22 results about "Graph operations" patented technology

The present application belongs to the cross field of maritime affairs, ship, shipping, data science, machine learning and artificial intelligence, and particularly relates to a ship lingering AIS trajectory rapid simplification method, which comprises: arranging and processing AIS data of a ship in time sequence to form an AIS trajectory of the ship; cutting the AIS trajectory to obtain a plurality of trajectory segments; performing self-intersection detection on each trajectory segment; constructing a trajectory directed graph for the trajectory segment with intersection points; and using a graph shortest distance solving algorithm to find out a simplified trajectory and output the same. The method provided by the present application can quickly remove and correct trajectory redundancy and disorder caused by lingering, can not change the trajectory form of other normal navigation segments, and can maximize the retention of useful information of the original trajectory; the constructed directed graph is a very sparse graph, and the operation speed is very efficient and fast, without the need of manually fine-tuning parameters, and is suitable for automatically and rapidly processing a large amount of trajectory data.

This application provides a method and apparatus for constructing a permission topology structure based on the lowest privilege level. The method includes: obtaining a non-public interface path template by structured parsing and classifying request data from an identity management platform and performing parameterized substitution; establishing a mapping relationship between the non-public interface path template and a preset identity management platform permission model to obtain a non-public interface parameter structure; merging the non-public interface parameter structure with an existing interface parameter structure to obtain a centralized interface parameter structure; under the lowest privilege role, starting from the initial node of the centralized interface parameter structure according to the task scheduling engine and graph operation model, performing a multi-hop breadth traversal along member relationships, role bindings, and permission operation paths to determine the corresponding permission graph; performing fine-grained dependency extraction on the permission graph; updating the permission graph according to the dependency relationships; and determining the corresponding permission topology structure. This application can improve the security and management efficiency of the identity and access management platform.

The invention provides a territorial space knowledge evolution model construction method, and belongs to the technical field of territorial space knowledge evolution models.The territorial space knowledge evolution model construction method includes the steps that a distributed graph storage framework is constructed, sub-graph partitions are divided through a community detection algorithm, and external event information is collected and converted into a graph operation instruction sequence; positioning a k-hop neighborhood sub-graph by using an incremental sub-graph matching algorithm and executing local updating to generate an intermediate knowledge graph, calculating node state embedding representation of a fusion time coding vector, capturing a time sequence dependence mode through a time perception graph attention aggregation algorithm, executing semantic consistency constraint detection to identify conflicting nodes, and obtaining a time perception graph attention aggregation algorithm; when the conflict degree exceeds a threshold value, the multi-version concurrency control protocol is triggered to roll back to the snapshot version, otherwise, the intermediate knowledge graph is reserved as final output, and the problem that it is difficult to give consideration to updating efficiency and semantic consistency guarantee when the territorial space knowledge graph responds to an external event for evolution updating is solved.

The invention belongs to the technical field of computer aided design, and relates to a cross-modal mapping and equivariant representation fused body-in-white parameterized grid generation system, which comprises an instruction analysis and encapsulation module for generating a graph operation instruction packet containing hash information based on a cross-modal instruction and a reference topological graph; the parameter injection and verification module is used for injecting parameters into the topological graph copy to generate an initial graph after verification is passed; the constraint guidance isovariant generation module is used for injecting an engineering constraint correction signal in the isovariant graph neural network interlayer iteration to generate isovariant representation; the auxiliary data generation module is used for generating a geometric validity mask and a local differential attribute tensor in parallel; and the guiding type deterministic decoding module is used for constructing contour surfaces by using the masks and guiding grid division and mapping according to the tensor. According to the method, the problems of long design iteration period and data incoherence caused by decoupling of design intention and geometric generation and lagging of engineering constraint verification in the prior art are solved.

Systems and methods for generating item recommendations from unstructured data using machine learning models are disclosed. One embodiment includes obtaining a dataset comprising a plurality of items, wherein each item includes at least one field containing unstructured data, preprocessing the dataset by performing filtering and text cleanup on the unstructured data, performing a coarse relatedness analysis by executing lookups on items in the dataset to identify potentially similar items and create links between items that are potentially interchangeable, performing coarse clustering by utilizing the links to organize related items into clusters using graph operations, performing fine clustering by constructing prompts for a large language model for each cluster to recluster items into subclusters and generate labels for canonical items and generating a list of interchangeable item recommendations based on the canonical items and their associated metadata.

