Summary on the Conference of WISE 2013
Download as PPT, PDF1 like899 views
The 14th International Conference on Web Information System Engineering (WISE 2013) took place in Nanjing, China, from October 13 to 15, 2013, featuring 48 research papers with a 24% acceptance rate from 38 countries. Keynotes addressed topics from web-based video processing to big data security, while sessions included various areas such as web mining, recommendation systems, and social networks. The conference also showcased innovative algorithms and frameworks for improving web applications across multiple domains.
Summary on the Conference of WISE 2013
- 1. Summary of WISE 2013 (13th ~15th Oct. 2013, Nanjing) 07/11/14 1Middleware, CCNT, ZJU Yueshen Xu xyshzjucs@zju.edu.cn
- 2. Overview Introduction The 14th International Conference on Web Information System Engineering (WISE) 13th ~ 15th , Nanjing, China Before: HK, Kyoto, Singapore, Roma, Brisbane, NY, Nancy, Poznan, etc. Statistics of acceptance Num. of Research papers: 48 Accepted rate: 24% Num. of Long papers: 25; Num. of Short papers: 23 10 Demos, 5 challenge reports Come from: 38 Countries around the world 07/11/14 Middleware, CCNT, ZJU 2
- 3. Overview General Co-chairs 07/11/14 Middleware, CCNT, ZJU 3 PC Co-chairs Yahho! Research Lab Victoria Univesity University of New South Wales Aristotle University AT&T Lab Industry Chairs Google Research HKUST Tutorial Co-chairs CUHK Poznan University
- 4. Overview Publicity Co-chairs 07/11/14 Middleware, CCNT, ZJU 4 Society Representative Aristotle University University of New South Wales University of Queensland Keynote Speaker Peking University Academician Towards web-based video processing UCSB, ACM Fellow Data-driven Methodologies for understanding, managing and analyzing Online Social Networks
- 5. Overview 07/11/14 Middleware, CCNT, ZJU 5 Keynote Speaker University of Technology, Sydney, Australia Senior Member, IEEE Big Data Related Research Issues and Progress New Jersey Institute of Technology Security of Cyber-Physical Systems Distinguished Young Scientists Forum on Big Data Jianmin Wang, Tsinghua Univ. Enhong Chen, USTC Aoying Zhou, East China normal Univ. Guoren Wang, Northeastern Univ. Etc.
- 6. Session Web Mining (2): 11 Web Recommendation (2): 9 Hidden Web: 4 Web Services: 4 Semi-structured Data and Modeling: 7 Social Web (2) : 11 Web Monitoring and Management: 6 Innovative Techniques and Creations (2): 8 Web Text Mining: 6 Networks and Graphs: 6 Demo (2): 5 07/11/14 Middleware, CCNT, ZJU 6
- 7. Web Mining(I) Ying Xu, Zhiqiang Gao, Campbell Wilson, Zhizheng Zhang, Man Zhu, Qiu Ji: Entity Correspondence with Second-Order Markov Logic. 1-14 Youliang Zhong, Lan Du, Jian Yang: Learning Social Relationship Strength via Matrix Co-Factorization with Multiple Kernels. 15-28 Shengsheng Shi, Wu Wei, Yulong Liu, Haitao Wang, Lei Luo, Chunfeng Yuan, Yihua Huang: NEXIR: A Novel Web Extraction Rule Language toward a Three-Stage Web Data Extraction Model. 29-42 Jun Deng, Liang Du, Yi-Dong Shen: Heterogeneous Metric Learning for Cross-Modal Multimedia Retrieval. 43-56 Margarita Karkali, François Rousseau, Alexandros Ntoulas, Michalis Vazirgiannis: Efficient Online Novelty Detection in News Online. 57-71 07/11/14 Middleware, CCNT, ZJU 7 In this paper we propose a KPMCF model to learn social relationship strength based on users’ latent features inferred from both profile and interaction information. The proposed model takes an uniformed approach of integrating Matrix Co-Factorization with Multiple Kernels. We conduct experiments on real-world data sets for typical web mining applications, showing that the proposed model produces better relationship strength measurement in comparison with other social factors. In this paper, we propose a Bayesian personalized ranking based heterogeneous metric learning (BPRHML) algorithm, which optimizes for correctly ranking the retrieval results. It uses pairwise preference constraints as training data and explicitly optimizes for preserving these constraints. To further encourage the smoothness of learning results, we integrate graph regularization with Bayesian personalized ranking In this paper, we propose a new novelty detection algorithm based on the Inverse Document Frequency (IDF) scoring function. Computing novelty based on IDF enables us to avoid similarity comparisons with previous documents in the text online, thus leading to faster execution times. At the same time, our proposed approach outperforms several commonly used baselines when applied on a real-world news articles dataset. Eric Xing CMU Yueting Zhuang, YanFei Wang, Fei Wu, Yin Zhang, Weiming Lu: Supervised Coupled Dictionary Learning with Group Structures for Multi-modal Retrieval. AAAI 2013, Regular Paper Deng Cai, Xiaofei He, Jiawei Han, Thomas S. Huang: Graph Regularized Nonnegative Matrix Factorization for Data Representation. IEEE Trans. Pattern Anal. Mach. Intell. 33(8): 1548-1560 (2011)
