Journal of Computer Science and Technology
国内计算机科学与技术的英文期刊，SCI影响因子0.650，计算机综合类
http://jcst.ict.ac.cn:8080/jcst/EN/volumn/home.shtml
收到26篇论文，接收了8篇，接收率31%，
如果不包括3篇邀请论文，那么投稿23篇，接收5篇，真实接收率22%。
论文下周正式刊出，那个时候给摘要和链接。
-------------------------------------------------------------------
Special Issue: Community Analysis and Information Recommendation
In the past few years, we have witnessed rapid proliferation of online social network services, such as Facebook, Twitter, and Del.icio.us, which greatly facilitate the collaboration, sharing, and other kinds of interactions among individuals. The term ‘social media’ has thus been coined to embrace all those new collaborative services or applications and also to indicate a new “social” approach to generating and distributing Web contents. The ever-increasing social network services produce large-scale social data for analyzing user behaviors, and then improving existing services. On the other hand, they also bring forth more severe information overload and require more flexible and robust computing platforms or service paradigms.
Generally, social network services mainly focus on two kinds of user needs, namely, social needs and information needs. Community is the typical outcome of users’ social needs and has practically become the brand of social network services. The research on community has become an important focus of social network analysis. As to users’ information needs, information overload poses a big challenge to the success of social network services. As an important approach to information filtering, information recommendation is thus broadly adopted in industrial and academic communities. This special issue aims to report the state-of-the-art and in-progress research on various aspects of community analysis and information recommendation. It totally received 26 submissions. Through a rigorous review process, where each submission is carefully peer-reviewed by at least three reviewers, eight high-quality papers are finally selected for inclusion in this special issue. The detailed contributions of accepted papers are outlined below.
As communities often play important roles in network systems, detecting communities have been widely viewed as an efficient way to identifying the key functions and their working mechanisms of the systems. A variety of community detection techniques have already been proposed in the open literature. Existing community detection approaches are usually based on the optimization of a priori decision, namely, a pre-defined quality function. Unlike the existing ones, in this special issue Shi et al. propose a posteriori decision based community detection approach that consists of two phases, namely, search phase and decision phase. The search phase finds a set of so-called Pareto optimal solutions that actually represent a group of community candidates at different scales using a multi-objective evolutionary algorithm, whilst the decision phase employs three model selection criteria and a possibility matrix method to identify actual communities. Liu and Murata study the community detection issue of K-partite K-uniform (hyper) networks, where each (hyper)edge is a K-tuple composed of K nodes from K different groups, respectively. They first propose a quality function for measuring the goodness of partitions of K-partite K-uniform (hyper)networks into communities, based on which they further present a comprehensive, parameter-free, and scalable community detection approach.
So far, information recommendation has emerged as a feasible and efficient way for overcoming the severe information overload situation and has thus been widely adopted in various areas, such as, news recommendation and music recommendation. Due to its timeliness and particularly its convenience, more and more people prefer to reading news online instead of from news papers. However, given the gigantic amount of online news, a challenging issue is the personalization of news recommendation, i.e., how to appropriately and efficiently recommend news to readers, which can match their reading preference as much as possible. In this special issue, Li et al. conduct a comprehensive study on existing personalized news recommenders. Through an empirical study, they evaluate the effects of different factors on personalized news recommendation. Besides, as most existing recommender systems tend to recommend popular items, which does not work well for brand-new users, Zhou et al. develop a heuristic based method for recommending music by focusing on identifying users’ personality as soon as possible. They verify that the developed method is more suitable for brand-new users than existing ones.
Recently, social tags with unambiguous semantics have been extensively exploited to make online recommendations. Based on this observation, Zhang, Zhang and Gao propose a random walk based model to capture the semantics of those popular tags, on which no consensus semantics has been reached, using the semantics of unpopular ones. Zhang, Zhou, and Zhang summarize the recent research progress on tag-aware recommendation systems and discuss the challenges confronted by tag-aware recommendation algorithms. They particularly survey the contributions of three mainstream methods of tag-aware recommendation systems, namely, network-based methods, tensor-based methods, and topic-based methods. Upon observing that trust based recommender systems suffer from novel recommendation attacks, which are different from the profile injection attacks in traditional recommender systems, Zhang addresses, for the first time, this problem and finds that "victim" nodes play a significant role in the attacks. Furthermore, he proposes a data provenance method for tracing malicious users and identifying victim nodes as distrust users of recommender systems. Holme and Huss point that the feedback loop from what we do, to the information this produces, to decisions what to do next, will likely be an increasingly important factor in human behavior on all levels from individuals to societies. They review some effects of this feedback and discuss how to understand and exploit them beyond mapping them on more well-understood phenomena.
In summary, it can be noted from the papers included in this special issue that both community analysis and information recommendation have been attracting more efforts from social media related research communities. The research on community detection and analysis not only unceasingly intends to propose fast detection algorithms, but also begin to take the special features of networks into account in order to invent network-topology-dependent, but more efficient, detection algorithms. As to information recommendation, besides proposing recommendation approaches suitable for different application fields, some more concrete research issues (e.g., recommendation attacks) have recently been identified and already begun to be investigated.
Finally, as the guest-editors, we highly appreciate the Editor-in-Chief, Professor Guojie Li, for hosting this special issue in the Journal of Computer Science and Technology and for his strong support and helpful guidance. We are very grateful to the editorial office staff of the journal for their excellent job during the course of preparing this special issue. We would like to express our heartfelt thanks to the authors for their contributions, including those whose papers were not included. Our special thanks go to the reviewers for their valuable time and careful evaluations on the manuscripts.
Guest Editors
Xue-Qi Cheng
Institute of Computing Technology,
Chinese Academy of Science, China
Department of Computer Science
Wayne State University, USA
Tao Zhou
School of Computer Science and Engineering
University of Electronic Science and Technology of China
