中国典型管理期刊文献主题发现与演化分析
|
摘要
随着计算机技术和网络技术的发展,文献数量指数倍增长,而文献是传递情报信息,交流学术思想的重要载体,能快速地反映社会性和科技的动态,现状及未来的发展趋势。因此,从大量的文献中获取文献主题信息以及主题发展演化信息是非常必要的。
本文在对文献主题分析方法比较的基础上,采用目前国外比较流行的相关主题模型(CTM)对我国管理期刊1984-2009年主题研究情况进行分析,主要包括主题组成分析、主流主题分析、期刊主题强度演变分析、期刊主题内容演变分析以及热点主题分析等。
(1)采用相关主题模型对我国管理期刊1984-2009年的主题组成进行分析,得出我国管理期刊此时间段的主题构成；应用本文定义的主题重要性指标,对主题进行重要性排序,分析出主流主题。
(2)对我国管理期刊1984-2009年的文献每两年为一个时间段进行划分,对每个时间段的文献数据运用相关主题模型进行主题发现,并根据相邻时间段主题关联矩阵构建主题演化网络。通过分析主题演化网络,可以得到主题在各年度之间的发展主线。按照主线随各年度的发展趋势,把主题划分为增长型、衰退型和稳定型三个类别；并根据主题概率和主题增长速度,分析了热点主题和冷点主题。
(3)在各时间段演变过程中,主题领域会包含多个子主题,子主题在一定时间段内会显示出自己的演变趋势。本文以主题企业管理为例,分析了主题领域子主题的演变和发展趋势。
本文的研究成果可以为科研人员和科研管理人员把握我国管理期刊文献的主题及主题演化提供帮助,可以避免科研工作的重复和浪费。
With the computer technology and network technology development, the quantity of the literature grows with index times. Literature is an important carrier to transfer information and to exchange academic ideas, which can quickly reflect social and technological dynamic, the current situation and future development trend. Therefore, it is very necessary to acquire the information of topics and topic evolution from a lot of literature.
In this paper, we compare the methods of document theme analysis and decide to adopt correlated topic models (CTM) to analyze the themes of Chinese management journal from 1984 to 2009. The main analysis content includes theme composition analysis, mainstream subject analysis, journal theme strength evolution analysis, journal theme content evolution analysis, hot topic analysis and so on.
(1) We use correlated topic models to analyze topic composition of Chinese management journal from 1984 to 2009 and obtain theme structure of Chinese management journal in this interim; we sort topic importance according to theme importance index defined in this paper and conclude mainstream themes.
(2) We divide literature data of Chinese management journal every two years for a time segment and use correlated topic models to discovery topics in each time segment. According to theme correlation matrix, we construct the network of topic evolution. Through analyzing the topic evolution network, we can get the developing main thread of topics between different segments. According to development trend of main thread, topics can be divided into three categories, which are increasing type, decreasing type and stable type; and according to topic probability and the speed of topic growth, we analyze hot topics and cold topics.
(3) In the course of topic evolution, topic area may include multiple sub-topics, which display their own evolution tendency over time. Taking the topic of enterprise management as an example in this paper, we analyze subtopic evolution and development trend of enterprise management.
The result of this paper can provide help for researchers and scientific management personnel to grasp the information of topics and topic evolution in the management journal, which can avoid the repetition and waste of research work.
本文在对文献主题分析方法比较的基础上,采用目前国外比较流行的相关主题模型(CTM)对我国管理期刊1984-2009年主题研究情况进行分析,主要包括主题组成分析、主流主题分析、期刊主题强度演变分析、期刊主题内容演变分析以及热点主题分析等。
(1)采用相关主题模型对我国管理期刊1984-2009年的主题组成进行分析,得出我国管理期刊此时间段的主题构成；应用本文定义的主题重要性指标,对主题进行重要性排序,分析出主流主题。
(2)对我国管理期刊1984-2009年的文献每两年为一个时间段进行划分,对每个时间段的文献数据运用相关主题模型进行主题发现,并根据相邻时间段主题关联矩阵构建主题演化网络。通过分析主题演化网络,可以得到主题在各年度之间的发展主线。按照主线随各年度的发展趋势,把主题划分为增长型、衰退型和稳定型三个类别；并根据主题概率和主题增长速度,分析了热点主题和冷点主题。
(3)在各时间段演变过程中,主题领域会包含多个子主题,子主题在一定时间段内会显示出自己的演变趋势。本文以主题企业管理为例,分析了主题领域子主题的演变和发展趋势。
本文的研究成果可以为科研人员和科研管理人员把握我国管理期刊文献的主题及主题演化提供帮助,可以避免科研工作的重复和浪费。
With the computer technology and network technology development, the quantity of the literature grows with index times. Literature is an important carrier to transfer information and to exchange academic ideas, which can quickly reflect social and technological dynamic, the current situation and future development trend. Therefore, it is very necessary to acquire the information of topics and topic evolution from a lot of literature.
