Antonis Sidiropoulos

Generalized Hirsch h-index for disclosing latent facts in citation networks

What is the value of a scientist and its impact upon the scientific thinking? How can we measure the prestige of a journal or a conference? The evaluation of the scientific work of a scientist and the estimation of the quality of a journal or conference has long attracted significant interest, due to the benefits by obtaining an unbiased and fair criterion. Although it appears to be simple, defining a quality metric is not an easy task. To overcome the disadvantages of the present metrics used for ranking scientists and journals, J. E. Hirsch proposed a pioneering metric, the now famous h-index. In this article we demonstrate several inefficiencies of this index and develop a pair of generalizations and effective variants of it to deal with scientist ranking and publication forum ranking. The new citation indices are able to disclose trendsetters in scientific research, as well as researchers that constantly shape their field with their influential work, no matter how old they are. We exhibit the effectiveness and the benefits of the new indices to unfold the full potential of the h-index, with extensive experimental results obtained from the DBLP, a widely known on-line digital library.

CDNsim: A simulation tool for content distribution networks

Content distribution networks (CDNs) have gained considerable attention in the past few years. Hence there is need for developing frameworks for carrying out CDN simulations. In this article we present a modeling and simulation framework for CDNs, called CDNsim. CDNsim has been designated to provide a realistic simulation for CDNs, simulating the surrogate servers, the TCP/IP protocol, and the main CDN functions. The main advantages of this tool are its high performance, its extensibility, and its user interface, which is used to configure its parameters. CDNsim provides an automated environment for conducting experiments and extracting client, server, and network statistics. The purpose of CDNsim is to be used as a testbed for CDN evaluation and experimentation. This is quite useful to both the research community (to experiment with new CDN data management techniques), and for CDN developers (to evaluate profits on prior certain CDN installations).

A new perspective to automatically rank scientific conferences using digital libraries

Citation analysis is performed in order to evaluate authors and scientific collections, such as journals and conference proceedings. Currently, two major systems exist that perform citation analysis: Science Citation Index (SCI) by the Institute for Scientific Information (ISI) and CiteSeer by the NEC Research Institute. The SCI, mostly a manual system up until recently, is based on the notion of the ISI Impact Factor, which has been used extensively for citation analysis purposes. On the other hand the CiteSeer system is an automatically built digital library using agents technology, also based on the notion of ISI Impact Factor. In this paper, we investigate new alternative notions besides the ISI impact factor, in order to provide a novel approach aiming at ranking scientific collections. Furthermore, we present a web-based system that has been built by extracting data from the Databases and Logic Programming (DBLP) website of the University of Trier. Our system, by using the new citation metrics, emerges as a useful tool for ranking scientific collections. In this respect, some first remarks are presented, e.g. on ranking conferences related to databases.

Prefetching in Content Distribution Networks via Web Communities Identification and Outsourcing

Content distribution networks (CDNs) improve scalability and reliability, by replicating content to the “edge” of the Internet. Apart from the pure networking issues of the CDNs relevant to the establishment of the infrastructure, some very crucial data management issues must be resolved to exploit the full potential of CDNs to reduce the “last mile” latencies. A very important issue is the selection of the content to be prefetched to the CDN servers. All the approaches developed so far, assume the existence of adequate content popularity statistics to drive the prefetch decisions. Such information though, is not always available, or it is extremely volatile, turning such methods problematic. To address this issue, we develop self-adaptive techniques to select the outsourced content in a CDN infrastructure, which requires no apriori knowledge of request statistics. We identify clusters of “correlated” Web pages in a site, called Web site communities, and make these communities the basic outsourcing unit. Through a detailed simulation environment, using both real and synthetic data, we show that the proposed techniques are very robust and effective in reducing the user-perceived latency, performing very close to an unfeasible, off-line policy, which has full knowledge of the content popularity.

A latency-based object placement approach in content distribution networks

Content distribution networks (CDNs) are increasingly being used to disseminate data in todays Internet. The growing interest in CDNs is motivated by a common problem across disciplines: how does one reduce the load on the origin server and the traffic on the Internet, and ultimately improve response time to users? In this direction, crucial data management issues should be addressed. A very important issue is the optimal placement of the outsourced content to CDNs servers. Taking into account that this problem is NP complete, a heuristic method should be developed. All the approaches developed so far assume the existence of adequate popularity statistics. Such information though, is not always available, or it is extremely volatile, turning such methods problematic. This paper develops a network-adaptive, non-parameterized technique to place the outsourced content to CDNs servers, which requires no a-priori knowledge of request statistics. We place the outsourced objects to these servers with respect to the network latency that each object produces. Through a detailed simulation environment, using both real and synthetic data, we show that the proposed technique can yield up to 25% reduction in user-perceived latency, compared with other heuristic schemes which have knowledge of the content popularity.

