基于时间序列分析的微博突发话题检测方法

doi:10.11959/j.issn.1000-436x.2016052

通信学报 ›› 2016, Vol. 37 ›› Issue (3): 48-54.doi: 10.11959/j.issn.1000-436x.2016052

基于时间序列分析的微博突发话题检测方法

贺敏^1,²,徐杰²,杜攀¹,程学旗¹,王丽宏²

¹ 中国科学院计算技术研究所，北京100080
² 国家计算机网络应急技术处理协调中心，北京100029

出版日期:2016-03-25 发布日期:2017-08-04
基金资助:
国家高技术研究发展计划（“863”计划）基金资助项目;国家科技支撑计划基金资助项目

Bursty topic detection method for microblog based on time series analysis

Min HE^1,²,Jie2 XU²,Pan1 DU¹,Xue-qi1 CHENG¹,Li-hong WANG²

¹ Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100080,China
² National Computer Network Emergency Response Technical Team/Coordination Center of China,Beijing 100029,China

Online:2016-03-25 Published:2017-08-04
Supported by:
The National High Technology Research and Development Program of China(863 Program);The National Key Technology Support Program

摘要/Abstract

摘要：

针对微博信息噪音大、新颖度难以判断的问题，在动量模型的基础上进行优化，提出了基于时序分析的微博突发话题检测方法。通过动量模型提取候选突发特征后，对特征的动量时间序列分别借鉴信号频域分析理论和股票趋势分析理论进行建模，分析特征的频域特性来识别频繁伪突发特征，分析特征的新颖程度来识别间歇性伪突发特征，合并过滤后的有效突发特征形成突发话题。微博数据实验表明，该方法有效提高了突发话题检测的准确率和F值。

关键词: 突发话题, 微博, 突发特征, 时序分析

Abstract:

Detecting bursty topics from microblogs was an important task to understand the current events attracting a large number of internet users.However,the existing hods suitable for news articles cannot be adopted directly for microblogs.Because microblogs have unique characteristics compared wi formal texts,including diversity,dynamic and noise.A detection method for microblog bursty topic was proposed based on time series analysis,which was an op-timization method of momentum model.The candidate bursty features were extracted by momentum model.The time se-ries of feature's momentum were modled by frequency domain analysis theory and stock trend analysis theory.The fre-quently pseudo-bursty features were filtered according to analysis results of frequency-domain characteristics.The inter-mittently pseudo-bursty features were filtered according to the novelty analysis result through stock trend theory.The bursty topics were finally emerged with combination of effective bursty features.The experiments are conducted on a real Sina microblog data set.It show that the proposed method improves the precis and F-measure remarkably compared with the momentum modle.

Key words: bursty topic, microblog, bursty feature, time series analysis

贺敏,徐杰,杜攀,程学旗,王丽宏. 基于时间序列分析的微博突发话题检测方法[J]. 通信学报, 2016, 37(3): 48-54.

Min HE,Jie2 XU,Pan1 DU,Xue-qi1 CHENG,Li-hong WANG. Bursty topic detection method for microblog based on time series analysis[J]. Journal on Communications, 2016, 37(3): 48-54.

图/表 5

图1

图2

图3

表1

表2

参考文献 14

[1]	FUNG G , YU J , YU P , et al. Parameter free bursty events detection in text streams[C]// Conference on 31th VLDB. Trondheim,Norway, c2005:181-192.
[2]	HE Q , CHANG K , LIM E . Analyzing feature trajectories for event detection[C]// Conference on 30th SIGIR. Amsterdam, c2007:208-214.
[3]	KLEINBERG J . Bursty and hierarchical structure in steam[C]// Con-ference on KDD'02. Edmonton,Alberta,Canada, c2002:91-101.
[4]	CUI A , ZHANG M , LIU Y , et al. Discover breaking events with popu-lar hashtags in twitter[C]// Conference on CIKM'12. Maui,HI,USA, c2012:1796-1798.
[5]	DU Y Y , HE Y X , TIAN Y . Microblog bursty topic detection based on userrelationship[C]// 6th IEEE Information Technology and Artificial Intelligence Conference. Chongqing,China, c2011:260-263.
[6]	SHIVA P K , PREM M , ARINDAM B . Emerging topic detection using dictionary learning[C]// Conference on CIKM'11. Glasgow,Scotland,UK, c2011:745-754.
[7]	ZHU M L , HU W M , WU O . Topic detection and tracking for threaded discussion communities[C]// IEEE/WIC/ACM International Conferences on Web Intelligences and Intelligent Agent Technology. c2008:77-83.
[8]	贺敏，杜攀，张瑾，等. 基于有意义串动量模型的微博突发话题检测方法[J]. 计算机研究与发展， 2015,52(5): 1022-1028. HE M , DU P , ZHANG J , et al. Microblog bursty topic detection me-thod based on momentum model[J]. Journal of Computer h and Development, 2015,52(5): 1022-1028.
[9]	贺敏 . 面向互联网的有意义串挖掘[D]. 北京: 中国科学院计算技术研究所， 2007. HE M . Web-oriented Chinese meaningful string mining[D]. Beijing: Insti-tute of Computing Technology,Chinese Academy of Sciences, 2007.
[10]	ALAN R , MAUSAM O E . Open domain event extraction from twit-ter[C]// Conference on KDD'12. Beijing,China, c2012:1104-1112.
[11]	ANDREW J , YASHAR M , JOEMON M . Building a large-scale cor-pus for evaluating event detection on twitter[C]// Conference on CIKM'13. San Francisco,CA,USA, c2013:409-418.
[12]	DIAO Q M , JIANG J , ZHU F D , et al. Finding bursty topics from microblogs[C]// The 50th Annual Meeting of the Association for Computational Linguistics. Jeju,Korea, c2012:536-544.
[13]	POPESCU A M , PENNACCHIOTTI M , PARANJPE D A . Extracting events and event descriptions from twitter[C]// Conference on WWW'11. Hyderabad,India, c2011:105-106.
[14]	LI CL , SUN AX , DATTA A . Twevent:segment-based event detection from tweets[C]// Conference on CIKM'12. Maui,HI,USA, c2012:155-164.