In this paper, we compare the methods of document theme analysis and decide to adopt correlated topic models (CTM) to analyze the themes of Chinese management journal from 1984 to 2009. The main analysis content includes theme composition analysis, mainstream subject analysis, journal theme strength evolution analysis, journal theme content evolution analysis, hot topic analysis and so on.
(1) We use correlated topic models to analyze topic composition of Chinese management journal from 1984 to 2009 and obtain theme structure of Chinese management journal in this interim; we sort topic importance according to theme importance index defined in this paper and conclude mainstream themes.
(2) We divide literature data of Chinese management journal every two years for a time segment and use correlated topic models to discovery topics in each time segment. According to theme correlation matrix, we construct the network of topic evolution. Through analyzing the topic evolution network, we can get the developing main thread of topics between different segments. According to development trend of main thread, topics can be divided into three categories, which are increasing type, decreasing type and stable type; and according to topic probability and the speed of topic growth, we analyze hot topics and cold topics.
(3) In the course of topic evolution, topic area may include multiple sub-topics, which display their own evolution tendency over time. Taking the topic of enterprise management as an example in this paper, we analyze subtopic evolution and development trend of enterprise management.
The result of this paper can provide help for researchers and scientific management personnel to grasp the information of topics and topic evolution in the management journal, which can avoid the repetition and waste of research work.
引文
[1]胡昌平,邱均平.科技文献分析研究基本方法初析[J].图书情报知识,1984(1)：50-53.
[2]马费成,张勤.国内外知识管理研究热点——基于词频的统计分析[J].情报学报,2006,25(2)：163-171.
[3]梁立明,谢彩霞.词频分析法用于我国纳米科技研究动向分析[J].科学学研究,2003,21(2)：138-142.
[4]李文兰,杨祖国.1993-2002年情报学科研究主题分析[J].图书情报工作,2003,48(7)：96-99.
[5]邱均平,谭春辉,文庭孝.2004年国内外情报学研究重点及其演进[J].图书馆论坛,2004,25(2)：71-76.
[6]邱均平,谭春辉,任全娥.“十五”期间情报学论文热点及其研究重点的发展趋势[J].情报学报,2006(s1)：287-289.
[7]周凤华,朱雪忠.2000-2004年中国知识产权研究的文献计量分析[J].科研管理,2006,27(4)：53-60.
[8]魏瑞斌.基于关键词的情报学研究主题分析[J].情报科学,2006,24(9)：1400-1404.
[9]王亮亮,彭晓东.国内外管理科学学科发展趋势研究——基于内容分析方法[J].科学学研究,2007,25(5)：959-962.
[10]岳洪江,刘思峰,梁立明.我国对技术创新的关注与研究—基于24年的文献计量分析[J].科研管理,2008,29(3)：43-52.
[11]苏娜.基于共词分析的数字图书馆领域研究主题及进展分析[J].情报杂志,2009,28(6)：15-19.
[12]倪丽娟,于淑丽.档案学研究热点分析——基于2004-2008年《档案学研究》、《档案学通讯》论文关键词的词频分析[J].档案学通讯,2010(1)：19-22.
[13]Robert D, Gautheir E, Ippersiel MP. The State of Nanoteehnology Researeh:Report to the National Research Couneil of Canada [R].1997,5.
[14]庞景安.科学计量研究方法论[M].北京：科学技术文献出版社,2002.
[15]Garfield E. Citation indexes for science-new dimension in documentation through association of ideas [J]. Science,1955,122(3159):108-111.
[16]常爱东.1999-2003年我国图书情报学博硕士论文及引文计量分析[J].情报资料工作,2006(2)：72-75.
[17]王孝宁,崔雷.2001-2006年国际情报学研究的引文分析[J].情报学报,2007,26(3)：399-407.
[18]赵丽红.基于高被引论文的情报学研究现状分析[J].现代情报,2008,28(12)：157-162.
[19]陈仕吉,王小梅.基于C-value与TF-IDF的文献簇主题识别研究[J].情报学报,2009,28(6)：821-826.
