Not logged in.

Contribution Details

Type Other Publication
Scope Discipline-based scholarship
Title Implementation and Evaluation of the SciMantic’s Semantic Content Infrastructure
Organization Unit
Authors
  • Hasan Hasan
Language
  • English
How Published SciMantic Project Public Deliverable
Date 2011
Abstract Text The technical results of SciMantic have been presented in D2.1 (Functional Requirements and Analysis of Mechanisms for a Semantic Content Infrastructure) and D2.2 (Architecture of a Secure and Scalable Semantic Content Infrastructure). This deliverable, D3.1 (Implementation and Evaluation of the SciMantic’s Semantic Content Infrastructure), is the last technical deliverable in this series which completes the design work in the previous deliverable with implementation and evaluation of the SciMantic’s Semantic Content Infrastructure (SCI). Although some refinements in the implementation to achieve modularity and better performance have been made, the implementation has principally been straightforward according to the design. This deliverable does not only focus on the API descriptions of services provided by the SCI, but also on the service implementation, where necessary, to enable developers to understand mechanisms used and potentially extend or improve the SCI implementation in future. The SCI implementation consists basically of OSGi bundles which can be divided into the following categories (groups): third party bundles, Apache Clerezza bundles, and SciMantic specific bundles. Developers implementing application logic writes application bundles which use services and functionality offered by the other bundles. The major part of implementation description in this deliverable is dedicated to Apache Clerezza bundles and SciMantic specific bundles. Apache Clerezza bundles are the result of development collaboration between SciMantic partners which see a potential in pushing further development of the resulting software by an open source community under Apache Software Foundation. SciMantic specific bundles are the results of an effort to apply Peer-to-Peer (P2P) technology in a content sharing environment which bases on Semantic Web for modeling linked distributed data. The Knowledge Sharing System (KSS) scenario defined in D2.1 has been prototypically implemented to show the benefit of combining P2P and Semantic Web technology. The results of the performance and scalability evaluation show that the KSS, implemented on top of the Apache Clerezza, is able to deliver a good response time in the range of a few hundreds of milliseconds to one second for several hundreds of semi concurrent accesses to its Web services. For this evaluation, tests were set up with negligible network delay by directly connecting the client with the server under test. Furthermore, the use of Distributed Hash Table (DHT) for storing keyword-based index of knowledge units allows for an efficient keyword-based search. The time needed to insert or retrieve a knowledge unit does not depend on the number of knowledge units already present in the DHT. However, it does grow with the growing number of nodes participating in the knowledge sharing network. Within the evaluated range of the number of participating nodes, the resulting insertion or retrieval time seems to increase linearly, if not logarithmically. Therefore, it can be concluded that the prototypical implementation of the KSS is feasible to be applied to content sharing scenarios where the Web services are not computationally intensive, and the sharing network may comprise up to several thousands of participating nodes.
Official URL http://www.csg.uzh.ch/research/previous-projects/scimantic/D3.1-SciMantic.pdf
Export BibTeX