Transforming and integrating heterogeneous datasets into structured and semantically enriched data models remains a critical challenge in data and knowledge engineering. Addressing this challenge requires a systematic approach, as data and knowledge integration is often the most time-consuming phase. This work proposes a model to identify the key dimensions for establishing a method to infer Domain Reference Models. Furthermore, this work also proposes how such method can be decomposed into components and steps that encompass data normalization, advanced knowledge and data alignment, and graph-based representations, in order to infer and integrate Domain Reference Models.