Reusing of the past information and lessons learned helps practitioners in better management of various aspects of construction projects, such as cost estimation, planning, contracting, and design. Measuring the similarity of construction projects improves the efficiency of the existing information systems in retrieval of relevant cases. It was hypothesized that the Work Breakdown Structure (WBS) of projects contains the necessary information to measure the semantic similarity of construction projects; therefore, WBS can be used as a potential representative of the projects. In this research project, a novel method is proposed to assess the semantic similarity of projects by application of natural language processing techniques. In this method, a new project is compared with the documented as-built projects based on their WBS similarity. This method is implemented using two metrics: (1) node similarity that compares the semantics of all nodes in two WBSs; (2) structural similarity which compares the topology of the work breakdown structures. The proposed system calculates a similarity score between 0 and 1 for each metric and the combination of these two scores provides the final similarity score between a pair of WBSs, thus it could rank the similarity of the documented cases to the new project based on their final scores. Experimental results indicated that the structural similarity produced about 15 percent higher degree of retrieval precision than the node similarity.