49 results about "Group discovery" patented technology

A method of social networking is disclosed in which users may indicate areas of interest to them and/or search for other users with the same or similar interests. Users may specify one or more areas of interest to them and enter those interests in a database, by means of a list of “tags” or keywords, called a taglist. The values of the tags may be weighted. A user wishing to find other users with similar interests may input a search taglist which is compared with the other taglists stored in the database, and a list of the users with the closest matching taglists is returned. The method may also be used to characterize documents, projects, media files or other data objects, so that such items may be searched in similar fashion.

The invention relates to a specific group discovery and expansion method based on microblog data, and belongs to the field of social network analysis and data mining. The specific group discovery and expansion method comprises the following specific steps: collecting relevant group information; carrying out information integration and mapping; aiming at text data to carry out characteristic extraction; calculating a user similarity degree; carrying out the self-detection of a category group; and extracting the attributes of the specific group, judging a category, and carrying out group expansion. The specific group discovery and expansion method artfully avoids the problem that group identification can not be carried out since data is sparse or incomplete when a network model is used, inputs large-scale data calculation and is high in stability.

A method and apparatus for enabling user services in a communication network with a plurality of terminal users. An initiating user terminal (200) executes a group discovery process with the terminals (202) to form a user group. Capabilities are exchanged with the terminals, including at least one of: functional resources, available services, equipment information, applications or data programs, and current traffic or connection characteristics When a group service request (R) is received from a user terminal (D) in the group, functions and actions required for the requested service are assigned to functional resources in the terminals based on the exchanged terminal capabilities. The requested group service can then be consumed by means of these functional resources without involving an operator controlled group management server.

The invention belongs to the field of network data mining, and discloses a group discovery method and system for an attribute network, which are used for accurately identifying a potential group structure in the attribute network, and comprises the following steps: obtaining user interaction behavior data of the attribute network; modeling an attribute network topology and a node attribute set by preprocessing the attribute network data; positioning a potential clustering center node based on the node degree centrality measurement and the relative distance between the nodes; converting the network adjacency matrix into a similar matrix according to the topological structure information, and synthesizing a node attribute matrix at the same time; carrying out deep fusion on structure information and node attributes at the same time by using a multilayer graph convolution model, and automatically identifying a complete group structure; and finally, evaluating a group discovery result. The method and system can face large-scale attribute network data, reveals the group structure under low time complexity, is high in universality to complex networks, and has high application value.

The invention provides a common neighbor similar triangle agglomeration-based hierarchical and overlapping community discovery method applicable to a traditional Chinese medicine herbs (TCMF) network. The method comprises the following steps: 1) common neighbor similar triad agglomeration stage a: seeking all triads; b: calculating the similarity of any two triads; c: giving a similarity threshold of the triads, and merging the triads with the similarities which are higher than the similarity threshold as initial communities; and d: ending; and 2) cluster merging stage a: calculating the distance between any two initial communities; b: setting a distance threshold of the initial communities, and merging the two initial communities with the distance which is smaller than the distance threshold; and c: ending. The TCMF network-based hierarchical and overlapping core medicine group discovery method provided by the invention provides a new method for TCMF network discovery; and by adopting the method, a high overlapping and hierarchical medicine group community structure of the TCMF network can be excavated by setting three parameters alpha, beta and gamma, and a solution is provided for core medicine group discovery in prescription compatibility.

The invention belongs to the technical field of personalized recommendation and discloses a cluster group discovery-based recommendation system and method and a personalized recommendation system. According to collection of user operation behavior data by an online system, a user-project matrix is extracted; an entire user-project data set is divided into multiple subgroups that are highly internally correlated to each other; via combination of Euclidean and other similarity calculation methods and clustering algorithms, recommendation results are obtained after a collaborative filtering algorithm is used directly in the subgroups. Via use of the cluster group discovery-based recommendation system and method and the personalized recommendation system, data preprocessing operation is performed before a recommendation algorithm is used directly, fuzzy clustering is adopted for grouping the data, and a final recommendation result is obtained after predicted scores of all groups are integrated. While normal operation of the entire recommendation system is ensured, a fuzzy cluster group discovery-based data preprocessing method allows the recommendation algorithm to be applied directlyto individual groups of data that are strong in correlation instead of all data, and therefore the recommendation system can be improved in expansibility and accuracy in the face of mass data.

