CINET: A cyberinfrastructure for network science

ACTA POLYTECHNICA HUNGARICA, 2013

The study of complex networks has become important in several fields of science such as biology, sociology and physics. The collection of network data and the storage, analysis and visualisation of these data have become important contributors to the knowledge of programmers working in these fields. Our cxnet software connects several software packages of the Python language to make these tasks easier. One of the main goals of this development is to provide a comfortable application programming interface for students to develop their own programs. The cxnet software package is able to create the software package network of the Ubuntu Linux distribution. This network is a directed network with several types of vertices and connections. It changes quite fast and can be created easily. These properties make it an ideal object of investigation. The present paper describes some useful measures of the properties of the complex networks, the usage of the cxnet package with some examples, and our experiences in the education.

Procedia Computer Science, 2013

Network science research aims to understand the underlying properties of complex networks. Large-scale modeling and simulation is the core of network science research. Existing systems take a long time to run large network science experiments with high performance computing resources. Scientific data management systems currently lack the performance efficiency needed to support this type of computation and data-intensive research. Memoization provides the ability to index, archive, and reuse frequently requested and expensive to re-compute datasets. In this paper, a domain independent memoization service to increase the computational execution process performance within cyberinfrastructure-based systems is described. The service is formulated as an extensible memoization framework for the computational and simulation network science domains that is built on top of well-defined metadata objects. We present extensible concepts, discuss the proposed algorithm and framework architecture, and examine the flexible nature of the framework. The framework was utilized as a part of the CINET cyberinfrastructure-based digital library (DL). Our experimental results indicate an increase in the efficiency of the system and recommendation of the service's inclusion in scientific DLs.

2014

Here we introduce a novel web-infrastructure for differential network analysis. The aim of the web-site is to provide a comprehensive collection of tools for network inference, network comparison and network reproducibility analysis. Four main processes are available through the web service: the network inference process which include 11 reconstruction algorithms, the network distance process with 3 available metrics, the network stability process which includes all the network reconstruction methods and network distances and the netwok statistic process which computes the most common measures for network characterization. We introduce here a novel infrastructure which allows the userinterface logic to be separated from computing services and the asynchronous task management. Task submission is implemented mimicking the high performance computing queue submission system which allows to run multiple jobs without affecting the front-end server. Availability and Implementation: The web-site is available at https://renette.fbk.eu, the implementation is based on the django framework and Apache, with all major browsers supported. Furthermore, the whole project is Open Source under GPLv2 and the code is available on GitHub at https://github.com/MPBA/ renette for local installation.

Complexity

Network data mining has attracted a lot of attention since a large number of real-world problems have to deal with complex network data. In this paper, we present NOESIS, an open-source framework for network-based data mining. NOESIS features a large number of techniques and methods for the analysis of structural network properties, network visualization, community detection, link scoring, and link prediction. The proposed framework has been designed following solid design principles and exploits parallel computing using structured parallel programming. NOESIS also provides a stand-alone graphical user interface allowing the use of advanced software analysis techniques to users without prior programming experience. This framework is available under a BSD open-source software license.

2010

Conan is a C++ library created for the accurate and efficient modelling, inference and analysis of complex networks. It implements the generation and modification of graphs according to several published models, as well as the unexpensive computation of global and local network properties. Other features include network inference and community detection. Furthermore, Conan provides a Python interface to facilitate the

Network Science, 2013

This is the beginning of Network Science. The journal has been created because network science is exploding. As is typical for a field in formation, the discussions about its scope, contents, and foundations are intense. On these first few pages of the first issue of our new journal, we would like to share our own vision of the emerging science of networks.

2002

The transformation of armed forces towards network oriented principles poses a number of methodological challenges for the field of defence analysis. This paper presents Project Metanet, an undertaking at the Swedish Defence Research Agency, with the aim to monitor developments within the broadly defined field of Network Analysis, and outlines three specific challenges. The first two have to do with extending the tools of network reliability theory to handle the case of complex, hybrid communication systems, where the set of nodes and the state of their communication links is rapidly changing over time. The other is the problem of how to find resources in such a dynamic, distributed system using only local topological information. The paper concludes with a discussion of general-purpose Network Information Systems.

2004

ABSTRACT This paper describes the architecture of a toolkit for large scale network analysis. The toolkit and the associated webinterface provide an extremely user-friendly manner to obtain network analysis results. The code library is programmed in Java with Perl-CGI for the front-end providing fast, efficient and scaleable system. The underlying classes and methods are written in a manner so as to facilitate the easy extension of the library.

2007 High-Speed Networks Workshop, 2007

Within the framework of the Diversified Internet where architecturally different metanetworks coexist on a shared communication substrate, we present a Large-scale Scientific Metanetwork (LSM) designated for special communication needs of large-scale science. LSM emphasizes performance and security over horizontal scalability and offers two services to users: prompt-delivery service for quick transfers of bulk data and assured-capacity service for applications that require delay or rate guarantees. The prompt-delivery service relies on messagegrained scheduling to provide near-minimal average message delay while treating each individual message fairly. In support of the assured-capacity service, LSM maintains differential tree data structures to store and update advance reservations of communication capacities. Explicit accounting for real-world entities involved in communications forms a basis for secure capacity allocation in LSM.

We introduce NetworKit, an open-source software package for high-performance analysis of large complex networks. Complex networks are equally attractive and challenging targets for data mining, and novel algorithmic solutions, including parallelization, are required to handle data sets containing billions of connections. Our goal for NetworKit is to package results of our algorithm engineering efforts and put them into the hands of domain experts. NetworKit is a hybrid combining the performance of kernels written in C++ with a convenient Python frontend. The package targets shared-memory platforms with OpenMP support. The current feature set includes various analytics kernels such as connected components, diameter, clustering coefficients, community detection, k-core decomposition, degree assortativity and multiple centrality indices, as well as a collection of graph generators. Scaling to massive networks is enabled by techniques such as parallel and sampling-based approximation algorithms. In comparison with related software, we propose NetworKit as a package geared towards large networks and satisfying three important criteria: High performance, interactive workflows and integration into an ecosystem of tested tools for data analysis and scientific computation.