Modeling Long- and Short-Term Project Relationships for Project Management Systems

In modern power grid project management, project management systems have become indispensable. However, the intricate and mutable nature of relationships among power grid projects presents significant challenges to accurate modeling using traditional approaches. To tackle this problem, we propose a sequential management model that consolidates long-term and short-term potential relationships between power grid projects, aiming to enhance the efficiency of the project management system. Specifically, we develop a project relationship network, which leverages graph convolutional neural networks and attention mechanisms to dynamically capture and integrate project relationship information. This innovative method enables a more refined representation of inter-project relationships within the power grid domain. Furthermore, to account for temporal shifts in project execution, we devise a method incorporating project temporal information to predict project progress. The method employs separate modules for long-term and short-term project execution, allowing us to distinguish between enduring and immediate impacts among power grid projects, thereby enriching the portrayal of project relationships. Experiments on public recommendation system datasets validate the efficacy of our proposed method in the context of power grid project management.