Having a goal of building attractive and competitive services, communication service providers use expert knowledge from various domains and face certain difficulties in an attempt to reach desired level of knowledge compatibility. Activities of such diverse teams, focused on building or managing services, often assume spending significant amount of time and effort to bring all team members to approximately same level of knowledge at which they can effectively collaborate. This thesis introduces a novel methodology for knowledge representation using granular approach for building granular knowledge structures which together with aspect formalism allow them to be reused without need for their further modifications. Any domain by using aspect, marked by its goal, application domain and activity domain can use same granular structures according to domain’s particularities. Due to great numbers of systems and network elements used in daily activities, communication service providers are well positioned to be able to collect a multitude of information that can be used for deriving knowledge. However, due to the nature of network equipment and platforms, information gathered in communication service provider’s environment is imperfect, inconsistent and incomplete. In order to minimize the impact of such limitations on deriving knowledge from acquired information, a methodology for deriving knowledge granules based on rough sets was introduced. Based on this methodology, a new system has been built and integrated in a communication service provider environment. The system was able to generate knowledge granules describing the influence of weather conditions on customer lines and in that way introduced new possibilities for existing systems and processes such as more advanced field technicians’ activities planning based on weather conditions forecasts and knowledge about influence of such conditions on customer’s line and service stability.