35 results about "Influence propagation" patented technology

The invention discloses a software FMEA (failure mode and effects analysis) method based on level dependency modeling. The method includes 1, understanding an object to be analyzed completely and deeply as required, and determining an analyzing target; 2, establishing a system-grade level dependency model by utilizing outline design as a reference; 3, selecting a module to be analyzed and determine a failure mode thereof, analyzing a failure influence propagation path and a failure reason tracing path to determine a failure reason and the failure influence and provide improvements according to system-grade level dependency model reachability node analysis; 4, selecting a detailed-grade FMEA analyzing object according to a system-grade FMEA result; 5 establishing a detailed-grade level dependency model on the basis of detailed design or a pseudo-code of the selected object to be analyzed; 6, selecting key variables to be analyzed according to the detailed-grade level dependency model; 7, determining a specific failure mode of the key variables to be analyzed, and analyzing producing reasons and failure influences of the variables and providing improvements according to detailed-grade level dependency model reachability node analysis.

An optical component having a diffractive element and / or refractive element is provided. The optical component includes a photosensitive glass and / or a photosensitive glass ceramic and a plurality of structures influencing propagation of light in at least a part of the photosensitive glass and / or a photosensitive glass ceramic.

The invention discloses an author influence propagation ability prediction method based on an interest similarity model. The author influence propagation ability prediction method includes the following steps that 1, academic relation information of author literature is extracted and excavated from literature basic information of a literature database and includes author-paper writing relation and literature-literature citation relation; 2, according to the author-paper writing relation and the literature-literature citation relation, an author citation relation network and an author cooperative relation network are set up; 3, based on co-citation relation, interest similarity among authors is calculated, wherein the co-citation relation means that if two pieces of literature citrates one piece of literature, the two pieces of literature have the co-citation relation; 4, the author citation relation network and the author cooperative relation network are used for excavating author influence propagation path; interest similarity is used as path weight, and influence propagation ability is obtained through weighted calculation.

The invention discloses a dynamic social network-oriented influence maximization analysis method. The method specifically comprises the steps of (1) obtaining an activation probability, and adding a time factor into the activation probability by using power law distribution of an influence delay distribution function; (2) building an influence propagation model LAIC; (3) executing a greedy algorithm, and calculating an initial marginal income of each node by utilizing the greedy algorithm; and (4) optimizing an original greedy algorithm by using a CELF algorithm, and improving the efficiency of searching for seed nodes through sub-mode characteristics of the influence function and an influence priority queue. By analyzing the effect of the time factor in influence propagation, the power law distribution of the distribution function, consistent with real social network node degree distribution is used, and finally an excellent result and reasonable running time are achieved in selectingTOP-K nodes with highest influence, so that the problem of dynamic social network influence maximization is effectively solved.

The invention discloses a method for solving an influence maximization problem based on a user behavior propagation model. According to the method, the individual user influence is calculated based onuser behaviors of the social network, the influence propagation probability is calculated based on the individual user influence, and the maximization scope of influenced users in a special social circle is calculated based on the influence propagation probability. Compared with a propagation model based on the network topology structure, a more considerable influence node set can be more easilyacquired in the social network, nodes with the larger influence can better influence other adjacent nodes, the success probability is correspondingly larger, the individual user influence can be solved through a PageRank method based on the time distribution user liveness to effectively eliminate zombie nodes, compared with a PageRank method based on the network topology structure, the acquired influence has greater timelines and accuracy, ranking of active users can be better improved, and ranking of inactive users can be reduced.

The invention discloses a dynamic MIMO channel modeling and parameter calculating method for the internet of vehicles, the invention provides a V2V dynamic MIMO channel modeling method, and various factors influencing propagation of radio waves, such as the movement of the transceiving vehicle, the movement of the scatterers, the movement speed, the track changes are comprehensively considered; the V2V dynamic MIMO channel modeling method and the parameter calculation method provided support time-varying channel model parameters, are suitable for any three-dimensional antenna array and the propagation scene of any vehicle moving track, and the continuity of the output channel fading phase can be ensured.

