摘要
提出了一种基于规则和学习相结合的元数据分类存储的方法,该方法通过统计分析,提取对元数据分类影响较大的通用特征规则,对无法用规则分类的元数据,采用文本学习方法,将元数据记录看成由多个属性字段组成的文本,通过构造分类器实现分类。实验结果表明,采用元数据分类存储方法具有良好的检索性能。
This article proposes a metadata classification storage method based on the combination of rule and learning, extractes universal feature rules that affect metadata classification greatly through statistical analysis; for those metadata that cannot be classified using the rules, categorizes them through constructing classifier by adopting text learning method and regarding metadata record as a text that is composed of many attribute fields. According to experimental results, the system has good search performance when adopting metadata classification storage method.
出处
《计算机工程》
CAS
CSCD
北大核心
2007年第13期68-70,共3页
Computer Engineering
基金
国家自然科学基金资助项目"受限域汉语问答系统研究"(60663004)
