社会网络中非参数化的事件演化分析
Analysis of Non-Parametric Event Evolution in Social Networks
摘要: 社会网络，如微博和Twitter，已经成为数十亿人关注事件的重要平台。人们不仅关注所发生的事情，更关注事件的演化。因此，监控社交网络中事件的发展是至关重要的。在新闻文章和短文本(社交网络中的文本)中有一些挖掘事件演化的研究。由于短文的形式比新闻文章短，适用于新闻文章的方法不能直接应用于社交网络。一些应用于短文本的方法不考虑语义信息，有些方法无法发现长期跨度事件的演化，特别是中间有间断的事件。鉴于此，我们提出了一种非参数的方法来发现事件演化(故事情节)。首先使用贝叶斯模型测量短文本的语义相关性。其次，使用基于嵌入表示的算法来生成长期和短期事件的故事线。我们进一步使用Dirichlet过程自动学习适当数量的主题。与其他方法相比，三个人工标记数据集的详细实验结果证明了我们方法的有效性。
Abstract: Social networks, such as Weibo and Twitter, have become important platforms where billions of individuals follow events. People not only concern about what happened, but also pay more attention to how the event gradually progresses. Therefore, it is crucial to monitor the development of events in social networks. There are some research mining event evolutions in news articles and short texts (texts in social networks). Methods designed for news articles cannot be directly applied to social network due to the fact that short texts are more ill-formed and shorter than news articles. Some methods that apply to short texts do not take into account semantic information, and some cannot discover evolution of long-term spanning events, especially intermittent events. In light of this, we propose a non-parametric method to discover event evolution (storylines). Firstly, a bayesian model is used to measure semantic correlation of short texts. Secondly, an embedded representation-based algorithm is used to generate storylines for long-term and short-term events. We further use dirichlet process to automatically learn an appropriate number of topics. In comparison with other methods, detailed experimental results on three manually labeled data sets demonstrate the effectiveness of our method.
文章引用：李莹莹. 社会网络中非参数化的事件演化分析[J]. 计算机科学与应用, 2018, 8(6): 976-985. https://doi.org/10.12677/CSA.2018.86109

参考文献

[1] Lee, P., Lakshmanan, L.V.S. and Milios, E.E. (2014) Incremental Cluster Evolution Tracking from Highly Dynamic Network Data. Proceedings of the IEEE 30th International Conference on Data Engineering, Chicago, IL, 31 March-4 April 2014, 3-14. [Google Scholar] [CrossRef] 
[2] Janani, K., Sumithra, V., et al. (2016) From Event Detection to Storytelling on Microblogs. Proceedings of the IEEE International Conference on Advances in Social Networks Analysis and Mining, San Francisco, CA, 18-21 August 2016, 437-442.
[3] Zhou, D., Xu, H. and He, Y. (2015) An Unsupervised Bayesian Modelling Approach for Storyline Detection on News Articles. Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, Lisbon, Portugal, 17-21 September 2015, 1943-1948. [Google Scholar] [CrossRef] 
[4] Wang, D., Li, T. and Ogihara, M. (2012) Generating Picto-rial Storylines via Minimum-Weight Connected Dominating Set Approximation in Multi-View Graphs. Proceedings of the 26th AAAI Conference on Artificial Intelligence, AAAI, Menlo Park, CA, 683-689.
[5] Yong, C., Hui, Z., Wu, J.J., et al. (2016) Modeling Emerging, Evolving and Fading Topics Using Dynamic Soft Orthogonal NMF with Sparse Representation. Proceeding of the 2015 IEEE International Conference on Data Mining, Piscataway, Atlantic City, NJ, 14-17 November 2015, 61-70.
[6] Zhou, D.Y., Xu, H.Y., Dai, X.Y., et al. (2016) Unsupervised Storyline Extraction from News Articles. Proceedings of the 25th International Joint Conference on Artificial Intelligence, Morgan Kaufmann, Burlington, MA, 3014-3021.
[7] Hua, T., Zhang, X.C., Wang, W., Lu, C.-T. and Ramakrishnan, N. (2016) Automati-cal Storyline Generation with Help from Twitter. Proceedings of the 25th ACM International Conference on Information and Knowledge Management, Indianapolis, Indiana, 24-28 October 2016, 2383-2388. [Google Scholar] [CrossRef] 
[8] Huang, L.F. and Huang, L.E. (2013) Optimized Event Storyline Generation Based on Mixture Event-Aspect Model. Proceedings of the 2013 Conference on Empirical Methods in Natu-ral Language Processing, Seattle, Washington, 18-21 October 2013, 726-735.
[9] Tang, S.L., Wu, F., Li, S., Lu, W.M., Zhang, Z.F. and Zhuang, Y.T. (2015) Sketch the Storyline with CHARCOAL: A Non-Parametric Approach. Proceedings of the 24th International Joint Conference on Artificial Intelligence, Buenos Aires, 25-31 July 2015, 3841-3848.
[10] Lin, C., Lin, C., Li, J.X., Wang, D.D., Chen, Y. and Li, T. (2012) Generating Event Storylines from Microblogs. Proceedings of the 21st ACM International Conference on Information and Knowledge Management, Maui, Hawaii, October 29-November 2 2012, 175-184. [Google Scholar] [CrossRef] 
[11] Kalyanam, J., Mantrach, A., Saez-Trumper, D., Vahabi, H. and Lanckriet, G.R.G. (2015) Leveraging Social Context for Modeling Topic Evolution. Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Sydney, NSW, 10-13 August 2015, 517-526. [Google Scholar] [CrossRef] 
[12] Lee, P., Lakshmanan, L.V.S. and Milios, E.E. (2014) CAST: A Context-Aware Storyteller for Streaming Social Content. Pro-ceedings of the 23rd ACM International Conference on Conference on Information and Knowledge Management, Shanghai, 3-7 November 2014, 789-798. [Google Scholar] [CrossRef] 
[13] Yan, X.H., Guo, J.F., Lan, Y.Y. and Cheng, X.Q. (2013) A Biterm Topic Model for Short Texts. Proceedings of the 22nd International World Wide Web Conference, Rio de Janeiro, 13-17 May 2013, 1445-1456. [Google Scholar] [CrossRef] 
[14] Cheng, X.Q., Yan, X.H., Lan, Y.Y. and Guo, J.F. (2014) BTM: Topic Modeling over Short Texts. IEEE Transactions on Knowledge and Data Engineering, 26, 2928-2941. [Google Scholar] [CrossRef] 
[15] Nie, F.P., Zhu, W. and Li, X.L. (2016) Unsupervised Feature Selection with Structured Graph Optimization. Proceedings of the 30th AAAI Conference on Artificial Intelligence, Phoe-nix, 12-17 February 2016, 1302-1308.