- 8. Web Mining(II) Daling Wang, Shi Feng, Dong Wang, Ge Yu: Detecting Opinion Drift from Chinese Web Comments Based on Sentiment Distribution Computing. 72-81 Peng Zhao, Xue Li, Ke Wang: Feature Extraction from Micro-blogs for Comparison of Products and Services. 82-91 Shahida Jabeen, Xiaoying Gao, Peter Andreae: Directional Context Helps: Guiding Semantic Relatedness Computation by Asymmetric Word Associations. 92-101 Jun Hou, Richi Nayak: The Heterogeneous Cluster Ensemble Method Using Hubness for Clustering Text Documents. 102-110 Abdul Wahid, Xiaoying Gao, Peter Andreae: Exploiting User Queries for Search Result Clustering. 111-120 07/11/14 Middleware, CCNT, ZJU 8 The proposed approach first determines possible drift timestamps according to the change of comment number, computes different sentiment orientations and their distributions at these timestamps, detects opinion drift according to the distribution changes, and analyzes the influences of related events occurring in the timestamps. Extensive experiments were conducted in a real comment set of Chinese forum.In this paper, we show our system namely OpinionAnalyzer, a novel social network analyzer designed to collect opinions from Twitter micro-blogs about two given similar products for an effective comparison between them. The system outcome is a structure of features for the given products that people have expressed opinions about. Then the corresponding sentiment analysis on those features is performed. Our system can be used to understand user’s preference to a certain product and show the reasons why users prefer this product. We propose a cluster ensemble method to map the corpus documents into the semantic space embedded in Wikipedia and group them using multiple types of feature space. A heterogeneous cluster ensemble is constructed with multiple types of relations i.e. document-term, document- concept and document-category. A final clustering solution is obtained by exploiting associations between document pairs and hubness of the documents Adaboost & Bagging George Mason
- 9. Web Recommendation(I) Xin Liu: Towards Context-Aware Social Recommendation via Trust Networks. 121-134 Weilong Yao, Jing He, Guangyan Huang, Jie Cao, Yanchun Zhang: Personalized Recommendation on Multi-Layer Context Graph. 135-148 Giseli Rabello Lopes, Luiz André P. Paes Leme, Bernardo Pereira Nunes, Marco Antonio Casanova, Stefan Dietze: Recommending Tripleset Interlinking through a Social Network Approach. 149-161 Chong Wang, Yao Shen, Huan Yang, Minyi Guo: Improving Rocchio Algorithm for Updating User Profile in Recommender Systems. 162-174 Kai Wang, Richong Zhang, Xudong Liu, Xiaohui Guo, Hailong Sun, Jinpeng Huai: Time-Aware Recommendation based on Tensor Factorization. 175- 188 07/11/14 Middleware, CCNT, ZJU 9 We employ random walk to collect the most relevant ratings based on the multi-dimensional trustworthiness of users in the trust network. Factorization machines model is then applied on the collected ratings to predict missing ratings considering various evaluation based on a real dataset demonstrates that our approach improves the accuracy of the state-of-the-art social, context-aware and trust-aware recommendation modelsIn this paper, we propose a Multi-Layer Context Graph (MLCG) model which incorporates a variety of contextual information into a recommendation process and models the interactions between users and items for better recommendation. Moreover, we provide a new ranking algorithm based on Personalized PageRank for recommendation in MLCG, which captures users’ preferences and current situations. Top-K Recommendation In this paper, we exploit a 3-way tensor to integrate context information. Based on this model, we propose a time-aware recommendation approach. In addition, a tensor factorization-based approach by maximizing the ranking performance measure is proposed for predicting the possible temporal-spatial correlations. SVM Supervised v.s. Unsupervised
