The information revolution has enabled us to acquire vast amount of data on any topic in any domain, hidden in email messages, web logs, technical papers, analyst's reports and so on. Exploitation of such data in harnessing relevant information for analysis and decision making needs to be automated. The facts and their inter-relationships can be represented as a graph. Techniques from Social Network Analysis (SNA) are used in this context. SNA models relations that exist among entities such as people, events, and organizations. The underlying graph can be quite complex even for simple applications. We propose a fuzzy theoretical approach to reduce this complexity through elimination of parallel edges. It results in a simpler graph that consolidates and derives hidden relationships between entities. A new measure of centrality called influence of an entity on the network is introduced and used to show the efficacy of consolidation operation.