Multi-agent participation becomes the mainstream paradigm of modern enterprises′ service innovation process, and different actors participate in the process of service innovation and create value through interaction. Service innovation usually includes the stages of demand generation, demand processing and demand realization, and presents the characteristics of subject interaction, resource integration and process complexity. Theoretical research on the process of service innovation is mainly carried out from micro and macro perspectives. From the micro perspective, it focuses on the division of different stages of service innovation process, aiming to explore which stages are there, and how these stages are connected. The macro perspective focuses on the interaction between different subjects in the process, as well as how the subjects participate in the process, and analyzes the roles and positions of different subjects. However, despite the growing research interest in service innovation, the issue of multi-agent participation in the view of service ecosystem remains under-explored. What are the differences in the relationship between actors in different stages of service innovation? What is the knowledge mobility among different subjects? At present, these problems have not been explained clearly in academic circles, and lack of relevant empirical research. 

Service ecosystem is a new theoretical perspective to study the process of multi-agent participation in service innovation. From the perspective of service ecosystem, multi-agent participation in service innovation process is a dynamic development process that relies on personalized innovation demand, interactive innovation subjects, integrated innovation resources, etc. to build a service ecosystem, resulting in resource sharing and integration and other interactive behaviors between subjects, so as to improve service innovation performance. Enterprises should create a good atmosphere of innovation, encourage multiple actors to participate in the process of service innovation, increase the breadth and depth of innovation search, so as to obtain more heterogeneous resources and improve the performance of service innovation. Against that backdrop, we adopt a service ecosystem perspective to analyze the service innovation process of multi-agent participation, and discusses the above problems from the perspective of multi-agent participation to fill the theoretical gap, so as to expand and improve the theoretical system of service innovation.

Our research design includes two parts, model construction and research method description. In a model construction sector, a multi-agent participation model of service innovation process is constructed from the perspective of service ecosystem, and the process of service innovation is divided into three stages: service innovation leading, generation and output. There are obvious differences in the structure of subject relationship in different stages. The main body structure of the leading stage is the core action main body, and the structure is multi network type in generation stage, and then in output stage is complex system type for mature benchmarking enterprise. The model mainly reflects the dynamic characteristics of the relationship between the subjects and the system structure in different stages of service innovation, which leads to significant differences in knowledge co creation between the actors in the process of interaction, and then affects the differences in service innovation performance. In a research methodology section, we show how to selection case, and then present case data collection process, describe the whole process of software service project implementation. 

In the case, the cooperation relationship between the actors is as follows. Actor A  is the service provider, which is the core service enterprise in the service ecosystem, actor B is the software service buyer, that is, the service recipient, C is the functional subject, including C1 and C2. C1 is the ERP software supplier of A, and C2 is the implementation partner of A. This is the fifth project cooperation between A and B, and the complexity and difficulty of the project are very high. Actor A adopts the way of multi-agent participation, the purpose is to integrate external resources to provide support for project delivery and ensure the successful launch of the project.

The software service implementation process is divided into project research, scheme determination, project design, project implementation, project delivery, and operation and maintenance services. Project research and scheme determination belong to the leading stage of service innovation, project design and project implementation are the description and discussion of service innovation generation stage, project delivery and maintenance services are part of the expression of service innovation output stage.

Through case analysis, this paper puts forward the following research propositions. The breadth and depth of participation of the actors have different impacts on the performance of service innovation, that is high participation breadth is easier to promote breakthrough service innovation, and high participation depth is easier to promote incremental service innovation. Knowledge co-creation is the core pre variable in the process of service innovation, outward oriented knowledge co- creation is easier to promote breakthrough service innovation, and internal knowledge co creation is easier to promote progressive service innovation. The scope of actors′ participation is easier to promote the formation of outward oriented knowledge co-creation, and the depth of actors′ participation is easier to promote the formation of inward knowledge co-creation. 

This article makes two important theoretical contributions to the research on service innovation. First, it is to expand the general process model of service innovation proposed, based on the perspective of service ecosystem, the participants such as customers and suppliers are regarded as "actor", effectively identify the attribute characteristics such as the differential subject relationship, knowledge co creation and the participation degree of action subjects in the process of multi-agent participation in service innovation. Second, our case study found the internal influence of the participation of the actors, knowledge co-creation and service innovation performance in the process of service innovation. It not only provides a theoretical framework for the follow-up study of the internal mechanism of multi-agent participation in service innovation process, but also helps service innovation enterprises adopt effective service innovation strategies.

service ecosystem, service innovation, multi-agent participation, knowledge co-creation