The invention provides a specific-group discovery method based on news data and related comment information. The method comprises the following steps: collecting the news data information and the related comment information in targeted media; classifying the news data information according to text contents thereof to obtain different class clusters; using the class cluster, which contains the newsdata information with a highest comment number, as a sample to acquire all comments of news data messages in the class cluster and users, who publish the comments, according to the related comment information; obtaining keywords through carrying out word segmentation on contents of all the comments, and using the keywords, of which occurrence frequency is higher than a threshold value, as high-frequency words; adopting a vector space model to represent the contents of the comments, clustering text of the comments through an agglomerative hierarchy, and obtaining comment user reference features of different class clusters according to a clustering result; and identifying a specific group according to the high-frequency words and the comment user reference features. A robot account can be quickly and intelligently discovered through analyzing the comment information contents, and thus processing is carried out in time.

The invention provides a suspicious group discovery method and device. The method comprises the steps of firstly obtaining a user and text contents published by the user; calculating the similarity between a large number of texts by utilizing a specific algorithm; by taking the similarity between the texts published by the user as a connection relation, establishing a similar user network; and finally discovering a suspicious group by community division of the user similar network. The similar text-based suspicious group discovery is realized.

The invention belongs to the field of graph data mining, and discloses a group discovery method, system and device for a multi-source heterogeneous relationship network, and a medium, and the method comprises the steps: obtaining interaction behavior data between any two users in the multi-source heterogeneous relationship network; obtaining an edge weight value between any two users according tothe interaction behavior data between the any two users, wherein the edge weight value is used for representing the connection closeness degree between the two users; constructing a graph structure according to the edge weight value between the any two users to obtain a multi-source heterogeneous relation graph; and performing sub-graph division on the multi-source heterogeneous relationship graphto obtain a plurality of maximum connected sub-graphs, and performing community division respectively to obtain a group discovery result of the multi-source heterogeneous relationship network. The method is low in time complexity, excellent in group discovery result, suitable for a large-scale relational network, capable of effectively reducing time resource expenditure in group discovery and improving the modularity of the discovery result, free of any priori knowledge, capable of being achieved completely through a network topology structure and high in applicability to a complex network.

The invention discloses a group discovery method based on path backtracking graph embedding, which comprises the following steps: establishing a topological graph for representing a network, selectinga node from the topological graph as an initial node, carrying out random walk to obtain a traversal node sequence, and sequentially cutting off the traversal node sequence into a plurality of traversal node sub-sequences with preset lengths; traversing nodes in the node subsequence according to each traversal node; counting backtracking to obtain the frequency of occurrence of each edge among the nodes to serve as the weight value of the edge, obtaining an edge weight matrix, expressing the nodes by adopting a randomly constructed graph embedding vector, optimizing the graph embedding vectorthrough the edge weight matrix, obtaining a graph embedding expression vector, and performing dimensionality reduction and clustering to form nodes contained in each category, namely the same group.The method has the characteristics of low calculation complexity and simple required data source, and can effectively reduce the calculation resource overhead in group discovery. The method does not need any priori knowledge, completely depends on a network topology structure, and is high in applicability to a real complex network.

Technology is disclosed herein that enhances organizational search capabilities. In an implementation, a representation of an organization graph that is associated with a user is surfaced in a user interface to an organization search application. A node in the organization graph may be selected, which may correspond to a particular person or other such entity in the graph. A set of groups may be associated with the node from which a subset may be identified as potential groups to be suggested to the user. Group suggestions may be surfaced in the user interface to the user that represent the subset of groups, thereby enhancing group discovery within an organization.

The disclosure relates to interference group discovery for a full duplex network architecture in a cable network environment. An example apparatus comprises a processor, and a memory element in communication with the processor. The processor configured to send a first message to at least a first cable modem of a plurality of cable modems. The first message indicates an upstream test signal to be generated by the first cable modem within a predetermined portion of a frequency spectrum allocated for downstream communication. The processor is further configured to send a second message to at least a second cable modem of the plurality of cable modems. The second message indicates a downstream interference measurement to be performed on the upstream test signal by the second cable modem. The processor is further configured to receive at least one interference measurement result indicative of the downstream interference measurement, and determine at least one interference group for the plurality of cable modems based upon the at least one interference measurement result.

