90 results about "Graph traversal" patented technology

Arrangements and methods for providing for the efficient implementation of ranked keyword searches on graph-structured data. Since it is difficult to directly build indexes for general schemaless graphs, conventional techniques highly rely on graph traversal in running time. The previous lack of more knowledge about graphs also resulted in great difficulties in applying pruning techniques. To address these problems, there is introduced herein a new scoring function while the block is used as an intermediate access level; the result is an opportunity to create sophisticated indexes for keyword search. Also proposed herein is a cost-balanced expansion algorithm to conduct a backward search, which provides a good theoretical guarantee in terms of the search cost.

For each of a number of landmarks in an image an initial position of the landmark is defined. Next a neighborhood around the initial position comprising a number of candidate locations of the landmark, is sampled and a cost is associated with each of the candidate locations. A cost function expressing a weighted sum of overall gray level cost and overall shape cost for all candidate locations is optimized. A segmented anatomic entity is defined as a path through a selected combination of candidate locations for which combination the cost function is optimized. During optimization towards the optimal segmented surface/volume graph traversal methods are exploited.

Embodiments described herein provide a more flexible, effective, and computationally efficient means for determining multiple intents within a natural language input. Some methods rely on specifically trained machine learning classifiers to determine multiple intents within a natural language input. These classifiers require a large amount of labelled training data in order to work effectively, and are generally only applicable to determining specific types of intents (e.g., a specifically selected set of potential inputs). In contrast, the embodiments described herein avoid the use of specifically trained classifiers by determining inferred clauses from a semantic graph of the input. This allows the methods described herein to function more efficiently and over a wider variety of potential inputs.

A software testing system uses a graph traversal algorithm to explore a model simulating a software product in order to identify errors in the software product. The model employs a Petri's net construct for maintaining state and governing transitions. In particular, the model mediates between a test driver and the software product. The model-based approach is usable both to validate the design of the software and verify the implementation of that design. Using the Petri net model, the test space is bounded.

The invention provides a bank financial atlas construction and analysis method. The method mainly comprises a financial atlas construction method and two typical analysis models based on financial atlass. The financial atlas construction method provides a universal bank data model with an atlas as a structure. The method mainly comprises three types of nodes of a bank, a client and a deposit account and five types of relations of transaction, guarantee, loan, correspondence and control among the nodes. The data in the bank relation database can be converted into the graph data, and a special graph database management system is adopted to store the graph data, so that the demand expression of associated data analysis is clearer, and the processing efficiency is greatly improved. The financial atlas analysis model comprises a consumer loan illegal entry real estate industry model and a guarantee ring model. According to the method, the analysis efficiency is effectively improved by usingtypical graph query and graph traversal modes respectively, required data can be quickly and accurately found out from massive bank data, and risk prevention of a bank and macroscopic decision of a bank supervision department are facilitated.

A finite state model-based testing system has a user interface to enable a user to enter state information and transition information about a software application to be tested. The user interface further enables a user to initiate a model generation engine to generate a model of the software application from the state information and transition information. A graph traversal menu enables a user to select a graph traversal program to generate a test sequence of inputs for the software application from the model, and a test execution menu enables a user to select a test driver program to read the test sequence of inputs for the software application, and execute the test sequence of inputs on the software application.

A system for generating and performing inference over graphs of sentences standing in directed discourse relations to one another, comprising a computer process, and a computer readable medium having computer executable instructions for providing: tree-structured encoder networks that convert an input sentence or a query into a vector representation; tree-structured decoder networks that convert a vector representation into a predicted sentence standing in a specified discourse relation to the input sentence; couplings of encoder and decoder networks that permit an input sentence and a “query” sentence to constrain a decoder network to predict a novel sentence that satisfies a specific discourse relation and thereby implements an instance of graph traversal; couplings of encoder and decoder networks that implement traversal over graphs of multiple linguistic relations, including entailment, contradiction, explanation, elaboration, contrast, and parallelism, for the purposes of answering questions or performing dialogue transitions; and a spiking neural network implementation of the aforementioned system components.

