative analysis of such data across municipalities is difficult.  This study presents a method of commonly classifying such data, regardless of different classification standards, by analyzing the contents of citizen reports by using text mining.  We then analyze the relationship between the trend of citizen reports and the occurrence trend of problems concerning the living environment and traffic safety, using the citizen report data of three large municipalities classified by this method, and infer the occurrence trend of problems.  This study has confirmed that citizen report data possibly contributes to municipalities’ prioritization of the maintenance and improvement of the living environment and traffic safety.