A table data question and answer method based on a data fact knowledge graph acquires original table data to be analyzed, automatically extracts data facts, and constructs a knowledge graph containing data fact nodes and semantic relationships thereof; a large language model is used to perform semantic analysis on a natural language query of a user, identify a query intent type, and extract corresponding structured intent elements; the natural language query is mapped to an executable graph operation sequence for the knowledge graph according to the query intent, operations are performed on the graph to retrieve or dynamically generate target data facts and semantic edges thereof; and a data fact obtained is input into a generative large language model as an enhanced context to generate an answer containing a visual chart and a text description. A table data question and answer system based on a data fact knowledge graph is also provided. The present application effectively suppresses hallucinations of a large model by using structured knowledge constraints, and improves the accuracy of question and answer.

This invention proposes an autonomous control method for agricultural machinery based on graph reinforcement learning, comprising: real-time acquisition of multi-source sensing information of the farmland environment and abstraction into a graph structure, wherein the nodes of the graph structure include agricultural machinery nodes, obstacle nodes, and target operation path point nodes, and the edges include Euclidean distance between nodes and collision threat level; inputting the state of the graph structure into a graph neural network, extracting the features of each entity through graph operations and performing aggregation operations to generate a state feature vector; inputting the state feature vector into a reinforcement learning policy network based on proximal policy optimization, and outputting end-to-end control commands for the steering angle and linear velocity of the agricultural machinery at the current moment to drive the agricultural machinery to complete autonomous farmland operations. This method achieves a highly real-time, highly robust, and highly stable autonomous control method for agricultural machinery.

Advanced persistent threats (APTs) have caused significant financial losses for enterprises, making the development of effective detection systems a critical priority. While existing provenance graph-based APT defenses demonstrate high accuracy, the high complexity and cost of graph operations (e.g., construction, iteration) require extensive processing time and computational resources, making them impractical for real-time detection. To address this challenge, we introduce Madeline, a graph-free, lightweight APT detection system that leverages historical system statistics. This graph-free statistical approach enables lightweight coarse-grained detection with minimal computational resources. Through a multi-step state score calculation for a set of behavioral attributes, Madeline meticulously captures subtle, gradual system changes indicative of stealthy APT activities. Using an unsupervised LSTM autoencoder, Madeline performs anomaly detection effectively without the need for prior knowledge or manual labeling of attacks. Additionally, Madeline supports continuous monitoring, enhancing the assessment of ongoing risks.

In response to a user invoking a graph operation on a graph in an in-memory graph algorithm (IMGA) runtime in a relational database management system (RDBMS), the RDBMS generates a set of one or more graph indexes in memory. The graph is represented as one or more vertex tables and one or more edge tables. The RDBMS generates the set of one or more graph indexes by generating a mapping of each database table vertex identifier to a corresponding internal identifier. For each edge table of the one or more edge tables, the RDBMS generates a graph index data structure representing edges of the edge table. The graph index data structure represents each edge in the edge table as a source internal identifier and a destination internal identifier. The RDBMS can then execute the graph operation in the IMGA runtime using the set of one or more graph indexes.

The invention relates to the technical field of water conservancy knowledge graph updating, and discloses a water conservancy knowledge graph dynamic real-time updating system and method based on physical-information deep coupling and semantic field intensity driving, and the system comprises a physical sensing layer which is used for carrying out the real-time collection, transmission and preprocessing of multi-source water conservancy physical data; the coupling mapping layer is used for mapping the multi-source water conservancy physical data into semantic state variables available for a knowledge graph; the field intensity calculation layer is used for calculating semantic field intensity based on semantic state variable changes and determining nodes and relations, needing to be updated, of the knowledge graph; the graph operation layer is used for executing knowledge graph updating decision making and writing operation according to a semantic field intensity triggering threshold value; and the knowledge service layer is used for providing dynamic knowledge service for an external intelligent water conservancy system. According to the invention, multi-source water conservancy data can be effectively integrated, comprehensive acquisition and preprocessing of real-time data are realized through the physical sensing layer, and the accuracy and integrity of the data are ensured.

