摘要
利用地理空间描述模型中的相关概念扩展自然语言中空间语义角色,通过空间语义角色标注、短语识别以及句法模式分析达到识别非受限文本中深层空间语义的目的。实验表明,该方法具有较好的准确率、召回率与通常的信息提取性能相当。
To make a translation from space descriptions to graphics, it is necessary to extract deep spatial semantics from unrestricted text. This paper presents an approach to recognize deep spatial semantics from unrestricted text by means of defining spatial semantic roles based on the space description model in geography information system （GIS）. It discusses the spatial semantic dictionary construction, the spatial semantic role annotation, the recognition grammar and the whole extracting flow. The primary experiment shows that the recognition has a good precision and a similar recall compared to common information extraction systems.
出处
《计算机工程》
EI
CAS
CSCD
北大核心
2006年第4期36-38,共3页
Computer Engineering
基金
国家"973"计划基金资助项目(G2000077906)
