Administrative Legislation Public Opinion Automatic Classification

Information has multiplied at an unexpectedly high rate since an information society develops rapidly with the Internet as its main vehicle. In order to promptly locate what you want in a vast sea of information, we need to develop information processing technology, thus sorting out piles of documents automatically. In recent years, text categorization technology emerges as an important tool in the field of text and data processing and comes into widespread use thanks to scientists’ in-depth research. For the purpose of improving the work of soliciting opinions from the public on draft regulations and rules, we conduct a thorough investigation and study on the solicitation of public opinion on draft regulations and rules formulated by the Legislative Affair Office of the State Council, local governments and departments of the State Council; as well we deal with the issue of technical support to public participation in US. We believe that the public have got used to participating in administrative rule-making through rapid channels like the Internet, to offer hundreds of thousands of pieces of comments and suggestions even during a brief spell. To quickly and accurately give a semantic analysis of and automatically categorize public opinion can raise efficiency of legislators in sorting out public comments.Based on automatic categorization technology for foreign language texts and particularly, Chinese texts, and by taking into account the rules of categorizing public comments by hand in administrative rule-making, this paper designs a set of work patterns and technical combinations for semantic analysis and automatic categorization of public comments on draft regulations and rules. By exploring such work patterns and technical combinations in an empirical model, we conclude that legislators, facing flexible and variable demands for autonomic categorization of public comments, need to adopt an intellectual classification algorithm such as clustering and categorization, or adopt a rule-based classification approach such as a combination search. Accordingly, three types of automatic categorization, which is, clustering, categorization and rule-based classification, as well as their flow charts, are proposed in this paper to meet legislators’ multi-propose needs for classification of public opinion. Furthermore, recommendations for making improvements in soliciting public opinion are produced in this paper based on an analysis of experience.