The present invention describes methods for performing large-scale graph traversal calculations on parallel processor platforms. The invention describes methods for on-the-fly hypothesis rescoring that utilizes graphic processing units (GPUs) in combination with utilizing central processing units (CPUs) of computing devices. The invention is described in one embodiment as applied to the task of large vocabulary continuous speech recognition.

The invention relates to a virtual machine distribution method based on topology partition. The method comprises building an undirected graph model according to a topological structure, achieving the topology partition then through graph traversal segmentation, calculating sub graph resources, and matching a bearing capability of a physical server. In the case that virtual machines are built in batches, the graph model is built, and virtual machine resource request information and a characteristic between a virtual network and a logistics network are comprehensively considered, so virtual machine nodes are reasonably distributed to the physical server, the resource utilization rate of the physical server is improved on the premise that virtual machine performance is guaranteed, the occupancy rate of the network bandwidth can be reduced, and then high efficiency of network communication between the virtual machines is guaranteed.

Systems and methods for using graph databases to make digital content recommendations are described. A graph database may be associated with tagged digital content. The graph database may include a node for each content tag and edges identifying a relationship between nodes. When a user accesses or searches a digital content item, the graph database may be traversed to identify and present related content recommendations to the user based on the traversed nodes. Node graph traversal may initiate from a root node associated with a tag of content item being presented to a user. Nodes may be searched based on their proximity to the root node as determined by edge connections of the graph. Initially searched nodes may include nodes in the graph database directly connected by edges to the root node. Subsequently searched nodes may include nodes one edge connected removed from the root node, and so forth.

The invention belongs to the technical field of network security, and particularly relates to a cross-architecture binary function similarity detection method and system based on a neural network, andthe method comprises the steps: traversing a binary file function list for different types of binary files, constructing and optimizing a function control flow graph, translating the program basic block byte code to obtain an intermediate representation, generating a semantic embedding vector of the basic block code, extracting function control flow graph nodes by using a breadth-first graph traversal algorithm, obtaining function embedding vectors according to semantic embedding vectors and control flow information of the nodes, and calculating the cosine distance between the function embedding vectors to measure the function similarity. The method is more beneficial to code intermediate representation, eliminates the difference between different instruction architectures, reduces the cross-architecture code similarity detection difficulty, and reduces the expansion workload and difficulty, based on a function embedding process of a PVDM model and a graph neural network, introductionof human prejudice is avoided, the improved graph neural network is faster in convergence speed and higher in overall efficiency and accuracy of the system.

A computer architecture for graph-traversal provides a processor for bottom-up sequencing through the graph data according to vertex degree. This ordered sequencing reduces redundant edge checks. In one embodiment, vertex adjacency data describing the graph may be allocated among different memory structures in the memory hierarchy to provide faster access to vertex data associated with vertices of higher degree reducing data access time. The adjacency data also may be coded to provide higher compression in memory of vertex data having high vertex degree.

The invention discloses a classified variable clustering method based on the attribute weight similarity. On the basis of the attribute weight similarity, the clustering process is converted into a process of searching for graph connected components, data points of data concentration serve as nodes, when the attribute weight similarity of two data points of data concentration is larger than or equal to theta, it is considered that one connecting line exists between the two data points (the parameter theta is given in advance), and when the attribute weight similarity of two data points of data concentration is smaller than theta, it is considered that no connecting line exists between the two data points. After an undirected graph is determined, each connected component of the undirected graph is a cluster, and the records in the clusters are peaks of the connected components. The classified variable clustering method based on the attribute weight similarity is substantially used for searching for the peaks contained in the connected components of the undirected graph, the clustering process can be guided through the thought of a graph traversal algorithm, and therefore the time and space complexity is low, and the clustering result is high in accuracy.