The present invention discloses a method for selecting initial users enabling social network cooperative influence maximization. According to the method, a cooperative influence propagation graph can be obtained according to respective influence propagation graphs of two kinds of commodities and the association rules of the two kinds of commodities; initial users are selected one by one; when the initial users are selected each time, each user in a non-initial user set is added to an initial user set, so that a plurality of alternative initial user sets can be formed; the propagation models of each alternative initial user are built; the ultimate benefit of each alternative initial user set is calculated; and users with maximum ultimate benefit values are added into the initial user set until the number of the initial users satisfies a requirement. According to the method of the invention, the influence propagation graph of the two commodities purchased by users is obtained based on the association rules of the commodities bought by the customers, so that the initial users who can maximize cooperative influence can be selected, and therefore, the promotion of the two kinds of commodities can be better realized.

The invention provides a power distribution network switch state identification method based on a probability graph model, and the method comprises the steps: enabling a plurality of interconnected power distribution transformers to be equivalent to a load group, and obtaining a simplified circuit diagram of a physical model of a power distribution network; analyzing the dependency relationship between the voltage correlation between the load groups and the switch state in the power distribution network, and constructing a probability graph model taking the voltage correlation and the switch state as nodes; calculating the initial probability distribution of each node and the conditional probability distribution between the nodes based on the historical operation data of the power distribution network, and completing the learning of the probability graph model; analyzing influence propagation among the nodes in the probability graph model, and determining necessary observation variables, so that the states of the rest nodes in the network can be deduced through effective traces; under the condition that necessary observation variables can be observed, the switch state of the wholepower distribution network is obtained through a confidence coefficient propagation algorithm. According to the method, the operation state of the whole power distribution network can be deduced by utilizing an artificial intelligence algorithm under the condition that part of power distribution transformer data is difficult to obtain.

The invention provides an influence maximization method and system based on a group in a social network, and the method comprises the steps: 1, mapping a node to a representation space through a random walk method in the social network, and keeping the influence propagation attribute of the node; 2, defining and calculating the propagation affinity between nodes, combining the adjacent node pairswith the highest propagation affinity in sequence until a set compression ratio is met so as to obtain a coarsened network, wherein each node corresponds to one group in the original network; and 3, constructing an influence propagation function of a candidate seed set according to attributes of propagation of the influence of the nodes in the groups and between the groups, and selecting a maximuminfluence user set containing a preset number of nodes according to a greedy algorithm. The method has higher time efficiency under a similar influence propagation effect, and has a better influencepropagation effect under the similar time efficiency.

The invention provides a method of building a learning incentive mechanism of an online community learning system. The method comprises the steps of obtaining a condition that a user uses the system by carrying out comprehensive assessment on a user activity in the learning system which takes an curriculum as a community, wherein the user activity is divided into two parts: the number of resources downloaded by a user and the number of problems answered by the user which form a user influence propagation map; obtaining a user influence propagation matrix by combining an indirect influence between a friend relationship of the user and the user; calculating the reputation gain of the user as the regular grade of the user in the online learning system by adopting a similar Page Rank algorithm designed by the user, so as to activate the user to join more activities of the curriculum community to improve the reputation gain of the user.

The invention provides a method for detecting an influence maximization node in a social network. The method comprises the steps of obtaining a network model; traversing all nodes in the network model, and calculating an influence propagation expected value of each node in a second-order neighbor range; by taking the influence propagation expected values of the nodes as indexes, correspondingly putting all the nodes into a big top heap which is initially empty in sequence; popping up a node with the maximum influence expected value from the top of the reactor, adding the node into a seed nodeset, and updating the influence expected values of all neighbor nodes of the node; and calculating the similarity between the node and all neighbor nodes, performing re-insertion operation on the neighbor nodes of which the similarity is not greater than a preset threshold in the large top heap until k nodes are popped up in the heap, and outputting the nodes as an influence maximization seed nodeset. By implementing the method, the calculation complexity is low, the efficiency is high, and the problem of influence coverage overlapping can be solved.

