Today, companies are seeking effective approaches to improve their innovation cycle time . Among them, it is possible to mention Inventive Design Methodology (IDM) as a TRIZ-based systematic inventive design process . However, the application of this approach is time-consuming due to requesting a complete map of a problem situation at the initial phase of the inventive design process. To solve this drawback, the Inverse Problem Graph (IPG) method has been developed to increase the agility of the process. Nevertheless, authors of IPG did not mention how the designers could achieve the innovative solutions by using the formulated problems. The purpose of the research presented in this article is to integrate the doc2vec method and machine learning text classification algorithms as Artificial Intelligence methods into the IPG process. This integration helps introduce an automatic approach for the inventive design process, helping to formulate the contradictions among TRIZ parameters in the contradiction matrix and extract the inventive principles in their intersection. The capability of the proposed methodology is finally tested through its application in a case study.