AI-based model of a researcher support service

Between 2020 and 2025, researchers faced challenges such as fragmented digital platforms, information overload, and limited personalisation capabilities. This underscored the need for services capable of providing comprehensive support for research activities. The aim of this study was to develop a conceptual model of an intelligent information service focused on personalised researcher support. The proposed system architecture was built using structural modelling, functional analysis, machine learning, and natural language processing techniques. It includes modules for recommendations, virtual collaboration, event management, and automated bibliography generation. A multi-layered user model was designed, taking into account scientific interests, interaction history, and research context. The combination of semantic analysis with behavioural patterns increased recommendation relevance by 20-30%. The prototype of the system was tested in March 2025 with the participation of 15 young scientists from three Ukrainian universities. The results of the survey and practical tasks showed that the average time spent searching for relevant literature was reduced by 35%, task planning efficiency increased by 40%, and user satisfaction with the service's functionality reached 87%. Respondents highly rated the convenience of the interface (4.5 out of 5), the relevance of recommendations (4.3), and co-authoring tools (4.6). Three new academic collaborations were initiated through the co-author selection module. The data obtained confirmed the effectiveness of the model in increasing research productivity, improving collaboration, and providing personalised user support. The proposed structure allows for scaling to different disciplines and has the potential to be implemented in digital platforms focused on scientific activity