Abstract: In the context of innovation-driven development, the organization and coordination of complex scientific and technological innovations have garnered significant interest from scholars, which revealing the nature of research and innovation activities shifting from individual pursuits to a stage of collaboration among researchers and organizations. Gaining insight into the dynamic mechanisms that underpin collaboration among organizations participating in scientific and research activities is fundamental for further organizing and optimizing further collaborative innovation. Facing above challenges, this study focuses on scientific and research collaboration networks, employing social network analysis methods and stochastic actor-oriented models (SAOMs) to analyze the evolutionary characteristics and mechanism of collaborations, as well as the primary factors that influence the dynamics of scientific and research collaborations. Specifically, this research employs academic publications from year 2011 to year 2022 in the top lithography conferences to construct organizational collaboration networks and knowledge element co-occurrence networks. In this context, organizations that published academic papers are regarded as innovation entities, while knowledge elements extracted from these papers serve as nodes in the corresponding knowledge networks. We then examine the dynamic changes in collaboration relationships based on the analysis of network topology characteristics, while also considering the influence of the knowledge element characteristics held by innovative entities on the dynamic evolution of the network. in this empirical study, the findings indicate that the scientific and research collaboration network in the area of lithography displays a trend of diminishing the number of core organizations, accompanied by a strengthening of collaborative relationships among participating innovative entities, suggesting a growing emphasis on building robust partnerships within the network. The findings of the study also highlight the influence of network status and transitivity on scientific and research collaboration. Knowledge proximity, knowledge diversity, and the triadic transfer of the collaboration network positively impact the formation of collaborative relationships among innovative entities. The positive coefficient of transitivity supports the facilitative effect of triadic closure on the formation of the collaboration network, reflecting the "friends of friends" phenomenon. The beneficial effect of knowledge proximity is also notable, indicating that organizations prefer to collaborate with those sharing similar knowledge structures and content, thereby reducing communication costs, and improving communication efficiency. In contrast, a high degree of knowledge combination opportunities reduces the likelihood of collaboration in this area. The negative coefficient of network status indicates that organizations in core positions are less inclined to establish new collaborations with other entities, since they may prefer to utilize their resources rather than engage in external collaborations. Moreover, we found that higher knowledge similarity reduces communication costs between related institutions, thereby increasing the likelihood of forming cooperative relationships. The findings demonstrate that the characteristics of the knowledge network significantly determine the modes of collaboration among participating organizations. Through a detailed analysis of the dynamic evolution of collaboration networks and knowledge networks, this study provides theoretical foundations and practical guidance for optimizing collaborative strategies and network decisions for innovative entities. These insights offer important references for the purposes of evaluating the collaborative strategies and further organizing research collaborations.