Despite the significant benefits of crowd ideation platforms, these introduced two main challenges: (1) many ideas generated are basic and repetitive (2) the high number of ideas generated makes it practically impossible for ideation experts to review ideas one by one in order to select novel and useful ones. A key feature to overcome these issues resides in understanding the ideas. General Knowledge Graphs describe the meaning of domain-independent terms in an computationally understandable way and therefore represent a promising solution in obtaining such meaning. In this paper, we describe our research in understanding ideas and preliminary findings in application for crowd-powered innovation.