摘要
对象识别是数据集成的一个重要问题,针对学术领域的对象集成问题,提出一个基于上下文环境的对象识别模型。利用作者名字的上下文环境,包括合作者、国际会议、论文时间、论文标题4维信息对作者进行对象识别。通过计算两个表象每一维信息的相似程度,采用感知器模型对于少量的专家标注的学习用例进行学习从而获得每一维合适的权重以及对应的阈值,最后利用构造的模型进行准确预测。实验结果表明该模型具有较高的可用性。
Object recognition is an important problem in data integration with uncertain.To integrate academic objects,the authors propose an object recognition model basd on the context.The authors use an appearance's context information,including co-author,international conference, publication time, paper title, to recognize it.The authors measure the similarity of four dimensions information in an appearance context and use sensors to learn the parameters by a few of examples labeled by field experts.According to the model,the authors can recognize academic objects accurately.
出处
《计算机工程与应用》
CSCD
北大核心
2008年第23期139-142,150,共5页
Computer Engineering and Applications
基金
国家自然科学基金(No.60703007)~~