The invention discloses a method for determining the change impact of software module based on dynamic simulation of complex networks, which belongs to the field of software complex networks. The method comprises the following steps of: first statically scanning the source code of the target software, and constructing a network of software attribute methods; then identifying the granularity of thesoftware module according to the actual needs, constructing a software complex network model, and dynamically simulating the influence propagation of the change after the determination of the software module to obtain the quantized value of the degree of influence of each node and the scope of the change impact; finally, summing the node with the attenuation coefficient of each change propagationgeneration as the quantized value of the degree of influence of the node, and depending on this quantized value to acquire a visualization result graph of change impact level of the software network.According to the method for determining a change impact of software module based on dynamic simulation of complex networks, considering the change propagation characteristics and the node characteristics are comprehensively considered for dynamic simulation, the entire measurement analysis process can be implemented in the background, and is established in a fully automated process to ensure theminimum human and time costs are reduced, by utilized the attenuation process of the change impact.

The invention discloses a social recommendation method based on social influence propagation, which comprises the following steps: 1. constructing a scoring matrix of a user to an article and a socialrelationship matrix between users; 2. constructing an initial characteristic matrix of that user and the article; 3. the contribution of user's social influence propagation to user fusion feature matrix being obtained according to K-times evolution; 4. calculating the contribution of the item to the user fusion feature matrix according to the item scored by the user history; 5. getting the predictive scoring matrix from users to articles by the inner product of the matrices. The invention can alleviate the data sparsity problem in the traditional recommendation model based on the social influence, calculate the contribution of the social influence propagation according to a plurality of evolutionary operations, realize the accurate modeling of the user fusion characteristic matrix, thereby realizing the accurate item recommendation to the user.

The invention discloses a dynamic influence maximization method based on cohesion entropy. The method specifically comprises the following steps: 1) proposing a CeCOPRA algorithm to perform overlapping community discovery on a social network; 2) selecting potential nodes in the accumulation area to construct a candidate seed set; (3) proposing a selectable dynamic influence propagation algorithm,calculating to obtain cohesion between adjacent nodes by utilizing various entropies, and determining whether the node has the capability to become a propagable precursor of another node or not, so that the information is continuously and effectively diffused; and 4) finally, verifying whether the DEIM algorithm can successfully influence the ideal number of users in different scenes or not through multiple experiments on multiple data sets. According to the method, edge nodes in the network can be filtered, the seed node selection range is reduced, the efficiency can be greatly improved, theindividual autonomy is reserved, and the information propagation process is more real.

The invention discloses a method for detecting weak connection overlapping communities, which comprises the steps of receiving a community detection graph model, and dividing graph communities by combining the number of processors; calculating an influence propagation model of each subarea, wherein the influence propagation model is used for adjusting the edge weight of the approximate active edgeof each subarea in combination with the domain edge density; and detecting the weakly connected overlapping communities by using a time interaction bias algorithm. According to the method, the influence propagation model is provided by predicting the future active trend of the users with influence, the high-frequency interaction of the users can be determined, the influence of the users on the adjacent users can be determined, the weak connection users are ensured to still have opportunities to be brought into the community, and the community detection accuracy is improved; meanwhile, by considering the structure of the community, the invention provides a time interaction bias (TIB) community detection method based on overlapping community detection, so as to obtain better overlapping community detection performance.

The embodiment of the invention provides a social account influence evaluation method and device and a storage medium. The method comprises the following steps: obtaining a plurality of article association relationships existing in a social application platform, namely association relationships such as forwarding relationships and reference relationships among articles published by social accounts, so as to carry out influence propagation operation on the social accounts, obtaining influence evaluation parameters of the corresponding social accounts and storing the influence evaluation parameters. So visible, Athe leader value of the social account having an article association relationship with a plurality of social accounts is considered, one artificial intelligence related article is taken as an example, the weights of the academic leader character likes and the common user likes of the article are distinguished, and the accuracy and reliability of calculating the influence of the social account are improved.