- 10. Web Recommendation(II) Fangfang Li, Guandong Xu, Longbing Cao, Xiaozhong Fan, Zhendong Niu: CGMF: Coupled Group-Based Matrix Factorization for Recommender System. 189-198 Zhengang Wu, Liangwen Yu, Huiping Sun, Zhi Guan, Zhong Chen: Authenticating Users of Recommender Systems Using Naive Bayes. 199- 208 Junyang Rao, Aixia Jia, Yansong Feng, Dongyan Zhao: Taxonomy Based Personalized News Recommendation: Novelty and Diversity. 209-218 Xiaochi Wei, Heyan Huang, Xin Xin, Xianxiang Yang: Distinguishing Social Ties in Recommender Systems by Graph-Based Algorithms. 219-228 07/11/14 Middleware, CCNT, ZJU 10 In this paper, we propose an innovative coupled group-based matrix factorization model for recommender system by leveraging the user and item groups learned by topic modeling and incorporating couplings between users and items and within users and items. Given a recommendation list, we improve a user’s satisfaction by introducing the taxonomy based novelty and diversity metrics to include novel, but potentially related items into the list, and filter out redundant ones. The experimental results show that the coarse grained knowledge resources can help a content-based news recommender system provides accurate as well as user-oriented recommendations. ::: Case Study In this paper, we investigate the issue of distinguishing different users’ influence power in recommendation systematically. We propose to employ three graph-based algorithms (including PageRank, HITS, and heat diffusion) to distinguish and propagate the influence among the friends of an active user, and then integrate them into the factorization-based social recommendation framework. Tomoharu Iwata, Amar Shah, Zoubin Ghahramani: Discovering latent influence in online social activities via shared cascade poisson processes. 266-274, SIGKDD, 2013
- 11. Social Web (I) Nguyen Quoc Viet Hung, Nguyen Thanh Tam, Lam Ngoc Tran. An Evaluation of Aggregation Techniques in Crowdsourcing, pp, 1-15 Zhunchen Luo, jintao Tang and Ting Wang. Propagated Opinion Retrieval in Twitter Meiling Wang, Xiang Zhou, Qiuming Tao, Wei Wu. Diversifying Tag Selection Result for Tag Clouds by Enhancing both Coverage and Dissimilarity Zhiang Wu, Alfredo Cuzzocrea. Community Detection in Multi-relational Socail Networks Maria Giatsoglou, Despoina Chatzakou. Community Detection in Social Networks by Leveraging Interactions and intensities Hemank Lamba and Ramasuri Narayanam. A Novel and Model Independent Approach for Efficient Influence Maximization in Social Networks 07/11/14 Middleware, CCNT, ZJU 11 We attempt to address this challenge by introducing a novel co-ranking framework, named MutuRank. It makes full use of the mutual influence between relations and actors to transform the multi-relational network to the single-relational network. We then present GMM-NK (Gaussian Mixture Model with Neighbor Knowledge) based on local consistency principle to enhance the performance of spectral clustering process in discovering overlapping communities. In this paper we present a community detection approach for user interaction networks which exploits both their structural properties and intensity patterns. The proposed approach builds on existing graph clustering methods that identify both communities of nodes, as well as outliers. The importance of incorporating interactions’ intensity in the community detection algorithm is initially investigated by a benchmarking process on synthetic graphs. In this paper, we precisely address this problem by proposing a new framework which fuses both link and interaction data to come up with a backbone for a given social network, which can further be used for efficient influence maximization. We then conduct thorough experimentation with several real life social network datasets such as DBLP, Epinions, Digg, and Slashdot Tomoharu Iwata, Amar Shah, Zoubin Ghahramani: Discovering latent influence in online social activities via shared cascade poisson processes. 266-274, SIGKDD, 2013
- 12. Social Web (II) Lijiang Chen, Yibing Zhao, Shimin Chen. Personalized List Recommenda- tion in Twitter, pp 88-103 John Pfaltz. The Irreducible Spine of Undirected Networks Fotios Psallidas, Alexandros Ntoulas. Soc Web: Efficient Monitoring of Social Network Acivities, pp 118-136 Xiang Wang, Lele Yu, and Bin Cui. A multiple Feature Integration Model to infer occupation from Social Media Records, pp 137-150 Jinpeng Chen, Zhenyu Wu, etc. Recommending Interesting Landmarks Based on Geo-tags from Photo Sharing Sites, pp 151-159 07/11/14 Middleware, CCNT, ZJU 12 To address the challenge of bootstrapping Twitter Lists, we envision a novel tool that automatically creates personalized Twitter Lists and recommends them to users. Compared with lists created by real Twitter users, the lists generated by our algorithms achieve 73.6% similarity. Demo In this paper, we propose a comprehensive framework to infer user’s occupation from his/her social activities recorded in micro-blog message streams. A multi-source integrated classification model is set up with some fine selected features. We