异构系统中实体识别研究

Research on Entity Identification in Heterogeneous Systems

已有的实体识别技术可以根据名称、地址进行摘词、排序,从而进行匹配,文章就不同商业企业中多个系统之间的数据集成时的数据匹配工作提出了一种基于线性相关性分析的实体识别方法。在多个系统均存在主数据和与主数据相关的记录项时,通过对记录项的平滑处理后,计算不同系统数据之间的线性相关性,对计算出来的相关性值进行分析,从而得到匹配关系的建议解。最后对相同实体识别研究方向进行了展望。

The existing way of entity identification can make the mapping based on name, address and the key words subtracted from them, this paper raises a way of identify raw data by using the correlation analysis technology. When there are both master data and detail records in different systems, with the exponential smoothing of the detailed records, we could calculate the lineal correlation value, with analyzing on those results, and give the suggested mapping relationship. The future research directions are discussed.