Refining the CDMS and employing the framework in interdisciplinary scenarios

Abstract

The Complex Networks field is a cornerstone for the analysis of emerging behavior in graph-structured data. From the pivotal introduction of the PageRank algorithm for information retrieval to the analysis of important phenomena in diverse areas such as social networks, epidemics, genomics, physics and linguistics. A fundamental concept in complex network analysis is that of network dynamics, which captures the influence of the underlying topology in the information flow. These dynamics can be represented by state values on nodes (e.g. reputation of websites as in PageRank). Complex network analysis typically relies on ad-hoc, off-line algorithms running in frameworks that offer little to none opportunities for exploratory analysis. Current graph databases cannot represent the required network dynamics due to the static nature of graph query matches.To tackle these shortcomings, we are specifying and building the Complex Data Management System (CDMS), which aims at providing a database-like interaction for Complex Network tasks. We propose data management and querying mechanisms adequate for interactive network analysis. Our query model can capture several aspects of network dynamics, which are used to compose metrics that are integrated in a declarative query language. The query language enables users to control several aspects of the intended analysis. This research exchange proposal aims at improving our current architecture in several fronts: (i) acquiring interdisciplinary data and use cases, (ii) tuning and optimizing the system in face of the new challenges, and (iii) refining our dynamics engine and assessing its effectiveness. (AU)