作者简介：

Dr.Jian Yang is a full professor at Department of Computing, Macquarie University. She received her PhD from The Australian National University in 1995. Before she joined Macquarie University,she worked as an associate professor at Tilburg University, Netherlands(2000-2003), a senior research scientist at the Division of Mathematical and Information Science, CSIRO, Australia (1998-2000), and as an assistantprofessor at The Australian Defence Force Academy, University of New SouthWales (1993-1998).

Dr. Yang has published about200 papers in the international journals and conferences such as IEEEtransactions, Information Systems, Data & Knowledge Engineering, CACM,VLDB, ICDCS, CAiSE, CoopIS, CIKM, etc. She is the co-founder of the International Conference on Service Oriented Computing and now serving as a steering committee member. She has served as program committee co-chairs andgeneral chair of in various international conferences. She is also a regularreviewer for journals such as IEEE Transactions on Knowledge & Data Engineering, Data & Knowledge Engineering, VLDB Journal, IEEE InternetComputing, etc.

Her main research interests are: web service technology; business process management;interoperability, trust and security issues in digital libraries ande-commerce; social network.

讲座摘要：

As one of the most popular social media, Twitter has attracted interests of business and academics to derive topics and apply the outcomes in a wide range of applications such asemergency management, business advertisements, and corporate/government communication. Since tweets are short messages, topic derivation from tweets becomes a big challnege in the area. Most of existing works use the.Twitter content as the only source in the topic derivation. Recently, tweet interactions have been considered additionally for improving the quality of topic derivation. In this talk, we introduce a method that incorporates social interactions such as mention, retweet, etc into Twitter content to derive topics. Experimental results showthat the proposed method with the inclusion of temporal features resultsin a significant improvement in the quality of topic derivation comparing to existing baseline methods.

In this talk, we will explain the general idea of Matrix Factorisation and how it is applied in topic derivation, the experiment set up, and experiment resultsanalysis.