The invention belongs to the technical field of big data, and particularly relates to a group discovery method based on data enhancement and non-negative matrix sparse decomposition. According to themethod, neighbor distribution of nodes with different attributes in a network with attributes of the nodes is learned through the recurrent neural network; a neighbor mode formed by similar neighbor distribution of a plurality of nodes is extracted to enhance data, a non-negative matrix is subjected to sparse decomposition through alternating least squares and Glynov regularization to find a group, and the group discovery accuracy is improved. According to the method, the edges of the network topology structure are filled through data enhancement, so that all-zero lines are not iterated as much as possible during matrix decomposition, and iteration stability is ensured; moreover, sparse group representation is obtained through a non-negative matrix sparse decomposition method, a large number of non-zero probabilities cannot be obtained, the interpretability of the group discovery method is enhanced, and the problem that a large number of non-zero probabilities are difficult to interpret group membership relations is solved.

The invention relates to a mail mining method based on coarsening and local overlapping modularity. The method comprises the following steps: constructing a mail contact network G according to mail header log information; repeatedly iteratively traversing, and fusing triangles in the G into a composite node to obtain a coarsened graph Gcn; initializing distances of nodes corresponding to all edges in the Gcn according to the Jaccard distance; and iteratively updating the distances between the neighbor nodes until all the distances converge; the nodes of which the distances are smaller than 1 belonging to the same group, and obtaining group division Ccn of the coarsened graph; restoring the network to obtain an initial group partition C of the mail contact network; adding the nodes with the distance of 1 into a group set with the maximum local overlapping modularity increment to obtain an overlapping group set Cover; and combining the groups of which the number of the group nodes is smaller than a set threshold value in the Cover to the group with the maximum compactness, and updating the Cover to finally obtain group division C' of the email contact network and output a final group division result. According to the method, the problem of low calculation efficiency of the traditional modularity is solved, and the precision of overlapped group discovery is improved.

The invention provides an attribute graph group discovery method based on maximum mutual information and a graph neural network. The method is characterized in that the method comprises steps of carrying out the representation learning of a to-be-processed matrix through a pre-trained graph neural network, obtaining a preliminary node representation, and carrying out the mutual information calculation of a to-be-processed attribute graph, and obtaining a global mutual information value; dividing the preliminary node representation to the centers of a plurality of groups by using soft clustering to obtain an allocation matrix; carrying out modularity and mutual information calculation in the to-be-processed attribute graph on the original group according to the allocation matrix to obtain amodularity value and group mutual information; and calculating total loss according to the modularity value, the group mutual information and the global mutual information value, and iteratively updating the graph neural network through gradient return according to the total loss till a group discovery result is obtained. According to the method, the end-to-end updating graph neural network doesnot need to be realized step by step, the node attribute relationship can be better captured, and a group discovery result with higher accuracy is obtained.

The invention discloses a large-scale micro-blog user interest group discovery method, belonging to the technical field of data mining. The method comprises the following steps of (1) acquiring data as a data source of micro-blog user interest group discovery; (2) data checking and preprocessing; (3) standardizing the data and converting the unstructured data into structured data for clustering analysis; (4) carrying out the improved SSLOK-Means clustering analysis based on Calinski-Harabasz; (5) using the CH validity discriminant function to determine the optimal number of clusters k, and constructing the model of interest group discovery. The invention makes it possible to cluster analysis and automatically determine the cluster number in the limited memory according to the big data of the micro-blog user, and provides support for the optimization of the micro-blog personalized service and the promotion of the marketing income.

The invention relates to a TDMA wireless ad hoc network different-group discovery and combination method. The method comprises the following steps: S1, each node of a main path circularly performs receiving work according to a superframe structure; s2, transmitting a beacon signaling; s3, transmitting an auxiliary beacon; s4, receiving a beacon signaling and an auxiliary beacon; s5, judging whether the received beacon signaling and the auxiliary beacon are different-group beacons or not, if yes, dismissing the ad hoc network with small network depth, performing networking again, and returning to the step S1; and if not, returning to the step S1. The method has the advantages that one beacon signaling or one auxiliary beacon can be transmitted at an average interval of one time slot, and meanwhile, any node is ensured to continuously receive the beacons at more than three time slots in an unpredetermined beacon transmitting multiframe time, so that the beacons which possibly appear at an interval of one time slot can be received at the continuous three time slot time; beacons can be found in time, and the purpose of fast integration of the ad hoc network is achieved.

