Grid Schedulers aggregate the Grid resource that belongs to multiple administrative domains in a transparent manner. The Grid resource providers have their own local resources policies for the resources that reflects the restrictions on the resource what they are providing to the users. However, the Grid users are submitting the user application requirements with the Quality of Service (QoS) parameters. Currently, most of the existing Grid Schedulers discover the suitable Grid resources to execute the application based on the capability of the Grid resources using the keyword based matchmaking that leads to failure of finding out the potential resources. In this paper we have proposed the semantic based resource discovery mechanism that considers both the resource and local resource policy information. The proposed work is incorporated with semantic similarity calculator that will finds out the suitable and potential resources by considering the capability of resources and user required QoS parameter of deadline, job completion time and etc. It is integrated with the existing scheduler of Community Scheduling Framework (CSF) and tested for the various test cases in the real-time experimental setup. The various performance metrics such as hit/miss ratio, matchmaking time has been measured and analyzed. The results infer that the proposed work enhances the various performance measures compared to the conventional scheduler.