基于神经网络的个性化信息检索模型研究  

Research on Personalized Information Retrieval Model Based on Neural Network

作　　者：胡旷达 

机构地区：[1]沈阳航空航天大学知识工程研究中心,沈阳110136

出　　处：《现代计算机（中旬刊）》2016年第4期18-23,共6页Modern Computer

摘　　要：针对目前企业内部对个性化信息检索需求日益迫切的问题,提出一种基于神经网络的个性化信息检索模型。通过用户行为,神经网络模型可以自动学习用户的兴趣特征,建立用户检索词与用户兴趣之间的联系,得到个性化检索结果。最后,通过计算用户兴趣与文档的相似度,并与BM25模型打分值进行线性加和,重新排序文档。实验结果表明,该方法从效果上优于其他同类个性化检索方法。Now, the need of the personalized information retrieval is very urgent in company. Aiming at this problem, proposes a personalized information retrieval model based on neural networks. By using user behavior, the neural network models can learn users interest feature automatically, establish links between users search terms and users interest, and get personalized information retrieval result. Finally, calcu-lates the similarity of user interest and documents. And the linear weighed method is used to combine the similarity and BM25 model.Then uses the score to sort the documents. Experimental results show that the proposed method is superior to others methods.

关 键 词：个性化信息检索 神经网络 用户行为 

分 类 号：TP393.4[自动化与计算机技术—计算机应用技术]