基于多源异构数据融合的技术演化路径预测研究:以食品质量安全快检技术为例

Technology forecasting is of great practical significance to grasp the direction of future technology development. The traditional quantitative technology forecasting is mostly dependent on poor time-efficient science and technology data. In the era of big data, how to use multi-source heterogeneous data to forecast technology evolution path timely and effectively has become an urgent problem to solve. This project intends to use data from papers, patents, projects, media and other data sources. Firstly, the project constructs a multi-source heterogeneous data fusion model based on the relationships of time series, spatial, correlation and relative importance within data to forecast technology evolution path. The technology evolution path is identified by methods such as bibliometrics, LDA, and PCA. Then, this project mines and fuses technology supply and demand information from data sources such as projects and media. The two-way driven (pull from technology demand and push from technology supply) technology evolution path is forecasted by using methods such as multi-dimension evidence theory. In view of the severe food safety situation in China, this project takes the rapid detection technology of food quality and safety as an example to verify the effectiveness of the theory and method proposed in this project. This project intends to construct a technology R&D decision support system from technology forecasting perspective which is based on multi-source heterogeneous data fusion. This project enriches the theory and method of technology forecasting, and also helps to guide the R&D of rapid detection technology which will contribute to the food quality and safety protection in China.

技术预测对把握未来技术发展方向具有重要现实意义。传统定量技术预测多依赖于时效性差的科技数据，在大数据时代，如何利用多源异构数据进行及时有效的技术演化路径预测成为亟需解决的问题。本项目拟利用论文、专利、项目、媒体等数据源，根据数据间的时序关系、空间关系、关联关系以及重要性等原则构建基于技术演化路径预测的多源异构数据融合模型；在利用文献计量学、LDA、PCA等方法识别技术演化路径的基础上，通过挖掘和融合项目、媒体等数据源中包含的技术供给与技术需求信息，采用多维证据理论等方法对技术演化路径进行技术供给推动和技术需求拉动的双向驱动技术演化路径预测；鉴于我国严峻的食品安全形势，以食品质量安全快检技术为例，验证本项目提出的理论和方法的实效性，构建基于多源异构数据融合的技术预测视角下的技术研发决策支持体系。本项目研究丰富了技术预测的理论和方法体系，同时有助于引导快检技术研发，保障我国食品质量安全。

大数据时代，如何高效利用科技大数据是政府、科研机构以及科研工作者面临的一大挑战。本项目围绕科技大数据质量控制、多源数据融合及其在技术演化路径预测、多准则项目评价等方面应用开展一系列研究，并取得一系列研究成果。项目主持人依托项目撰写的资政报告获中央政治局委员肯定性批示1次，获教育部智库专刊《教育要情》全文采纳刊发1篇。项目主持人在International Journal of Project Management，Technological Forecasting and Social Change，Journal of Informetrics，Scientometrics等FMS管理科学高质量期刊发表论文13篇，其中ESI高被引论文2篇，研究成果直接服务科研评价、项目评审改革、多源科技数据融合与应用。本课题为Web of Science/Scopus等基础性权威科技大数据源的完善做了一些基础性铺垫工作。