A Concept Graph is a graph in which nodes are concepts and edges indicate the relationship between the concepts. In these graphs, a concept is usually represented by a single term or a phrase. Statistical methods can be used for concept graph construction. These methods are language independent and computationally efficient. One of the applications of concept graphs is finding other related concepts to the user query in a context dependent manner. This set of concepts can be used for automatic or manual query expansion. In this paper we study and evaluate a statistical method for concept graph construction and utilize the concept graph for query expansion to improve the precision of retrieval systems. The Wikipedia corpus is used to construct the concept graph and CACM collection is used to investigate the usability of our concept graph for query expansion and extracting deeper information.