融媒体信息推荐模型构建与信息推荐方法研究  

Construction of Information Recommendation Model and Method for Information Recommendation Media Convergence

作　　者：崔金栋[1] 陈思远 CUI Jin-dong;CHEN Si-yuan(School of Economic and Management,Northeast Electric Power University,Jilin 132012,China)

机构地区：[1]东北电力大学经济管理学院,吉林吉林132000

出　　处：《情报科学》2020年第7期52-58,共7页Information Science

基　　金：国家社会科学基金项目“基于信息生态的微博信息管理机理研究”(16BTQ068)。

摘　　要：【目的/意义】为解决融媒体平台海量资源主题提取难、信息推荐精度低问题,笔者利用大数据和NLP技术对现有推荐方法加以优化,改进后的模型在精准度和匹配度方面得到显著提升。【方法/过程】首先利用MapReduce对融媒体资源数据进行过滤处理,再借助NLP技术对资源数据提取主题特征、同时对用户进行需求特征描述,然后将二者进行匹配以确定推荐内容,最后使用对比分析法检验本研究推荐方法的可行性。【结果/结论】与传统方法相比,本研究方法所推荐内容更符合用户需求,其推荐精确度和匹配度也有显著提高。推荐方法评价结果显示,随着推荐列表数量变动,曲线会有一定波动,其波动原因有待深入研究与分析。【Purpose/significance】Using big data and NLP technology to extract themes of rich and diverse resources of the media,and to describe the images according to user needs,can effectively improve the accuracy and matching of the recommendations of the media.【Method/process】Firstly,MapReduce is used to filter the data of the media platform.Then,the NLP technology is used to extract the topic features of the resource data,describe the requirements of the user,and match the two to determine the recommended content.Finally,comparative analysis is used to test the feasibility of the recommended method.【Result/conclusio】Compared with traditional methods,the content recommended by this research method is more in line with user needs,and its recommendation accuracy and matching degree have also been significantly improved.The evaluation results of recommendation methods show that as the number of recommendation lists changes,the curve will fluctuate to some extent,and the reasons for its fluctuations need to be further studied and analyzed.

关 键 词：融媒体 信息推荐模型 大数据 自然语言处理技术 

分 类 号：G254.9[文化科学—图书馆学]