本網(wǎng)訊（通訊員 唐思思）近日，由美國電氣電子工程師學(xué)會（the Institute of Electrical and Electronics Engineers,，IEEE）主辦的大數(shù)據(jù)分析國際會議（International Conference on Big Data Analysis,，IEEE ICBDA 2016）在杭州召開。我校信息管理與工程學(xué)院張文宇教授應(yīng)邀以“Fueling Made-in-China with Internet+ Plan”為主題在大會上作主旨報告,。
此次會議涵蓋了大數(shù)據(jù)模型與算法,、數(shù)據(jù)架構(gòu)、數(shù)據(jù)管理,、保存與隱私保護(hù),、數(shù)據(jù)挖掘應(yīng)用、企業(yè)政府社會實踐等多個主題,，旨在交流大數(shù)據(jù)領(lǐng)域的最新成果和實踐經(jīng)驗,。來自美國、德國,、南非及我國等的百余名專家學(xué)者參加了會議,。
附張文宇教授主旨報告摘要：
Made-in-China is facing a distributed manufacturing environment which requires unprecedented levels of interoperability to integrate distributed manufacturing resources across organizational boundaries to support cross-enterprise collaboration. The recently emerged Internet+ plan has been utilized as a promising approach to integrate mobile Internet, cloud computing, big data and the Internet of Things with modern manufacturing to address the above issue. Manufacturing enterprises have been putting their efforts to virtualize their core competencies or advantageous resources as manufacturing services and publish, discover and share them on the open Internet for developing Industry 4.0 applications. However, the large size, dynamic nature and heterogeneous expression of distributed manufacturing resources brings forth a serious challenge in scalability and efficiency. This trend demands intelligent and robust models to address information overload in order to enable efficient discovery of manufacturing services. As an illustrative example of Industry 4.0 service technology, we present a personalized manufacturing service recommendation approach, which combines a PageRank-based reputation model and a collaborative filtering technique in a unified framework for recommending the right manufacturing services to an active service user for supply chain deployment. The novel aspect of this research is adapting the PageRank algorithm to a network of service-oriented multi-echelon supply chain in order to determine both user reputation and service reputation. In addition, it explores the use of these methods in alleviating data sparsity and cold start problems that hinder traditional collaborative filtering techniques.
