融合需求与评价的科技服务平台机构推荐方法研究  

Research on Technology Service Platform Institution Recommendation Method Combining Demand and Evaluation

作　　者：徐尚英[1,2] 徐书情 陈冬林 吴天昊[1,2] Xu Shangying;Xu Shuqing;Chen Donglin;Wu Tianhao(School of Economics,Wuhan University of Technology,Wuhan 430070;Hubei Provincial Research Center for E-Business Big Data Engineering Technology,Wuhan 430070)

机构地区：[1]武汉理工大学经济学院,武汉430070 [2]湖北省电子商务大数据工程技术研究中心,武汉430070

出　　处：《情报杂志》2022年第12期140-148,共9页Journal of Intelligence

基　　金：国家重点研发计划项目“长江中游城市群综合科技服务集成与应用技术”(编号:2018YFB1404303);国家自然科学基金项目“面向最优时间窗推荐的网络客户行为轨迹预测研究”(编号:71702138);中央高校基本科研业务费专项资金资助项目“2021Ⅲ026GX基于大数据的平台经济精准监管模式研究”(编号:212415001)。

摘　　要：[研究目的]随着科技服务需求日趋多样化复杂化,科技服务机构大批涌现,为科技服务平台需方定位优质机构愈发困难,故提出机构推荐方法以解决平台供需双方匹配问题。[研究方法]基于平台发展提出科技服务机构推荐方法CDAE,通过构建科技服务需求模型并计算需求双层相似度后,辅以购买关联生成机构候选集合;再从科技服务平台评分和机构评价标签双维度构建科技服务机构推荐指数,对集合机构重排以返回机构推荐列表。[研究结论]基于真实数据集进行实验,从精准度、覆盖度与契合度多重维度,通过F1、NDCG、COV及ILS多个指标充分证明CDAE方法优于CF-D和CBF这2种对比方法且提高了推荐性能。[Research purpose]With the increasing diversification and complexity of the technology service demand,a large number of technology service institutions emerge,and it is increasingly difficult to locate high-quality institutions for the demanders of technology service platforms.Therefore,the institutions recommendation method is proposed to solve the matching problem between the suppliers and demanders of the platforms.[Research method]Based on the development of the platform,the recommendation method of technology service institutions CDAE is proposed.After constructing the demand model of technology service and calculating the double-level similarity of demand,the institution candidate set is generated by purchasing association.Then,the recommendation index of technology service institutions is constructed from the two dimensions of technology service platform score and institution evaluation label,and the institutions of the set are rearranged to return to the institution recommendation list.[Research conclusion]Based on the real data set,the experiment shows that the CDAE method is superior to CF-D and CBF and improves the recommendation performance from multiple dimensions of accuracy,coverage and fit,and through multiple indicators of F1,NDCG,COV and ILS.

关 键 词：科技服务需求 评价标签 科技服务机构推荐指数 机构推荐列表 

分 类 号：G354[文化科学—情报学]