基于自然语言处理的地理要素变化信息抽取研究

Research on Change Information Extraction of Geographic Elements Based on Natural Language Processing

传统的地理要素更新通过遥感解译的方法进行,存在无法实时监测、难以获取属性信息等问题。本文组合多种自然语言处理技术,以依存句法分析为核心,通过地理要素特征词汇匹配确定变化主体、动词、时间、地点、属性等信息。以交通建设新闻和行政区划变更通告为测试语料,验证了模型算法的有效性。研究减轻了信息抽取对知识库的依赖,变化信息抽取的准确率能达到80%以上,研究成果可用于地理要素实时变化发现和信息采集,提高数据生产效率。

The traditional updating method of geographical elements is based on remote sensing interpretation,which can not be monitored in real time and it is difficult to obtain attribute information.In this paper,we combined a variety of natural language proc essing technologies,with dependency parsing as the core,determined the changed subject,verb,time,place,attribute and so on.The effectiveness of the model algorithm is verified by testing corpus from traffic construction news and administrative division change notice.The research reduced the dependence of information extraction on knowledge base and the accuracy rate of change information extraction can reach more than 80%.The research results can be used for real-time change detection and information collection of geographical elements and improve the data production efficiency.