The embodiment of the specification provides a detection method, a device and a server for abnormal fund dispatch plan. In the detection method for abnormal fund dispatch plan, after a new fund dispatch plan of a certain user is created, a plurality of accounts of the user can be determined. One or more fund dispatch plans associated with multiple accounts are obtained. At least one fund dispatchdirected graph is generated according to the one or more fund dispatch plans. Among them, the nodes in the directed graph correspond to the accounts, and the directed edges in the directed graph correspond to the relationships between the accounts and the funds. According to a preset graph traversal algorithm, whether a loop exists in the at least one fund scheduling directed graph is detected. Ifa loop exists, the funds transfer frequency of each node in the loop is obtained. According to the frequency of fund transfer, it is determined whether there are any abnormalities in the new fund dispatch plan.

The invention relates to a graph-based regression test case priority ranking method, and belongs to the field of software testing. The method comprises: firstly, collecting test cases and statement execution information during running, and constructing a statement coverage matrix; secondly, calculating the distance between paired case coverage sets through a Jaccard distance function, and constructing a dissimilarity matrix between test cases; then, encoding the test cases, and constructing a graph according to the code coverage information and the test case dissimilarity matrix; and finally,taking the case with the largest statement coverage in the test cases as an initial vertex, traversing the graph through a graph traversal algorithm based on test case search, and generating a test case priority sorting sequence based on the graph. According to the method, the problems of overlong execution time, poor result stability and the like of the existing regression test case priority ranking method are solved, and the test efficiency is improved.

The invention discloses an intelligent second-level power restoration method and system for regulation and control services. The method comprises the steps: 1) realizing power grid fault automatic identification according to power grid steady-state data, monitoring alarm information and protection action information in combination with a Petri network; 2) realizing automatic screening and evaluation of a power restoration path according to the result of power grid fault automatic identification by combining traversal and search of a graph and a load flow calculation technology; 3) graphicallydisplaying a fault influence range and a power supply recovery path; and 4) automatically generating and executing a remote control sequence according to the power restoration path. Power grid fault full-flow automatic and intelligent processing can be realized, more accurate fault judgment, influence range and fault processing information can be provided for regulation and control personnel, andthe power failure time can be remarkably shortened through automatic execution of the power recovery control sequence so that the safe and stable operation level of the power grid is improved.

The invention relates to a method of information extraction for ocean eddy evolution process and belongs to the technical field of ocean information. The method is mainly applied to extract detailed evolution process information of eddies automatically from an eddy tracking data set and, technically, includes: establishing a process-oriented hierarchical data organization model, abstracting and recording the evolution process information of eddies by process, event and scenario objects, tracing the eddy evolution process by depth-first graph traversal algorithm and the like, automatically extracting detailed spatial-temporal information of process, event and process interaction, explicitly expressing and recording the process information, originally implied in the eddy tracking results, into a spatial-temporal database, and providing effective technical and data support for the analysis and discovery of eddy motion feature and law.

The invention relates to a target community recognition method and apparatus, a computer device and a storage medium. The method comprises the steps of obtaining an recognition request; determining acorresponding individual set according to the recognition request to obtain a to-be-identified set; determining individuals with an association relationship in the to-be-identified set to obtain a candidate individual set; traversing the candidate individual set by adopting a graph traversal algorithm to obtain an individual set with a direct or indirect connection relationship in the graph structure network so as to form a target community; and sending the related information of the target community to the terminal, so that the terminal displays the related information of the target community. According to the invention, individuals with an association relationship are determined according to the recognition request; a graph structure network is constructed, a graph traversal algorithm isused to traverse the graph structure network, all associated individuals are found out, a target community is determined, a false insound group can be quickly and accurately found out in a service scene of false insound group recognition, the target community is convenient to identify, the case handling time of inspection case handling personnel is shortened, and the accuracy is high.

