[目的/意义] 根据基金项目数据的特点，提出基于基金项目数据的研究前沿主题探测方法，以期识别出前瞻价值更高的研究前沿，为研究前沿识别相关研究提供借鉴思路。[方法/过程] 首先，基于PLDA模型识别蕴含在基金项目文本中的研究主题；然后基于主题-文档矩阵建立主题和基金项目文档的映射关系，在此基础上，利用主题的资助时间、资助金额和中心性指标进行研究前沿主题探测，从而识别出值得关注的研究前沿主题；最后，利用主题演化可视化分析方法，对研究前沿主题进行演化分析，以预测其发展趋势。[结果/结论] 研究结果表明，该方法可以根据基金项目数据的特点识别出蕴含在其中的研究前沿主题，并且能够分析研究前沿主题的分裂、融合等演化过程。
[Purpose/significance] In this paper, we try to use the fund data to detect the research front topic and forecast its development trend. According to the characteristics of the fund project data, we propose the method of research front topic detection based on the fund project data, in order to identify the research front topic with higher prospective value, providing references for the related research.[Method/process] First of all, the research subject contained in the project was identified by using the PLDA model. s Then, based on the mapping document matrix, we built the mapping relation between the theme and fund project documents. We detected the research front topics based on the subsidy time of topics, the amount of subsidies and the center of the index. Finally, using the method of the topic evolution visualization, we analyzed the evolution of research topics to predict the development trend. [Result/conclusion] The result shows that, according to the characteristics of the fund project data, the method can identify the research front topic, and analyze the evolution process of the splitting and merging of the research front topic.
[1] 刘小平,冷伏海,李泽霞.国际科技前沿分析的方法和途径[J].图书情报工作,2012,56(12):60-65.
[2] BLEI D M, Ng A Y, JORDAN M I. Latent Dirichlet allocation[J].Journal of machine learning research,2003(3): 993-1022.
[3] BLEI D M, LAFFERTY J. Dynamic topic models[C]//Proceedings of the 23rd international conference on machine learning. NewYork:ACM,2006:113-120.
[4] NEWMAN D, ASUNCION A U, SMYTH P, et al. Distributed inference for latent allocation[C]//Advances in neural information processing systems 20. New York:Curran Associates,Inc.,2007:1-6.
[5] WANG Y, BAI H J, STANTON M, et al. PLDA: Parallel latent Dirichlet allocation for large-scale applications[C]//Algorithmic aspects in information and management.Berlin:AAIM,2009:301-314.
[6] 范云满,马建霞.基于LDA与新兴主题特征分析的新兴主题探测研究[J].情报学报,2014,33(7):698-711.
[7] 祝娜,王效岳,杨京,等.基于LDA的科技创新主题语义识别研究[J].图书情报工作,2015,59(14):126-134.
[8] 郑彦宁,许晓阳,刘志辉.基于关键词共现的研究前沿识别方法研究[J].图书情报工作,2016,60(4):85-92.
[9] 赵蓉英,赵浚吟,陈必坤. 我国"图书馆、情报与档案管理"学科的研究现状分析[J].情报杂志,2013, 32(7):106-112.
[10] 秦佳航.图书馆、情报与文献学热点研究主题分布及其发展趋势[J].图书与情报,2013(6):112-115.
[11] 静发冲,李晨英,韩明杰,等. 基于文本挖掘的美国 NSF 生物科学部新兴前沿项目主题分析[J].现代情报,2014, 34(12):107-112.
[12] 李广建,祝振媛. 基于硕博士论文和国家社科基金项目的我国情报学研究热点分析[J].图书情报研究,2016(4):3-11.
[13] 梁伟波.美国NSF资助物流项目的知识图谱分析[J].情报杂志,2016,35(10):114-119.
[14] 王旭仁,姚叶鹏,冉春风,等.一种并行LDA主题模型建立方法研究[J].北京理工大学学报,2013,33(6):590-593.
[15] 高阳,严建峰,刘晓升.朴素并行LDA[J].计算机科学,2015,42(6):243-246.
[16] TU Y N. SENG J L. Indices of novelty for emerging topic detection[J].Information processing management, 2012,48(2):303-325.
[17] 黄鲁成,唐月强,吴菲菲,等.基于文献多属性测度的新兴主题识别方法研究[J].科学学与科学技术管理,2015,36(2):34-43.
[18] BONACICH P B. Factoring and weighting approaches to status scores and clique identification[J].Journal of mathematical sociology,1972, 2 (1):113-120.
[19] 刘自强,王效岳,白如江.多维主题演化模型构建与实证研究[J].情报理论与实践, 2017,40(3):92-98.
[20] 刘自强,王效岳,白如江.多维度视角下学科主题演化可视化分析方法研究[J].中国图书馆学报,2016,42(6):67-84.
[21] JIMENEZ S, GONZALEZ F A, GELBUKH A. Mathematical properties of soft cardinality: enhancing Jaccard, Dice and cosine similarity measures with element distance[J].Information processing management, 2012,48(2):303-325.
[22] 向三明,许柱,李宝荣,等.碳纳米管的制备、改性以及应用研究进展[J].轻工科技,2017(3):29-31.
[23] 张媛,张伟.碳纳米管材料发展现状[J].化工技术与开发,2017,46(3):24-36,48.
[24] 王浩然.碳纳米管及其应用[J].山东化工,2017,46(1):59-61.
