基于增强学习与主成分提取的资源信息分析技术

Resource information analysis based on reinforcement learning and principal component extraction

针对从海量数据中难以获取有效信息的问题,设计了一种基于增强学习和主成分分析的信息推荐系统模型。该模型通过分析用户的行为偏好并从中挑选相关特征,再利用核主成分分析法进行降维,同时使用基于增强学习的用户特定深度Q学习方法,根据其状态来获得最优策略,以实现对用户兴趣数据的精准推荐。在电力人力资源数据集上进行的实验表明,所设计模型对于人力资源数据的推荐精准率可达到93.3%,召回率和F1值则分别为90.7%及91.2%,与其他推荐算法模型相比,该模型的综合性能较为理想,能够实现海量数据的精准信息提取与分析推荐。

To solve the problem that it is difficult to obtain effective information from massive data,this paper designs an information recommendation system model based on reinforcement learning and principal component analysis.This model analyzes user behavior preferences and selects relevant features from them,and then uses kernel principal component analysis to reduce dimensions,and then uses user specific depth Q learning method based on reinforcement learning to obtain optimal strategies according to user status,so as to achieve accurate recommendation of user interest data.The experiment on the electric power human resource data set shows that the recommended accuracy of the designed model for human resource data can reach 93.3%,the recall rate and F1 are 90.7%and 91.2%respectively.Compared with other recommendation algorithm models,the comprehensive performance is relatively ideal,and it can achieve accurate information extraction,analysis and recommendation of massive data.