The invention discloses a super computer big data processing capability test method based on vertex reordering, and aims to improve the super computer big data processing capability test speed. technical scheme comprises the following steps of generating an adjoining matrix for a map and a construction map, sorting the vertexes in the graph based on the vertex degrees; carrying out BFS search on the graph by utilizing the sequenced vertex set, wherein the characteristic that the probability of edge association of vertexes with high degrees is also high during the same layer of traversal is utilized, transversing and searching child nodes of nodes in the current layer of vertex set, and transversing and checking the vertexes with high degrees preferentially, so that invalid memory access traversal is reduced to the maximum extent. By adopting the method, the hit rate of the edge relation between the nodes can be improved, invalid memory access times are reduced, unnecessary memory access is avoided to the maximum extent, graph traversal is accelerated, and the testing speed of the big data processing capacity of the super computer is increased.

The automatic test case generation method disclosed by the invention is high in test efficiency, strong in adaptability and high in reusability. According to the technical scheme, the method comprises the steps that a state diagram test model is converted into a binary state tree on the basis of a workflow test model, and parameters and judgment conditions in an expression input by a user or an expression defined by the user according to the requirement of the user are extracted in combination with boundary values and equivalence classes on the basis of the binary state tree; a target path existing in the test model is obtained through a graph traversal algorithm, all states are traversed to edges of the states, sequence tables of all nodes in the binary state tree are obtained respectively, values are analyzed through boundary values and equivalence classes to obtain test items, an expression defined by a user is analyzed, judgment conditions of parameters are obtained, a value corresponding to the judgment condition is generated, a test case is automatically generated, and finally reverse verification isperformed according to the generated test case to judge whether the case is correct or not.

The invention discloses a path discovery method and device based on a knowledge graph. The method comprises the steps: constructing the knowledge graph which reflects a character relation network based on the information data of people, events, places, objects and organizations; receiving attribute values of a starting point entity and a target entity and a maximum depth parameter N, finding out the starting point entity in the graph according to the attribute value of the starting point entity, and finding out a relationship network extending outwards to traverse N layers by taking the starting point entity as a starting point based on a knowledge graph of a character relationship network as a sub-graph; traversing in the sub-atlas, searching all paths containing the target entity, and taking the paths as a path set; and drawing a relation graph from the starting point entity to the target entity according to the path set, and displaying the relation graph to a user. According to the method, through graph traversal query, the efficiency is thousands of times or even tens of thousands of times of that of a traditional relational database, and the efficiency is remarkably improved in a path discovery analysis scene.

The invention discloses a complex event scheduling system and method based on a priority-assigned event graph. The complex event scheduling system comprises an atomic event collecting module, a complex event graph model constructing module and a complex event detecting module, wherein the atomic event collecting module is used for collecting an atomic event from a sensing device and storing the atomic event in an atomic event storage; the complex event graph model constructing module is used for structuring a priority-assigned event graph model structure of a complex event according to a semantic rule of the user, and assigning and constructing a scheduling table for the priority of each event to form an event scheduling table to be stored in an event scheduling table set; and the complex event detecting module is used for reading event queues in the atomic event storage and a complex event storage, detecting priority-based graph traversal events according to the complex event graph model and the event scheduling table, and storing the detected complex event to the complex event storage. By using the complex event scheduling system and method based on the priority-assigned event graph, the complex event is scheduled in real time, and the system real-time response capability is improved.

A graph of nodes and edges is received. An identification of a starting node in the graph is received. Traversal walks on the graph from the starting node are automatically performed, wherein performing each of the traversal walks includes traversing to a randomly selected next node until any of one or more stopping criteria is met. One or more processors are used to determine one or more metrics based on the traversal walks. At least a portion of the one or more metrics is used to predict an illicit activity or entity.

The invention relates to a method for artificially identifying financial anti-fraud based on a knowledge graph, which comprises the following steps of: establishing the knowledge graph by utilizing graphical processing of data, assisting finance, integrating data (structured and unstructured) of different sources together, and constructing an anti-fraud rule in a manual checking standard manner. The method has the advantages that data insertion is facilitated, query operation is visual, graph search and graph traversal are convenient, and the speed is high.