摘要
以往数字图书信息分级推荐方法受到干扰信息影响,导致推荐效果较差,为了避免以往方法带来的弊端,提出了基于多层级决策分析的分级推荐方法。采用多维度建模方法,构建数字图书信息特征模型,针对不同图书特征要素在模型中的重要性,分析词与词之间语义关系,实现基于特征模型的词分级扩展。为排除信息干扰,引入词性标注层,通过中文分词工具进行词性标注,以便得到完善相似词集。在该集合中,计算多层级决策分析相似度,结合用户对项目测评结果设计分级推荐方案,构建数字图书信息项目本体,依据五元组形式对相似度进行排序,将相似度高的项目推荐给用户。对数字图书信息进行分级保密,避免外界信息干扰,从而保证推荐方案安全。通过实验对比结果可知,该方法最高推荐精准度可达到93%,为用户高效使用数字图书信息提供可行性方案。
In order to avoid the drawbacks of previous methods,a hierarchical recommendation method based on multi-level decision analysis is proposed. Using multi-dimensional modeling method,this paper constructs a digital book information feature model. According to the importance of different book feature elements in the model,the semantic relationship between words is analyzed,and the word hierarchical expansion based on the feature model is realized. In order to eliminate information interference,part-of-speech tagging layer is introduced,and part-of-speech tagging is carried out through Chinese word segmentation tools in order to improve the similar word set. In this set,the similarity of multi-level decision analysis is calculated,the hierarchical recommendation scheme is designed based on the user’s evaluation results,and the digital book information project ontology is constructed. The similarity is sorted according to the five-tuple form,and the items with high similarity are recommended to users. The digital book information is classified and kept secret to avoid interference from outside information,so as to ensure the safety of the recommendation scheme. The experimental results show that the highest recommendation accuracy of this method can reach 93%,which provides a feasible scheme for users to use digital book information efficiently.
作者
张新吉
ZHANG Xin-ji(Xi'an Peihua University,Xi'an 710125,China)
出处
《电子设计工程》
2019年第13期180-184,共5页
Electronic Design Engineering
基金
陕西省教育厅服务地方专项计划项目(16JK1328)
关键词
多层级决策分析
数字图书信息
分级推荐
多维度
分级扩展
multi-level decision analysis
digital book information
hierarchical recommendation
multi-dimensional
hierarchical expansion
