مشخصات پژوهش

Community search in static graphs has been a focal point of significant research [20], but in many real-world networks—such as those in web, social media, communication, transportation, and neuroscience—the interactions between entities evolve over time. This has led to growing interest in community search within temporal networks, with several models proposed to address the dynamic nature of these interactions [21, 22]. However, these temporal models face three primary limitations: 1. Neglecting Temporal Dynamics: Many existing approaches focus on efficiently updating community patterns, such as the 𝑘-truss method [23], and evaluating community quality based on structural cohesiveness at independent timestamps. However, they overlook the evolving nature of community structure over time, missing the critical temporal properties of dynamic communities. 2. Structural Inflexibility: These methods are often constrained by predefined patterns or rules (e.g., [24]), but communities in dynamic networks are inherently flexible. Rigid rules are unlikely to generate high-quality communities because the structure of a community is fluid and subject to change. 3. Ignoring Node Attributes: Many current approaches primarily consider the structural features of networks, yet real-world networks frequently include attributes tied to the nodes themselves, which can also evolve over time. These attributes are crucial for accurate community detection but are often overlooked. In addition, even for static networks, traditional community search methods face challenges. Most of these approaches adopt a progressive method to identify communities [20], but the process requires users to iteratively adjust their queries, select appropriate attributes, and modify community sizes to achieve optimal results. This iterative nature adds complexity and can reduce the quality and efficiency of the search process.