摘要
专家发现是实体检索领域的一个研究热点,针对经典专家发现模型存在索引术语独立性假设与检索性能低的缺陷,提出一种基于贝叶斯网络模型的专家发现方法。该方法模型采用四层网络结构,能够实现图形化的概率推理,同时运用词向量技术能够实现查询术语的语义扩展。实验结果显示该模型在多个评价指标上均优于经典专家发现模型,能够有效实现查询术语语义扩展,提高专家检索性能。
Expert finding is a research hotspot in the field of entity retrieval.Aiming at the shortcomings of the classical expert discovery model,such as the assumption of indexing term independence and the low retrieval performance,and an expert discovery method of Bayesian network with query semantic extension is proposed.The model adopts four-layer network structure,which can realize graphical probabilistic inference,and the semantic extension of query terms can be realized by word vector technology.Experimental results show that the new model is superior to the classical expert discovery model in terms of multiple evaluation indexes,indicating that the new model can effectively extend the semantics of query terms and improve the performance of expert retrieval.
作者
郑伟
侯宏旭
班志杰
ZHENG Wei;HOU Hongxu;BAN Zhijie(College of Computer Science,Inner Mongolia University,Hohhot 010021,China;College of Science,Hebei North University,Zhangjiakou,Hebei 075000,China)
出处
《计算机工程与应用》
CSCD
北大核心
2020年第13期194-198,共5页
Computer Engineering and Applications
基金
国家自然科学基金(No.616662053)
内蒙古自然科学基金(No.2018MS06005)
河北省社会科学基金(No.HB18XW004)。