The invention provides a group discovery algorithm model based on big data mining, and belongs to the technical field of big data mining. The method comprises the following steps: acquiring each pieceof trajectory data of a target, and preprocessing the trajectory data; taking each piece of effective trajectory data of the target as a midpoint, and intercepting the preprocessed data for a specified time length; recording the occurrence frequency of other people in the cut slice; preliminarily determining persons in the same row by utilizing the slices; acquiring trajectory data of a target and persons in the same row and preprocessing the trajectory data; taking each effective footprint as a starting point, and intercepting the time sequence trajectory data for a known fixed length; and calculating the occurrence frequency of all targets and persons in the same row. According to the method, people who may participate in group activities together with the targets are searched through attributes such as time, coordinates and names of people entering an area uploaded by each station and the known targets.

The invention discloses a group discovery method in police information industry. According to the method, firstly, data is subjected to data extraction, conversion and loading; an Hash table is used for storing co-concurrence data; the conversion of structured data to semi-structured data is realized; then, noise data is filtered; multi-dimensional co-concurrence data is merged to build a key atlas; the conversion from the semi-structured data to the non-structured data is realized; finally, social network analysis is performed on the key atlas; an on-map parallel clustering algorithm is used for obtaining member relationship inside groups and between groups, and is shown by a direct viewing mode of a pattern structure. The group discovery method provided by the invention has the advantages that the behavior feature data of individuals is used; a co-concurrence relationship ideal is used for building the key atlas; the on-map clustering analysis is performed; the key contact persons of a plurality of groups or between the groups can be discovered from the behavior feature data of individuals; the auxiliary support is provided for police information analysis.

The invention discloses a group discovery method based on a depth map convolutional network. The group discovery method is used for solving the problem that an existing method is low in group structure recognition rate in an attribute network. The method specifically comprises the following steps: acquiring attribute network user interaction behavior data; modeling an attribute network topology and determining a small number of node tags by preprocessing the attribute network data; pre-training existing node tags by using a tag propagation algorithm to expand a tag set; and constructing a depth map convolution model, performing deep fusion on structure information and node attributes at the same time, and automatically identifying a complete group structure. By adopting the technical scheme of the invention, mining of group features in a large-scale attribute network is facilitated, and meanwhile, the accuracy of group identification is effectively improved.

The invention discloses a social network group discovery system and method and a storage medium, and the method comprises the steps: firstly obtaining an online social network data set, and building an online social network topological graph; establishing an adjacent matrix according to the online social network topological graph, and performing dimension reduction processing on the adjacent matrix by utilizing a depth stack type auto-encoder to obtain a dimension reduction matrix; obtaining node embedding vectors by utilizing a graph embedding method; and finally, clustering the node embedding vectors to obtain a clustering result, namely a social network group discovery result. According to the invention, by obtaining the online social network data group and extracting the adjacency matrix, the description of the relationship between online social network users is realized; the integrity of an online social network structure is effectively reserved by utilizing a deep stack type auto-encoder and graph embedding, and the accuracy of a group discovery result is ensured; the node embedding vectors after dimension reduction embedding are clustered to obtain the discovery result, so that the time complexity is reduced, and the discovery result can be obtained more quickly and accurately.

The invention discloses a large-scale data group searching method based on time sequence density clustering, which comprises the following steps: on the basis of collection of hundred million nodes online, carrying out preliminary pretreatment on the collected nodes, constructing an original cluster and a clustering graph for expressing a node relationship, and finding a group where a merging node is located according to a communication relationship of representative cluster nodes. Along with each round of iteration of the algorithm, node contribution degree scores at different moments are calculated, and range query is executed on the nodes according to the score values. Under the condition that the correctness of final group discovery is ensured, the search efficiency of a large-scale data network can be well improved by adopting a time sequence-based density clustering method. By the adoption of the scheme, the key problems of high-energy physical data intensive access and diversified data query requirements can be solved by reducing I/O and cross-domain communication traffic.