*5.1. Recommendation of the Ontology-Driven MAS Development for the Energy Domain*

Based on the review result, this paper finds out the following aspects in the ontology-driven MAS development for the energy domain should be further discussed, developed or emphasized:

• The ontology development process in MAS design

Although the importance of ontology in the energy domain has been emphasized, especially for the MAS interoperability. However, from the ontology engineering perspective, the ontology development process has not been addressed well in the MAS design, especially with the consideration of the ontology categories. This paper recommends the further work can combine the categories of ontology [98] and the ontology development process [76] into the MAS design with two aspects: multi-agent communication and MAS interoperability.

• The detail design process and realization of the ontology-driven MAS development

The selected publications well discuss the 'what' and 'why' of their designed/developed MASs. However, the 'how' is missing in the majority of the selected publication. Therefore, it is difficult for readers to re-produce their methodologies of MAS development. Therefore, this paper recommends the further work can focus on this aspect, and it is especially important for the MAS interoperability.

• Open standard implementation and adoption

Open standards for both MAS design, agent communication and energy domain are discussed in the selected publication, especially regarding the MAS interoperability. This paper finds that the MAS interoperability issue is not solely due to the inter-MAS communication barriers, and upper ontology design cannot solve this issue if the designed upper ontology or the selected open standard is not adopted by other MASs. Therefore, further work on the open standard implementation and adoption for the ontology-driven MAS development is recommended.

• Higher intelligent MAS development

The MAS interoperability is important for the distributed energy systems, and ontology improvement (upper ontology or generic ontology) seems like the only solution in the majority of the selected publications. This paper recommends the future work can consider developing higher intelligent MASs that allow the 'fuzzy communication' between MASs.

• Inter-domain MAS development

Although this paper tries to search literature in the energy domain for both electricity and heating. However, the search result only shows in the electricity domain, and the literature on MAS and ontologies for the heating sector is missing. Heating is an important subdomain in the energy sector and is also strongly connected to the electricity sector through combined heat and power generation, and electrical heating. Hence, heating should be equally addressed in the studies of MAS and ontologies for the entire energy sector. The priority for future work in this field should focus on the interoperability with further external systems and cover the simulation of other areas in the energy system, including heating. However, the inter-domain ontology design will be more complex and difficult compared to only under the electricity-related domain.

• Agent listing

Agent types, roles, and interactions are well introduced in the selected publication. Meanwhile, the domain analysis in the MAS design methodologies is introduced. Some studies have done illustrations of agents in smaller scales, e.g., [3,47,48,52]. However, a systematic approach to list all related agents with a clear MAS boundary is missing. In a MAS, agents are specialized to perform tasks based on their individual goals [133]. Meanwhile, a MAS with stakeholder listing can give a good overview of the whole system. The literature shows that there are different ways to illustrate the identified agents together with their relationships. Some authors [53,81] introduce agents with descriptions, and others [18,52] use diagrams to graphically present agents. One example of the graphical illustration is the Harmonised Electricity Market Role Model by ENTSO-E [134]. This Harmonised Electricity Market Role Model represents agents, their roles, and information flow between them. This role model provides a common definition of roles and domains employed in the electricity market. It enables a common language in the development of information interchange.

Another way to present and describe stakeholders is by using business models. The research by Xia et al. [135] investigates the Swedish mobile phone business ecosystem. The stakeholder listing is represented by the Osterwalder and Pigneur business model canvas. An overview of the agents, their interrelations, and information flows can be illustrated in the business model canvas. The homogenous setup provided by the business model canvas highlights and organizes the identified information. This simplifies the information search. Furthermore, the business model canvas can easily be extended with new stakeholders by following the canvas approach. Both stakeholder listing by diagrams and the business model canvas provide well-organized information about complex systems. The canvas approach makes it possible to include supplementary information about the stakeholders.
