Despite the benefits of as-built building information models for facilities, the generation of these models is laborious due to time-consuming and error-prone process required to compare with as-designed condition manually. Therefore, an automated solution is significantly important for the improvement of the current practice. This study presents a guideline for recognition researchers to address this limitation. The proposed method includes the investigation of common building elements in the IFC data model standard automatically, determination of property ranges of each element experimentally and the acquisition of the properties currently measured by vision data. The obtained data is then used to build a decision tree as a case study. The distinctive properties and their value ranges for common building elements are found out in the conclusion.