基于自组织映射网络与时间权重的电力企业大学资源快速精准检索方法研究  

作　　者：郝腾飞 郭建龙 刘晓 冯伟夏 熊山 

机构地区：[1]广东电网有限责任公司培训与评价中心,广州510520

出　　处：《科技创新与应用》2022年第25期121-126,共6页Technology Innovation and Application

基　　金：中国南方电网有限责任公司科技项目(GDKJQQ20161164(038700KK52160002))。

摘　　要：电力企业大学在线学习平台是培训电力员工的有效手段,为快速精准检索平台资源,该文提出一种基于自组织映射网络SOM(Self-Organizing Feature Map,SOM)和时间权重的资源快速精准检索方法。该方法首先利用SOM神经网络算法对用户特征集进行聚类,并根据聚类群组生成关联资源集群构建用户资源快速推荐库;然后根据用户浏览历史记录与本次搜索时间的远近对浏览资源赋予时间权重,根据所赋时间权重更新用户特征集并再聚类,获得再聚类群组生成的关联资源集群,进而得到接近用户期望的资源快速推荐库,从而实现对平台资源快速精准检索。通过电力企业大学在线学习平台对该方法进行验证。结果表明,该方法在采用指数型权重函数时,新老员工在电力企业大学在线学习平台资源搜索具有较高的正确率和快速性,可实现对平台资源的快速精准检索。The online learning platform of electric power enterprise university is an effective means to train electric power employees. In order to quickly and accurately retrieve platform resources, a fast and accurate resource retrieval method based on SOM and time weight is proposed. The method first uses the SOM neural network algorithm to cluster user feature sets, and generates related resource clusters based on the clustering groups and further construct a user resource quick recommendation library. Then, according to the distance of this search time of user’s browsing history resources, time weight is assigned to the browsing resources, and the user feature set is updated according to the assigned time weight and re-clustered, to obtain the associated resource cluster generated by the re-clustering group, further obtain a quick recommendation library of resources close to user expectations, so as to achieve rapid and accurate retrieval of platform resources. The verification results show that when the method proposed in this paper uses an exponential weight function,new and old employees have a higher accuracy and rapidity in searching resources on the online learning platform of electric power enterprise universities, and can realize fast and accurate retrieval of platform resources.

关 键 词：SOM 时间权重 资源 快速精确检索 个性化推荐 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]