摘要
采用地理编码和汉语切分词相结合的方法对空间命名实体进行在线识别。通过全文粗扫描获取相关的地理编码来锁定文中涉及的空间范围,然后按照一定的策略在确定的空间范围内进行单句识别,从而显著地减少了词典的加载量,较好地解决了因命名实体词典数量庞大而导致的低效率问题。实验表明,该方法能有效降低内存的消耗量和识别时间,提高识别精度,基本满足空间命名实体在线识别和位置信息服务的要求。
The paper presents a method of spatial named entity(SNE) recognition for on-line location-based services(LBS). This paper discusses the establishment of spatial semantic lexicon combined Chinese word segmentation with geocoding, the match algorithm, the extraction of spatial semantic rules and SNE disambiguation. The experiment shows this algorithm can reduce computer memory consumption and system response time, and the recognition precision is also improved. It can meet the requirement of on-line LBS on the whole.
出处
《计算机工程》
EI
CAS
CSCD
北大核心
2005年第20期49-50,53,共3页
Computer Engineering
基金
国家"973"计划基金资助项目(G2000077906)
国家"863"计划基金资助项目(2001AA135090)
关键词
空间命名实体识别
自然语言处理
地理编码
位置服务
Spatial named entity recognition
Natural language processing
Geocoding
Location-based services (LBS)