We all are living in the world of clouds. User has become adaptive to latest technology trends of information and communication technologies. Cloud computing has made our lives practical and keeps providing us services like consulting, data management and storage in efficient way. Cloud computing is one of emerging areas that is prevailing in industries at grandiose rate. It is combination of Internet and centralized network servers forming a mesh called CLOUD. In this paper, various aspects of cloud computing and its applications are presented. Some differences are provided between network models of cloud computing that are governing organizations and enterprisers. For achieving fault tolerance strategy, there is need to introduce multiple clouds and resource management architecture. It is believed that existing model must be improved time to time to ensure efficient computing of tasks. The paper presents an improved version of cloud computing detailed architecture. It lists deficiencies between existing cloud information architecture and proposed ontology based architecture. Two additional modules have been introduced in model viz Query Retrieval and Query Refinement. Refinement of queries is done using Rocchio formula that extracts results based on relevance criteria i.e. by distinguishing relevant and non relevant results. They are introduced in order to get efficient indexed results after transforming user query.