摘要
国内针对社会稳定形势的分析工作主要集中在理论、体系、指标构建方面,定量分析研究较少。提出了基于网络敏感信息挖掘、精确语义匹配和量化分析的社会稳定形势监控技术框架。对互联网新闻文本中与社会环境、民族和谐、民生幸福相关的敏感信息进行挖掘,识别热点关键词以及由该词引导的敏感事件的变化趋势,构造敏感信息知识库;建立社会稳定理论模型和计算模型,利用社会调查和迭代反馈分析法习得模型参数,实现社会稳定形势的定量评估。基于该技术构造了原型系统,对新疆、西藏等边疆六省份社会稳定形势定量分析的平均准确率达到73.72%,具有一定决策参考价值。
Research on domestic social stability analysis mainly focuses on the construction of social stability theory, architecture and index, but little attention is paid to quantitative analysis. In this paper, we propose a social stability supervising framework based on sensitive Web information mining, semantic pattern matching and quantitative calculation. By analyzing the sensitive information about social environment, national harmony and happy index of people lives in natural language texts from Internet, and identifying hot keywords as well as the event trends led by the keywords, we construct a sensitive information knowledge base, and design a social stability index theoretic model and a quantitative calculation model to evaluate the social stability quantitatively. Parameters of the calculation model are deter- mined by employing social investigations and an iterative feedback learning method. A prototype system is built on the proposed framework and experiments are conducted in 6 frontier provinces, such as Xin-jiang and Tibet. The result of an average accuracy of 73.29% has reference value in decision-making to some extent.
出处
《计算机工程与科学》
CSCD
北大核心
2015年第6期1214-1220,共7页
Computer Engineering & Science
基金
国家自然科学青年基金资助项目(61309022)
陕西省自然科学基金资助项目(2013JQ8031)
武警工程大学军事应用研究项目(WJY201515)
关键词
敏感信息
社会稳定指数
网络文本挖掘
sensitive information
social stability index
web text mining