The invention discloses a software FMEA method based on hierarchical dependency modeling, including the first step, proceeding from the requirements, comprehensively and deeply understanding the analyzed object, and determining the analysis target; the second step, taking the outline design as a reference, constructing the system The third step is to select the module to be analyzed and determine its failure mode, analyze the reachability nodes of the system-level dependency model, analyze the propagation path of failure effect and the path of failure cause tracking, and determine the cause of failure and the effect of failure. Propose improvement measures; the fourth step is to select the detailed level FMEA analysis object according to the system level FMEA results; the fifth step is to build a detailed level dependency model based on the detailed design or pseudo code of the selected analysis object; the sixth step is to According to the detailed level dependence model, select the key variable to be analyzed; the seventh step is to determine the specific failure mode of the variable to be analyzed, analyze the accessibility node of the detailed level dependence model, analyze the cause of the variable and the failure effect, and propose improvement measures .

The invention relates to an influence maximization method based on the multi-layer potential and community structure. According to the method, the influence propagation has two stages: multi-layer-potential-based expansion between communities at a first stage and influence propagation in communities at a second stage. To be specific, at the first stage, a seed node v tries to activate a neighbour node in a non-activated state, wherein the neighbour node is expresses as follows: {u|u belonging to N(v), active(u)=0}; the activated node during the process is marked as S1 (shown as a formula), wherein N(S) meets the following formula: N(S)=U(v belonging to S) N(v); and the S1 tries to activate another neighbour node in a non-activated state again, wherein the neighbour node is expresses as follows: {u|u belonging to N(S1)\S, activate(u)=0}, wherein the activated node is marked as S2. At the second stage, for a node expressed by a condition that any v belongs to the S2, the influence range of the node is limited in a communication where the node is located; for any communication Ci belonging to C, the influence scale depends on two factors: the value |Ci| of the community Ci and the numbe being intersection of the S2 and the Ci of the nodes, falling into the community, of the S2. According to the influence maximization method disclosed by the invention, the efficiency and the accuracy of the method are higher than thoes of the existing latest algorithms like an IPA algorithm and other heuristic algorithms.

The invention relates to a configuration item association and association graph display method and system, and is used for creating a plurality of configuration items, including setting configurationitem types and setting association relationships among a plurality of different configuration item types, the association relationships of the configuration items can be modified in a user-defined manner, and each configuration item is associated with one or more configuration items; creating an influence propagation mechanism based on the association relationship of the configuration items, and when one or more of the plurality of associated configuration items are changed; performing one or more configuration item states associated with the changed configuration items according to the influence propagation mechanism; and carrying out real-time display and user-defined editing on the running state of the configuration item by using an interactive interface. The method has the beneficial effects that association relationships among all configuration items are allowed to be conveniently established in various ways; allowing to define an impact propagation mechanism between the configuration items; and the incidence relation and the influence relation between the configuration items are allowed to be displayed in a very intuitive and convenient-to-analyze manner.

The invention discloses an influence maximization method and device based on a three-hop speed attenuation propagation model. The method comprises the following steps: abstracting a social network into a directed graph data structure G (V, E), g represents a directed graph, v represents a set of nodes, wherein the node represents a user or a group; e represents a set of edges, wherein each edge represents the relationship between two nodes; scanning graph G, obtaining the out-degree w of each node and storing the out-degree w; defining a distance attenuation factor mu and a time attenuation factor; defining an influence propagation rate v according to the mu and the time attenuation factor; defining a node influence target function sigma according to mu; propagating in a graph G accordingto the influence propagation rate v; constructing a three-hop propagation path of each node, quantifying the influence of each node by utilizing an influence objective function, performing algorithm iteration, searching for and outputting the first z nodes with the maximum influence, Meanwhile, further discloses an influence maximization node calculation device based on a three-hop speed attenuation propagation model.