方法	P	R	F
TF-IDF＆UF-ITUF	70.27%	72.22%	71.23%
动量模型	87.71%	87.22%	87.47%
动量模型+频域分析	91.18%	86.11%	88.57%
动量模型+趋势分析	92.35%	87.22%	89.71%
动量模型+频域分析+趋势分析	96.27%	86.11%	90.91%

方法	识别数量	准确数量	准确率
动量模型	1 262	1 089	86.29%
动量模型+频域分析	1 024	933	91.11%
动量模型+趋势分析	1 185	1 087	91.73%
动量模型+频域分析+趋势分析	945	933	98.73%

[1]	石磊,杜军平,梁美玉. 基于RNN和主题模型的社交网络突发话题发现[J]. 通信学报, 2018, 39(4): 189-198.
[2]	张艳梅,黄莹莹,甘世杰,丁熠,马志龙. 基于贝叶斯模型的微博网络水军识别算法研究[J]. 通信学报, 2017, 38(1): 44-53.
[3]	张宇翔,孙菀,杨家海,周达磊,孟祥飞,肖春景. 新浪微博反垃圾中特征选择的重要性分析[J]. 通信学报, 2016, 37(8): 24-33.
[4]	陈福,林闯,薛超,徐月梅,孟坤,倪艺函. 短句语义向量计算方法[J]. 通信学报, 2016, 37(2): 11-19.
[5]	于淼,杨武,王巍,申国伟. 面向微博的多实体稀疏关系数据联合聚类[J]. 通信学报, 2016, 37(1): 151-159.
[6]	张光磊,徐雅斌. 面向热门话题的微博观点挖掘方法[J]. 通信学报, 2014, 35(Z2): 220-227.
[7]	卜俊丽,彭灿,郑毅,黄九鸣,周斌. 面向舆情监测的微博转世账户研判模型[J]. 通信学报, 2014, 35(Z2): 228-232.
[8]	张光磊，徐雅斌. 面向热门话题的微博观点挖掘方法[J]. 通信学报, 2014, 35(Z2): 30-227.
[9]	卜俊丽，彭灿，郑毅，黄九鸣，周斌. 面向舆情监测的微博转世账户研判模型[J]. 通信学报, 2014, 35(Z2): 31-232.
[10]	赖清楠1，马皓2，宋维佳2，李婷婷3，蒋广学3，张蓓2. 高校BBS与微博的用户社交行为特征分析[J]. 通信学报, 2013, 34(Z2): 20-106.
[11]	赖清楠,马皓,宋维佳,李婷婷,蒋广学,张蓓. 高校BBS与微博的用户社交行为特征分析[J]. 通信学报, 2013, 34(Z2): 99-106.
[12]	贺敏,王丽宏,杜攀,张瑾,程学旗. 基于有意义串聚类的微博热点话题发现方法[J]. 通信学报, 2013, 34(Z1): 256-262.
[13]	王巍1，李锐光2，周渊2，杨武1. 基于用户与节点规模的微博突发话题传播预测算法[J]. 通信学报, 2013, 34(Z1): 11-91.
[14]	贺敏1,2，王丽宏2，杜攀1，张瑾1，程学旗1. 基于有意义串聚类的微博热点话题发现方法[J]. 通信学报, 2013, 34(Z1): 34-262.
[15]	王巍,李锐光,周渊,杨武. 基于用户与节点规模的微博突发话题传播预测算法[J]. 通信学报, 2013, 34(Z1): 84-91.

基于时间序列分析的微博突发话题检测方法

Bursty topic detection method for microblog based on time series analysis

在线阅读

PDF下载

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 5

参考文献 14

相关文章 15

Metrics

推荐阅读 0