Replication Based on Objects Load under a Content Distribution Network

Users tend to use the Internet for “resource-hungry” applications (which involve content such as video, audio on-demand and distributed data) and at the same time, more and more applications (such as e-commerce, elearning etc.) are relying on the Web. In this framework, Content Distribution Networks (CDNs) are increasingly being used to disseminate data in todays Internet, providing a delicate balance between costs for Web content providers and quality of services for Web customers. The growing interest in CDNs is motivated by a common problem across disciplines: how does one reduce the load on the origin server and the traffic on the Internet, and ultimately improve response time to users? In this direction, crucial data management issues should be addressed. A very important issue is the optimal placement of the outsourced content to CDN’s servers. Taking into account that this problem is NP complete, an heuristic method should be developed. All the approaches developed so far either take as criterion the network’s latency or the workload. This paper develops a novel technique to place the outsourced content to CDN’s servers, integrating both the latency and the load. Through a detailed simulation environment, using both real and synthetic data, we show that the proposed method can improve significantly the response time of requests while keeping the CDNs’ servers’ load at a very low level.

Ranking and identifying influential scientists versus mass producers by the Perfectionism Index

The concept of h-index has been proposed to easily assess a researcher’s performance with a single number. However, by using only this number, we lose significant information about the distribution of citations per article in an author’s publication list. In this article, we study an author’s citation curve and we define two new areas related to this curve. We call these “penalty areas”, since the greater they are, the more an author’s performance is penalized. We exploit these areas to establish new indices, namely Perfectionism Index and eXtreme Perfectionism Index (XPI), aiming at categorizing researchers in two distinct categories: “influentials” and “mass producers”; the former category produces articles which are (almost all) with high impact, and the latter category produces a lot of articles with moderate or no impact at all. Using data from Microsoft Academic Service, we evaluate the merits mainly of PI as a useful tool for scientometric studies. We establish its effectiveness into separating the scientists into influentials and mass producers; we demonstrate its robustness against self-citations, and its uncorrelation to traditional indices. Finally, we apply PI to rank prominent scientists in the areas of databases, networks and multimedia, exhibiting the strength of the index in fulfilling its design goal.

Gazing at the skyline for star scientists

Admittedly, despite the plethora of scientometric indices proposed to rank scientists, none of them can fully capture the performance and impact of a scientist, since each index quantifies only one or a few aspects of his/her multifarious performance. Therefore, the task of scientometric ranking can be seen as a multi-dimensional ranking problem, where the different indices comprise the dimensions. The application of the skyline operator comes then as a natural solution to the problem. In this article we apply the skyline operator to scientist ranking to identify those scientists whose performance cannot be surpassed by others’ with respect to all attributes. This technique can be used as a tool for short-listing distinguished researchers in case of award nomination.

The Science of Science and a Multilayer Network Approach to Scientists' Ranking

The deluge of data on scholarly output created unique opportunities for identifying the drivers of modern science, for studying career paths of scientists, and for measuring the research performance. These massive data and processing methodologies have given rise to an exciting new field, namely Science of Science (SoS) as the successor of what is called scientometrics or informetrics for many decades. Science of Science is the offspring of the fertile cooperation of many disciplines, such as network science, statistics, machine learning, mathematical analysis, sociology of science and so on. In this article, we provide a comprehensive coverage of recent advances in SoS related to network analysis, prediction and ranking, and investigate the issue of scientist ranking from a multilayer network perspective. Towards this goal, we contrast by experiments the well-known h-index and the recently proposed indicator C3-index to a generalization of PageRank for multilayer networks, namely BiPlex PageRank, which is based on solid tensor analysis. Both the obtained results and the brief survey of SoS will deepen our faith to SoS and stimulate further efforts in this transdisciplinary field.

The Rainbow over the Greek Departments of Computer Science/Engineering: a Bibliometric Study

Various scientometric indices have been proposed in an attempt to express the quantitative and qualitative characteristics of scientific output. In this paper, we revisit several scientometrics indicators and apply the Rainbow Ranking method [1] [2] to categorize the academic personnel of the Departments of Informatics (Computer Science and Engineering) of Greek Universities. The dataset consists of ~700 Greek university professors along with all the relevant data and metadata about their publications and citations as identified in Microsoft Academic Search.