Forschungs- & Entwicklungsinformationsdienst der Gemeinschaft - CORDIS

Periodic Report Summary 2 - GRAPHRULES (GraphRules: Rule Discovery, Exploration and Visualization of Collaborative Graph Structures)


Graphrules dealt with the development of new algorithms for analysis of graph datasets, particularly within an enterprise setting. We addressed issues that dealt with analytic operations, such as recommendations, from static graphs. We also examined how to maintain a highly dynamic graph particularly when dealing with streaming data.


We focused on the following topics:

a) For non-dynamic graphs, we examined how to visualize large graph networks, and how to derive graph-driven techniques for making product recommendations. The graphs that we consider encode relationships between customers and products. We try understand what are the common characteristics that lead customers to buy a particular product. We apply our notions on real-world data, obtained from the host institution.
b) For dynamic-graphs, we examined methods for compressed storage and search over the compressed data. We applied our algorithms on the challenging case of streaming network data. We also examine how to use Graphical-Processing-Units (GPUs) for speeding up the indexing and search process.


Our work on clustering and recommendation of non-dynamic graphs has been patented by our host institution. We have also published this work on international conferences.

Our method for visualization and exploration of large graphs has also been patented, and published in the proceedings of peer-reviewed international conferences and a journal.

Finally, we work on dynamic-graph maintenance and search has been published in both conferences and journals, and several related patents have been filed on behalf of the host institution.


The above research work has been disseminated in diverse research articles. The host institution has patented a big portion of the developed technology. Finally, this work has been used in several pilots within the host institution.

We expect the final impact of this work to be even greater as the developed pilots scale out to more extended geographies within the host institution.

Reported by



Scientific Research
Folgen Sie uns auf: RSS Facebook Twitter YouTube Verwaltet vom Amt für Veröffentlichungen der EU Nach oben