AG Netzbasierte Informationssysteme

The DigiPolis project at AG NBI is developing a Self-organized Semantic Storage Service („S4“). A research paper written by NBI staff members Hannes Mühleisen, Tilman Walther and Prof. Robert Tolksdorf discussing the handling of different index levels within this system was just accepted for the 2011 IEEE Congress on Evolutionary Computation (CEC2011) , New Orleans, USA.
The publication is titled „Multi-Level Indexing in a Distributed Self-Organized Storage System“, here the abstract:

In many systems providing storage and retrieval operations on data, indices are used to make these operations more efficient. Distributed storage systems provide means to distribute the burden of storing and retrieving data on multiple different computers. Routing indices can answer the central question in these systems: Where should one look for a specified data item? To be able to query for different columns in a relation or different entries in tuples, indexing for multiple dimensions is necessary. Our group applies a swarm-based approach to distributed storage leading to a new class of distributed systems, which are fully self-organized in their behavior and lack any shared global data structures. Here, we research whether multiple levels of routing indices can be maintained and used in such a

The DigiPolis project at AG NBI is developing a Self-organized Semantic Storage Service („S4“). A research paper written by NBI staff members Hannes Mühleisen, Tilman Walther, Anne Augustin and Marko Harasic under the supervision of Prof. Robert Tolksdorf describing the impact of various configuration parameters on this system was just accepted for the The 6th International Workshop on Scalable Semantic Web Knowledge Base Systems (SSWS2010) at the 9th International Semantic Web Conference (ISWC2010), Shanghai, China.
The publication is entitled „Configuring a Self-Organized Semantic Storage Service“, here the abstract:

Scalability requirements for semantic stores lead to distributed hardware-independent solutions to handle and analyze massive amounts of semantic data. We use a different approach by imitating the behaviour of swarm individuals to achieve this scalability. We have implemented our concept of a Self-organized Semantic Storage Service (S4) and present preliminary evaluation results in order to investigate to what extent the performance of a distributed and swarm-based storage system is dependent on its configuration.