[20]刘雪立,王兆军.2004-2008年我国情报专题研究高被引论文的统计与分析[J].情报杂志,2010,29(1)：64-67.
[21]魏海燕,尹怀琼,刘莉.基于引文分析的情报学与相关学科的研究[J].情报杂志,2010,29(2)：38-43.
[22]Price D. Networks of Scientific Papers [J]. Science,1965,14(9):510-515.
[23]Roosa Leimu, Julia Koricheva. What determines the citation frequency of ecological papers? [J] Trends in Ecology & Evolution,2005,20(1):28-32.
[24]Leicht E A, Clarkson G, Shedden K, et al. Large-Scale structure of time evolving citation networks [J].Eur. Phys. J. B.,2007,59:75-83.
[25]Quirin A, Cordon 0, Santamaria J, et al. A new variant of the pathfinder algorithm to generate large visual science maps in cubic time [J]. Information Processing & Management,2008,44(4):1611-1623.
[26]思萌.引文分析法的作用、局限性及其改进[J].图书馆建设,1992(6)：17-22.
[27]薛为民,陆玉昌.文本挖掘技术研究[J].北京联合大学学报(自然科学版),2005,19(4)：59-63.
[28]Feldman R, Dagan I. KDT-knowledge discovery in texts [C]. In Proceedings of the First International Conference on Knowledge and Data Mining (KDD), Canada, 1995:112-117.
[29]堪志群,张国煊.文本挖掘研究进展[J].模式识别与人工智能,2005,18(1)：65-74.
[30]Losiewicz P, Oard D W, Kostoff R N. Textual Data Mining to Support Science and Technology Management [J]. Journal of Intelligent Information Systems, 2000(15):99-119.
[31]屈鹏,吴龙婷,隆捷,等.国际情报学研究主题的聚类分析—基于1996-2003年的LISA数据库[J].情报学报,2007,26(6)：909-918.
[32]丁堃,李鑫.我国知识管理研究主题变化的计量分析[J].科学学研究,2008,26(2)：373-379.
[33]李海燕,崔蒙,崔雷.近十年来国外对中草药的关注点——基于高频主题词的共现聚类分析[J].情报学报,2009,28(3)：395-403.
[34]宋丹.基于语义和链接的话题跟踪方法[D].大连：大连理工大学,2007.
[35]Le M H, Ho T B, Nakamori Y. Detecting Emerging Trends from Scientific Corpora [J]. International Journal of Knowledge and Systems Science,2005,2(2):53-59.
[36]Qiaozhu Mei, ChengXiang Zhai. Discovering Evolutionary Topic Patterns from Text--An Exploration of Temporal Text Mining [C]. Proceedings of the 2005 ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD'05), Chicago, 2005:198-207.
[37]Yi-Ning Tu, Jia-Lang Sen. Research intelligence involving information retrieval-An example of conferences and journals [J]. Expert Systems with Applications,2009,36(10):12151-12166.
[38]骆国靖.基于主题模型的模块化网络和社区挖掘研究[D].杭州：浙江大学,2008.
[39]王金龙,徐从富,耿雪玉.基于概率图模型的科研文献主题演化研究[J].情报学报,2009,28(3)：347-355.
[40]Scott CD, Susan TD, Thomas KL, et al. Indexing by latent semantic analysis[J]. Journal of the American Society for Information Science,1990,41(6):391-407.
[41]Hofmann T. Probabilistic latent semantic analysis[C]//Proc of the Fifteenth Conference on Uncertainty in Artificial Intelligence, Stockhom:Morgan Kaufmann, 1999,289-296.
[42]Blei DM, Andrew YN, Michael IJ.2003. Latent Dirichlet Allocation [J]. Journal of Machine Learning Research,2003,3:993-1022
[43]Griffiths TL, Steyvers M. Finding scientific topics [J]. The National Academy of Sciences,2004,101:5228-5235
[44]Rosen-Zvi M, Griffiths T, Steyvers M, et al. The author-topic model for authors and documents [C]. Proceedings of the 20"'conference in uncertainty in Artificial Intelligence, Banff,2004:487-494.
[45]Xuerui Wang, Andrew McCallum. Topic over time:A Non-Markov Continuous-Time Model of Topical Trends [C]. ACM SIGKDD-2006, Philadelphia,2006:424-433.
[46]Blei D, Lafferty J. Dynamic Topic Models [C]. Proceedings of the 23rd International Conference on Machine Learning, Pittsburgh,2006:113-120.