The invention provides an influence propagation relation model construction and alarm influence evaluation method, computer equipment and a storage medium, and the method comprises the steps of determining a root cause alarm template, a secondary root cause alarm template, a tail end alarm template and an independent alarm template according to an alarm work order; constructing a state quantity Q = {R, M, L, S} according to the root cause alarm template, the secondary root cause alarm template, the tail end alarm template and the independent alarm template; constructing an observation set V, wherein the observation set V comprises a plurality of reference alarm templates; and based on the observation set V and the state quantity Q, training the hidden Markov model according to a preset training algorithm. According to the invention, based on the influence propagation relation model obtained through training, the subsequent influence corresponding to the alarm generated by the equipment can be evaluated to determine the influence degree, so that the operation and maintenance efficiency is effectively improved.

The invention discloses a method and device for determining an infection source in a propagation network, and the method comprises the steps: firstly, building an influence propagation model to calculate the infection probability of each node in the network at different moments when different nodes are used as infection sources; calculating the probability of generating collected observation datawhen different nodes serve as infection sources according to the probability that the nodes are infected; and finally, measuring the possibility that the node is a propagation source according to theprobability that the node generates observation data, so that a final propagation infection source is selected. The situation that the structure of a propagation network is a tree-shaped propagation network is avoided , the method is suitable for a common propagation network structure, meanwhile, the operation efficiency of the algorithm is high, and the problem of infection source inference in the propagation network can be rapidly and effectively solved.

The invention discloses an academic team influence propagation prediction method and a device, and a storage medium. Internal organization structure characteristics and external propagation characteristics of the team are considered; a community node vector in a cooperative relationship network with an author cooperation relationship is combined with a propagation node vector in an influence propagation directed network with an author reference relationship; a deep full-connection neural network link prediction model is adopted to realize propagation prediction of team influence; according tothe method, the deep learning model is used for solving the team influence propagation link prediction problem in the complex network, the universality is good, and the limitation of link prediction of a traditional method and the defect that an existing deep learning method is difficult to predict the team influence propagation direction are overcome. Experiments of a real data set show that themethod has high prediction accuracy.

The invention discloses an initial user selection method for competition influence propagation in a social network with attributes. The method comprises the following steps: constructing an undirected graph of the social network with attributes according to the data in the social network, constructing a kernel-attribute-tree according to the undirected graph of the social network, for a commodity A of a needed initial user, under the condition that an initial user set IB of a competitive commodity B in the social network is known, obtaining a candidate user set CSA of the commodity A, calculating marginal influence propagation values after each candidate user is used as the initial user based on a competitive influence propagation model, selecting the current candidate user with the maximum marginal influence propagation value as the initial user, and so on until the number of the selected initial users meets the requirement. According to the invention, initial user selection of commodities can be realized more effectively in a competitive propagation environment.

The present invention discloses a complex network node influence maximization method based on a depth auto-encoder. The complex network node influence maximization problem is solved. The method comprises the steps of: constructing complex network data; determining initial influence of each node of a complex network; estimating influence propagation values in two layers of ranges; constructing a complex network influence approximate matrix; constructing a coding / decoding model; training the coding / decoding model, optimizing a target function of the coding / decoding model, and obtaining influencefeatures of each node of the complex network; selecting the influence node set of the complex network. Different from other technologies avariciously selecting nodes, the complex network node influence maximization method selects a depth auto-encoder to effectively execute the deep influence features of the complex network nodes and find a node set with potential influence through adoption of anunsupervised method, and the selected nodes can cause large-scale influence propagation. The complex network node influence maximization method based on a depth auto-encoder can be applied to variouscomplex networks in a real world.

