5 results about "Community search" patented technology

This invention provides a privacy-preserving community search method and system for graph data streams. It utilizes a secret shared domain comprised of multiple servers. Within this domain, each server can only access the first encrypted data of the graph data stream's list structure and the second encrypted data of the search parameters. Throughout the community search process, the search parameters and all information in the graph data stream except for the time of generation are processed within the encrypted domain and are not leaked to the querying server, thus achieving privacy protection during the community search process for graph data streams.

The application discloses a community search method for a graph federation, constructs a federated graph index through confusion encryption security, realizes equivalent plaintext calculation on ciphertext, completes index construction offline and completes fast retrieval online, and greatly improves the query efficiency and scalability of the federated community search under the premise of strictly guaranteeing the local graph structure privacy of each data party. Based on the uniqueness and inclusiveness of the federated k-core, the application efficiently calculates according to a predetermined tree structure, greatly improves the efficiency; by combining the domination vector and the minimum bounding rectangle, the R-Tree index about the k-vector is constructed, the efficient search is realized, and the storage cost is reduced; by traversing the equivalent relationship model between the connected components and the R-Tree index and aggregating the k-shell, the final community is obtained by decrypting and restoring the vertex set, and a reliable implementation scheme is provided for the efficient and safe community search task in the graph federation privacy data collaboration scene.

A cross-dataset universal community search method, belonging to the field of graph data mining and information retrieval technology, comprises the following steps: Step 1: Constructing the required sample set and training graph; Step 2: Using spectral awareness for feature alignment to form spectral awareness aligned training features; Step 3: Constructing hybrid token sequences for nodes in the training graph; Step 4: Encoding the token sequences using a Transformer encoder to obtain a unified community-aware representation; Step 5: Reconstructing the representation using a diffusion model; Step 6: Training the model; Step 7: Inputting the target dataset into the model obtained in Step 6 for community search to obtain the results. This invention, through the above method, allows the target dataset to be directly input into the model for inference and community search during the application stage, without the need for retraining or fine-tuning, thereby improving the efficiency and search quality of cross-dataset deployment.

The application provides a community search method, device, medium and equipment, the method comprises the following steps: according to complex network data set, the multiple graph model of dynamic attribute network is constructed, input specified parameter l, k and θ, wherein l represents sequence length threshold value, k represents support threshold value, and θ represents search score threshold value;The range of the edge time stamp in the network is determined to determine the time span, and the multiple graph model is divided into different time snapshots;In each time snapshot, the number of triangles formed by each edge in the current time snapshot is calculated as the support of the edge, and each dynamic edge is traversed, and the dynamic edge whose support is less than k or does not meet the existence in the continuous time snapshot of length l is deleted in a loop;Based on the search score threshold value θ, the remaining nodes of all time snapshots are integrated, and the (l, k, θ)‑truss structure model is obtained, and the induced subgraph of the (l, k, θ)‑truss structure model in the original multiple graph model G is taken as the result community.

The application provides a community searching method and device based on maximum clique search on uncertain graph, and relates to the field of data processing.The method comprises the following steps: obtaining an initial graph corresponding to a social network to be processed; determining reference values of each vertex in the initial graph; deleting vertices with reference values not greater than heuristic solution values in the initial graph to obtain a vertex set corresponding to the heuristic solution values, and determining an intermediate subgraph based on the vertex set; performing multiple search operations on each vertex in the intermediate subgraph, determining a target subgraph with the largest number of vertices based on candidate subgraphs obtained through the search operations, and ensuring that there is an edge between any two different vertices in the target subgraph and the product of the existence probabilities of all edges is greater than a preset probability value; and determining a target social user community in the social network to be processed based on the target subgraph.The application can search for a social user community based on uncertain correlation data between social users.