Subgraph isomorphism search is a basic task in querying graph data. It consists to find all embeddings of a query graph in a data graph. It is encountered in many real world applications that require the management of structural data such as bioinformatics and chemistry. However, Subgraph isomorphism search is an NPcomplete problem which is prohibitively expensive in both memory and time in massive graph databases. To tackle this problem, we propose a new approach based on concepts widely different from existing works. Our approach relies on a summarized representation of the graph database that minimizes both the amount space required to store data graphs and the processing time of querying them. Experimental results show that our approach performs well compared to the most efficient approach of the literature.