The Internet of Things (IoT) is driving the technology and advances of the Internet and every day it is becoming more popular to talk about IoT systems in multiple domains: Smart Cities, Agriculture, and Industrial Internet, etc. Increasingly, IoT systems will need to interact and be interconnected for offering the always-promoted everything-connected paradigm. Current IoT systems rely on semantic web technologies for integrating data and ensure web services interoperability. However there are yet a gap to ensure semantic interoperability among IoT systems. Most of the existing proposed (open) approaches and solutions lack on formal methodologies for interoperability in technology and standard format of the data. We studied and analyzed most available semantic-based IoT approaches to identify the main requirements hindering IoT semantic interoperability. In this paper, we present SEG 3.0 a methodology to federate, unify and provide semantic interoperability. SEG 3.0 emerges from methodologies for ontology engineering and the idea of unification and federated systems. We propose SEG 3.0 and apply it to Internet of Things (IoT) and particularly on use cases for smart cities as proof of concept. Firstly, we define characteristics required for the methodology. Secondly, we describe the processes and the different formal steps. Thirdly, we provide a proof of concept framework and architecture applying this methodology, thus the benefits of using SEG 3.0 methodology in IoT domains are described. Finally, we demonstrate that the SEG 3.0 methodology is applied to three use cases: (1) the M3 framework to assist developers in designing semantic-based IoT applications, (2) the VITAL EU project for smart cities, and (3) the FIESTA-IoT EU project for IoT semantic interoperability. SEG 3.0 is a formal methodology generic enough to be applied to other domains than IoT and smart cities, since the main benefit of the SEG 3.0 is integrating heterogeneous data and adding value to it to build innovative applications.