The invention discloses a non-contact man-machine interactive space 3D display device and a display control method. A composite technique of combining infrared control and image identification is adopted. An infrared control technique is that infrared ray signals received by a receiver are changed through influencing propagation of infrared rays, further digital signals output by the receiver generate hopping transition, the signals are input in a computer 4, and therefore, corresponding assemblies of a video player are controlled. The image identification technique is that when fingers go deep into a special region, an image collection camera photographs images of collected gestures and transmits the images to the computer through a USB interface, a built-in function is invoked by the computer for identifying and analyzing the transmitted images and changing the images on the display screen, and a purpose of controlling the images by an audience is realized. The method and the device have the advantages that non-contact interaction is realized by using simple and available raw materials, the numbers of transmitting tubes and receiving diodes are increased, more complex gestures can be identified, and the method and the device has great development prospects.

The invention discloses an establishment method of an influence propagation model based on viewpoint vectorization. An existing method has limitation, a model cannot effectively calculate the influence among users, and a certain effect can only be achieved in a single social network or group environment generally. The method comprises the following steps: crawling related information of original content of a user in a social network, and establishing a propagation network based on viewpoint influence; searching a local influence user set consistent with the user viewpoint in the influence propagation network by utilizing random walk; searching a global influence user set similar to the user viewpoints in the influence propagation network according to the viewpoint similarity; searching a user set opposite to the user viewpoint in the influence propagation network; and finally, establishing an influence propagation model based on viewpoints. According to the method, the influence user set opposite to the user viewpoint is considered, and the influence propagation model based on the viewpoint can be more accurately reflected.

The invention discloses a social recommendation method based on social influence propagation, the steps of which include: 1. Constructing a user's rating matrix for items and a social relationship matrix between users; 2. Constructing an initial feature matrix of users and items; 3. According to the K evolution, the contribution of the user's social influence propagation to the user fusion feature matrix is ​​obtained; 4. The contribution of the item to the user fusion feature matrix is ​​calculated according to the items rated by the user's history; 5. The user's contribution to the item is obtained through the matrix inner product operation Predicted scoring matrix. The present invention can alleviate the data sparsity problem in the traditional recommendation model based on social influence, and calculate the contribution of social influence propagation according to multiple evolution operations at the same time, realize the accurate modeling of user fusion feature matrix, and thus realize accurate item selection for users recommend.

The invention provides a method for detecting nodes with maximum influence in a social network, which includes obtaining a network model; traversing all nodes in the network model, and calculating the expected value of influence propagation of each node within the range of second-order neighbors; Propagate the expected value as an indicator, and put all the nodes into the initially empty big top heap in turn; pop the node with the largest expected influence value from the top of the heap to join the seed node set, and update the expected influence value of all neighbor nodes of this node; calculate The similarity between this node and all neighbor nodes, in the big top heap, reinsert the neighbor nodes whose similarity is not greater than the preset threshold, until k nodes pop up in the heap, and these nodes are output as the set of influence maximization seed nodes . The implementation of the present invention not only has low computational complexity and high efficiency, but also solves the problem of overlapping influence coverage.

The invention discloses an online influence maximization method independent of a network structure. The method comprises the following steps: firstly, randomly selecting a seed node from a node set which is not selected as a seed node; using the current seed node for executing an influence propagation process on a real network for b times, and calculating the influence accessibility of the seed node; inferring an activation probability between the nodes according to all historical infection state results; simulating an influence propagation process b times on the inference network, and estimating the influence accessibility of each node which is not selected as a seed node; identifying the first k nodes with the maximum influence according to the obtained influence accessibility by using a greedy algorithm; and finally, if the node with the maximum influence is not selected yet, selecting the node as a new seed node for repetition, otherwise, randomly selecting and repeating the process. Influence reachability estimation can be carried out under the condition that a network structure is unknown, meanwhile, only of seed and non-seed nodes is updated, and a group of k nodes with the most influence are identified more accurately.