[47]Blei D, Lafferty J. Correlated Topic Models [C]. Advances in Neural Information Processing Systems 18.2006.
[48]Blei D, Lafferty J. A correlated topic model of Science [C]. Annals of Applied Statistics,2007,1(1):17-35.
[49]Griths T, Steyvers M. Finding scientific topics [C]. Proceedings of the National Academy of Sciences of the United States of America, Washington,2004:5528-5235.
[50]Levent Bolelli, Seyda Ertekin, Ding Zhou, et al. Finding topic trends in digital libraries [C]. JCDL 2009, Texas,2009:69-72.
[51]Levent Bolelli, Seyda Ertekin, C. Lee Giles. Topic and Trend Detection in Text Collections Using Latent Dirichlet Allocation [C]. ECIR 2009, Manufacture des Tabacs, University of Toulouse 1,2009:776-780.
[52]Qi He, Bi Chen, Jian Pei, et al. Detecting topic evolution in scientific literature: how can citations help? [C] CIKM 2009, Hong Kong,2009:957-966.
[53]Salton G, Wong A, Yang C S. A vector space model for automatic indexing [J]. Communications of the ACM,1975,18(11):613-620.
[54]冯书晓,徐新,杨春梅.国内中文分词技术研究新进展[J].情报杂志,2002,21(11)：29-30.
[55]中文分类主题词典[M].北京：北京图书出版社,2005,ISBN：7-89996-811-9.
[56]Salton G, Buckley C. Term weighting approaches in automatic text retrieval [J]. Information Processing & Management,1998,24(5):513-523.
[57]Salton G, Allan J, Buckley C. Automatic structuring and retrieval of large text files [J]. Communications of the ACM,1994,37(2):97-108.
[58]Aitchison J. The statistical analysis of compositional data (with discussion). J. Roy. Statist. Soc. Ser. B.,1982.44 139-177. MR0676206.
[59]Aitchison J. A general class of distributions on the simplex. J. Roy. Statist. Soc. Ser. B.1985.47 136-146. MR0805071.
[60]Aitchison J, Shen S. Logistic normal distributions:Some properties and uses [J]. Biometrika,1980,67(2):261-272. MR0581723.
[61]Bishop C, Spiegelhalter D, Winn J. VIBES:A variational inference engine for Bayesian networks [C]. In Advances in Neural Information Processing Systems 15(Becker S, Thrun S, Obermayer K, et all.)2003,777-784. MIT Press, Cambridge, MA. MR2003168.
[62]Blei D, Jordan M. Variational inference for Dirichlet process mixtures [J]. Journal of Bayesian Analysis,2006,1(1):121-144. MR2227367.
[63]Xing E, Jordan M, Russell S. A generalized mean field algorithm for variational inference in exponential families [C]. In Proceedings of UAI, Morgan Kaufmann, San Francisco, CA,2003:583-591.
[64]Jordan M, Ghahramani Z, Jaakkola T, et all. Introduction to variational methods for graphical models [J]. Machine Learning,1999,37:183-233.
[65]Wainwright M, Jordan M. Graphical models, exponential families and variational inference [R]. Technical Report 649, Dept. Statistics, U.C.Berkeley.2003.
[2]马费成,张勤.国内外知识管理研究热点——基于词频的统计分析[J].情报学报,2006,25(2)：163-171.
[3]梁立明,谢彩霞.词频分析法用于我国纳米科技研究动向分析[J].科学学研究,2003,21(2)：138-142.
[4]李文兰,杨祖国.1993-2002年情报学科研究主题分析[J].图书情报工作,2003,48(7)：96-99.
[5]邱均平,谭春辉,文庭孝.2004年国内外情报学研究重点及其演进[J].图书馆论坛,2004,25(2)：71-76.
[6]邱均平,谭春辉,任全娥.“十五”期间情报学论文热点及其研究重点的发展趋势[J].情报学报,2006(s1)：287-289.
[7]周凤华,朱雪忠.2000-2004年中国知识产权研究的文献计量分析[J].科研管理,2006,27(4)：53-60.
[8]魏瑞斌.基于关键词的情报学研究主题分析[J].情报科学,2006,24(9)：1400-1404.
[9]王亮亮,彭晓东.国内外管理科学学科发展趋势研究——基于内容分析方法[J].科学学研究,2007,25(5)：959-962.
[10]岳洪江,刘思峰,梁立明.我国对技术创新的关注与研究—基于24年的文献计量分析[J].科研管理,2008,29(3)：43-52.
