Clients Purchasing Tendency Community Classification in E-Commerce Scenario: Multi-Feature Search for Densely Distributed Clients in Huge Network

We propose a scalable, multi-feature graph-based mining framework. It segments large populations of online users—such as those on digital commerce or social networking platforms—into well-defined communities. These communities reflect shared purchasing tendencies. The term “huge network search” in the title refers to conducting dense subgraph search on our constructed huge network. This network models the relations of massive-scale clients. Our approach captures and encodes user preferences (e.g., favored dress colors or styles). It extracts a broad range of behavioral features, including those from weakly supervised learning. A geometric feature selection process refines these attributes, ensuring that only the most informative factors are retained. This provides a precise depiction of each client’s buying behavior. User preferences are modeled as probabilistic distributions across multiple latent topics. This allows us to project complex purchasing patterns into a structured, hidden feature space. From this space, we construct a similarity-driven huge network. This network maps the relational landscape of users with comparable tastes. A dedicated community search algorithm uncovers densely interconnected subgroups. These subgroups are termed purchasing tendency communities. They exhibit consistent preference profiles, such as a collective inclination toward bright, saturated tones versus subtle, muted color palettes. To further enhance the personalization experience, the framework incorporates a tailored recommendation engine. This engine ranks and suggests products—such as clothing items or color themes—based on each userąŕs affiliation with learned communities. This ensures that suggestions are contextually appropriate and finely attuned to the individual’s behavioral signature. The method was validated on a real-world dataset of over one million users. It demonstrates the ability to reliably uncover distinct consumer communities and drive large-scale, individualized recommendation services. The results affirm the commercial and experiential value of the system. This positions it as a powerful tool for user-centric product personalization in dynamic online environments.