first identify some beneficial basic content features, and then we proceed to tailor a community discovery based latent dimension solution to extract community features. By using DFCM, we can cluster a large-scale geo-tagged web photo collection into groups (or landmarks) by location. And then, we provide more friendly and comprehensive overviews for each landmark. Subsequently, we model the users’ dynamical behaviors using the fusion user similarity, which not only captures the overview semantic similarity, but also extract the trajectory similarity and the landmark trajectory similarity. Social Media/ Video SearchBei Pan, Yu Zheng, David Wilkie and Cyrus Shahabi. Crowd Sensing of Traffic Anomalies based on Human Mobility and Social Media. SIGSPATIAL, 2013. Jing Yuan, Yu Zheng, Xing Xie. Discovering regions of different functions in a city using human mobility and POIs. SIGKDD, 2012
- 13. Web Text Mining Seema Nagar, Kanika Narang, Sameep Mehta, L. V. Subramaniam, Kuntal Dey. Topical Discussions on unstructured Microblogs: Analysis from a Geographical Perspective, pp. 160-173 Lili Yang, Chunping Li, etc. Discovering Correlated Entities from News Archives, pp. 174-187 Min Peng, Jiajia Huang, etc. High Quality Microblog Extraction Based on Multiple Features Fusion and Time Frequency Transformation, pp. 188- 201 David S. Batista, Rui Silva, Bruno Martins, etc. A Minwise Hashing Method for Addressing Relationship Extraction from Text, pp. 216-230 Roberto Rodriguez, Victor m.Pavon, Dernando Macias, etc. Generating a Conceptual Representation of a Legacy Web Application, pp. 231-240 07/11/14 Middleware, CCNT, ZJU 13 we identify and characterize topical discussions at different geographical granularities, such as countries and cities. We observe geographical localization of evolution of topical discussions. Experimental results suggest that these discussion threads tend to evolve more strongly over geographically finer granularities: they evolve more at city levels compared to country levels, and more at country levels compared to globally. We propose an extraction framework to get high quality information by considering different features globally in social media. Specially, in order to reduce computing time and improve extraction precision, some important social media features are employed and transformed into wavelet domain and fused further, to get a weighted ensemble value. A large scale of Sina microblog dataset is used to evaluate the framework’s performance.
- 14. Networks and Graphs Shanshan Huang and Xiaojun Wan. AKMiner: Domain-Specific Knowledge Graph Mining from Academic Literatures, pp. 241-255 Dayong Ye and minjie Zhang. A Study on the Evolution of Cooperation in Networks. pp 285-298 Natwar Modani, Kuntal Dey, Ritesh Gupta, Shantanu Godbole. CDR Analysis Based Telco Churn Prediction and Customer Behavior Insights: A Case Study, pp 256-269 Helan Liang, Yanhua Du, Sujian Li. An Improved Genetic Algorithm for Service Selection under Temporal Constraints in Cloud Computing, pp. 309-318 07/11/14 Middleware, CCNT, ZJU 14 In this paper, we propose a novel system called AKMiner (Academic Knowledge Miner) to automatically mine useful knowledge from the articles in a specific domain, and then visually present the knowledge graph to users. Our system consists of two major components: a) the extraction module which extracts academic concepts and relations jointly based on Markov Logic Network, and b) the visualization module which generates knowledge graphs, including concept- cloud graphs and concept relation graphs. In this paper, a self-organisation based strategy is proposed for the evolution of cooperation in networks, which can utilise the strengths of current strategies and avoid the limitations of current strategies. The proposed strategy is empirically evaluated and its good performance is exhibited. Moreover, we also theoretically find that, in static networks, the final proportion of cooperators evolved by any pure strategies fluctuates cyclically irrespective of the initial proportion of cooperators. In this case study paper, we present our experience of participating in a competitive evaluation for churn prediction and customer insights for a leading Asian telecom operator. We build a data mining model to predict churners using key performance indicators (KPI) based on customer Call Detail Records (CDR) and additional customer data available with the operator. Further, we analyze the social network formed between the (prepaid and postpaid) churners as well as the entire subscriber base. ::: Case Study
- 15. Thank You ! Q&A Thank You ! Q&A 07/11/14 15Middleware, CCNT, ZJU Summary of WISE 2013