[11]苏娜.基于共词分析的数字图书馆领域研究主题及进展分析[J].情报杂志,2009,28(6)：15-19.
[12]倪丽娟,于淑丽.档案学研究热点分析——基于2004-2008年《档案学研究》、《档案学通讯》论文关键词的词频分析[J].档案学通讯,2010(1)：19-22.
[13]Robert D, Gautheir E, Ippersiel MP. The State of Nanoteehnology Researeh:Report to the National Research Couneil of Canada [R].1997,5.
[14]庞景安.科学计量研究方法论[M].北京：科学技术文献出版社,2002.
[15]Garfield E. Citation indexes for science-new dimension in documentation through association of ideas [J]. Science,1955,122(3159):108-111.
[16]常爱东.1999-2003年我国图书情报学博硕士论文及引文计量分析[J].情报资料工作,2006(2)：72-75.
[17]王孝宁,崔雷.2001-2006年国际情报学研究的引文分析[J].情报学报,2007,26(3)：399-407.
[18]赵丽红.基于高被引论文的情报学研究现状分析[J].现代情报,2008,28(12)：157-162.
[19]陈仕吉,王小梅.基于C-value与TF-IDF的文献簇主题识别研究[J].情报学报,2009,28(6)：821-826.
[20]刘雪立,王兆军.2004-2008年我国情报专题研究高被引论文的统计与分析[J].情报杂志,2010,29(1)：64-67.
[21]魏海燕,尹怀琼,刘莉.基于引文分析的情报学与相关学科的研究[J].情报杂志,2010,29(2)：38-43.
[22]Price D. Networks of Scientific Papers [J]. Science,1965,14(9):510-515.
[23]Roosa Leimu, Julia Koricheva. What determines the citation frequency of ecological papers? [J] Trends in Ecology & Evolution,2005,20(1):28-32.
[24]Leicht E A, Clarkson G, Shedden K, et al. Large-Scale structure of time evolving citation networks [J].Eur. Phys. J. B.,2007,59:75-83.
[25]Quirin A, Cordon 0, Santamaria J, et al. A new variant of the pathfinder algorithm to generate large visual science maps in cubic time [J]. Information Processing & Management,2008,44(4):1611-1623.
[26]思萌.引文分析法的作用、局限性及其改进[J].图书馆建设,1992(6)：17-22.
[27]薛为民,陆玉昌.文本挖掘技术研究[J].北京联合大学学报(自然科学版),2005,19(4)：59-63.
[28]Feldman R, Dagan I. KDT-knowledge discovery in texts [C]. In Proceedings of the First International Conference on Knowledge and Data Mining (KDD), Canada, 1995:112-117.
[29]堪志群,张国煊.文本挖掘研究进展[J].模式识别与人工智能,2005,18(1)：65-74.
[30]Losiewicz P, Oard D W, Kostoff R N. Textual Data Mining to Support Science and Technology Management [J]. Journal of Intelligent Information Systems, 2000(15):99-119.
[31]屈鹏,吴龙婷,隆捷,等.国际情报学研究主题的聚类分析—基于1996-2003年的LISA数据库[J].情报学报,2007,26(6)：909-918.
[32]丁堃,李鑫.我国知识管理研究主题变化的计量分析[J].科学学研究,2008,26(2)：373-379.
[33]李海燕,崔蒙,崔雷.近十年来国外对中草药的关注点——基于高频主题词的共现聚类分析[J].情报学报,2009,28(3)：395-403.
[34]宋丹.基于语义和链接的话题跟踪方法[D].大连：大连理工大学,2007.
[35]Le M H, Ho T B, Nakamori Y. Detecting Emerging Trends from Scientific Corpora [J]. International Journal of Knowledge and Systems Science,2005,2(2):53-59.
[36]Qiaozhu Mei, ChengXiang Zhai. Discovering Evolutionary Topic Patterns from Text--An Exploration of Temporal Text Mining [C]. Proceedings of the 2005 ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD'05), Chicago, 2005:198-207.
[37]Yi-Ning Tu, Jia-Lang Sen. Research intelligence involving information retrieval-An example of conferences and journals [J]. Expert Systems with Applications,2009,36(10):12151-12166.
[38]骆国靖.基于主题模型的模块化网络和社区挖掘研究[D].杭州：浙江大学,2008.
[39]王金龙,徐从富,耿雪玉.基于概率图模型的科研文献主题演化研究[J].情报学报,2009,28(3)：347-355.
[40]Scott CD, Susan TD, Thomas KL, et al. Indexing by latent semantic analysis[J]. Journal of the American Society for Information Science,1990,41(6):391-407.
[41]Hofmann T. Probabilistic latent semantic analysis[C]//Proc of the Fifteenth Conference on Uncertainty in Artificial Intelligence, Stockhom:Morgan Kaufmann, 1999,289-296.
[42]Blei DM, Andrew YN, Michael IJ.2003. Latent Dirichlet Allocation [J]. Journal of Machine Learning Research,2003,3:993-1022
[43]Griffiths TL, Steyvers M. Finding scientific topics [J]. The National Academy of Sciences,2004,101:5228-5235
[44]Rosen-Zvi M, Griffiths T, Steyvers M, et al. The author-topic model for authors and documents [C]. Proceedings of the 20"'conference in uncertainty in Artificial Intelligence, Banff,2004:487-494.
[45]Xuerui Wang, Andrew McCallum. Topic over time:A Non-Markov Continuous-Time Model of Topical Trends [C]. ACM SIGKDD-2006, Philadelphia,2006:424-433.
[46]Blei D, Lafferty J. Dynamic Topic Models [C]. Proceedings of the 23rd International Conference on Machine Learning, Pittsburgh,2006:113-120.
[47]Blei D, Lafferty J. Correlated Topic Models [C]. Advances in Neural Information Processing Systems 18.2006.
[48]Blei D, Lafferty J. A correlated topic model of Science [C]. Annals of Applied Statistics,2007,1(1):17-35.
[49]Griths T, Steyvers M. Finding scientific topics [C]. Proceedings of the National Academy of Sciences of the United States of America, Washington,2004:5528-5235.
[50]Levent Bolelli, Seyda Ertekin, Ding Zhou, et al. Finding topic trends in digital libraries [C]. JCDL 2009, Texas,2009:69-72.
[51]Levent Bolelli, Seyda Ertekin, C. Lee Giles. Topic and Trend Detection in Text Collections Using Latent Dirichlet Allocation [C]. ECIR 2009, Manufacture des Tabacs, University of Toulouse 1,2009:776-780.
[52]Qi He, Bi Chen, Jian Pei, et al. Detecting topic evolution in scientific literature: how can citations help? [C] CIKM 2009, Hong Kong,2009:957-966.
[53]Salton G, Wong A, Yang C S. A vector space model for automatic indexing [J]. Communications of the ACM,1975,18(11):613-620.
[54]冯书晓,徐新,杨春梅.国内中文分词技术研究新进展[J].情报杂志,2002,21(11)：29-30.
[55]中文分类主题词典[M].北京：北京图书出版社,2005,ISBN：7-89996-811-9.
[56]Salton G, Buckley C. Term weighting approaches in automatic text retrieval [J]. Information Processing & Management,1998,24(5):513-523.
[57]Salton G, Allan J, Buckley C. Automatic structuring and retrieval of large text files [J]. Communications of the ACM,1994,37(2):97-108.
[58]Aitchison J. The statistical analysis of compositional data (with discussion). J. Roy. Statist. Soc. Ser. B.,1982.44 139-177. MR0676206.
[59]Aitchison J. A general class of distributions on the simplex. J. Roy. Statist. Soc. Ser. B.1985.47 136-146. MR0805071.
[60]Aitchison J, Shen S. Logistic normal distributions:Some properties and uses [J]. Biometrika,1980,67(2):261-272. MR0581723.
[61]Bishop C, Spiegelhalter D, Winn J. VIBES:A variational inference engine for Bayesian networks [C]. In Advances in Neural Information Processing Systems 15(Becker S, Thrun S, Obermayer K, et all.)2003,777-784. MIT Press, Cambridge, MA. MR2003168.
[62]Blei D, Jordan M. Variational inference for Dirichlet process mixtures [J]. Journal of Bayesian Analysis,2006,1(1):121-144. MR2227367.
[63]Xing E, Jordan M, Russell S. A generalized mean field algorithm for variational inference in exponential families [C]. In Proceedings of UAI, Morgan Kaufmann, San Francisco, CA,2003:583-591.
[64]Jordan M, Ghahramani Z, Jaakkola T, et all. Introduction to variational methods for graphical models [J]. Machine Learning,1999,37:183-233.
[65]Wainwright M, Jordan M. Graphical models, exponential families and variational inference [R]. Technical Report 649, Dept. Statistics, U.C.Berkeley.2003.