Methods, systems, and computer storage media for providing a multi-tenant system that executes graph language requests using graph operations of a graph language. A graph language request—that configures tenant data for tenants in a multi-tenant system—is executed using a graph operations engine. The graph operations engine receives and parses a graph language request that includes a list of tenants and a definition of data operations. The set of data operations of the definition are executed on a tree of data operation nodes comprising a plurality of leaf nodes and a root node. Executing the data operations is based on graph language actions (e.g., composition, transformation, and aggregation) that support asynchronously returning results data associated with configuring the Tenant data. Executing the data operations of the definition causes generation of results data (e.g., root node results or leaf node results) configuration of the tenant data in the multi-tenant system.

The invention relates to a dynamic graph hardware accelerator based on search space perception and a method thereof. The hardware accelerator comprises a control module, a loading module and a cache module. The control module analyzes the received graph query task to determine a task type, determines the range of vertexes and edges needing to be accessed by the graph query task by querying graph structure metadata, and generates a task descriptor; sending the task descriptor to a loading module; the loading module performs analysis and address mapping on the received task descriptor, and judges an execution path based on the cache state of the cache module; when it is detected that the cache is not hit, a data access mechanism for external storage is automatically triggered, and batch loading and cache updating of graph data are completed; sending the processed task data and the distribution logic to a target processing unit in an idle state in a computing unit cluster; the target processing unit receives the task data and the distribution logic and executes graph operation; the caching module is used for caching, address mapping and distribution of target data.

This invention discloses a simulation verification method and system for power communication network maintenance based on digital twins, relating to the field of intelligent operation and maintenance technology for power and communication networks. It solves the problems of existing technologies where maintenance approval relies on manual experience, lacks dynamic simulation methods, and cannot assess the risk of superimposed faults. The key technical points are: converting the pending power communication maintenance application form into a graph operation command; inputting the graph operation command into a pre-constructed digital twin model for topology simulation, outputting the maintenance-state network topology; wherein the digital twin model is constructed from the full topology graph of the power communication network; under the maintenance-state network topology, performing a full-network static connectivity analysis, traversing and simulating the network survivability after a single fault occurs on the remaining critical links, and calculating the maintenance risk index by combining the voltage level and type weight of the carried services; comparing the maintenance risk index with a preset safety threshold, and outputting the simulation verification result.

The invention discloses an inductive subgraph retrieval completion method and system for entity update knowledge graph, and belongs to the technical field of information, and the method comprises the following steps: constructing a graph operation function, initializing subgraph parameters, obtaining relation representation through attention aggregation, and calculating entity representation by combining message passing and main neighbor aggregation, and updating the representation by using gated residual connection, expanding the sub-graph and performing scoring to complement the triad. Through local sub-graph retrieval and inductive reasoning, the problem of sparsity during dynamic updating of the knowledge graph is solved, complementation efficiency and accuracy are improved, and the method is suitable for an information retrieval and recommendation system.

This application relates to the fields of software engineering and system architecture design, and provides a method, system, and computer-readable storage medium for boundary context identification based on business semantic analysis. The method includes: performing semantic layer reasoning on an acquired business requirement document to obtain business semantic elements; mapping the business semantic elements of the business requirement document to nodes and connecting edges according to preset graph construction rules to obtain a business semantic coupling graph; identifying high-cohesion node sets in the business semantic coupling graph to form communities using graph algorithms, obtaining the initial boundary context corresponding to each community; applying heuristic rules to perform dynamic graph operations on the initial boundary context to obtain the final boundary context; and generating a visual interface to display the final boundary context and the set of interaction relationships between different final boundary contexts. This method automatically discovers boundary contexts in the early design stage based on business requirement documents, without relying on existing code and text similarity, ensuring the accuracy of automatic context identification.

The invention discloses an asymmetric topological key manufacturing method based on an odd graceful topological coding graph. An asymmetric topological key is composed of two essentially different mathematical branches, namely a topological structure and a mathematical code. Wherein the theory of the topological structure is based on a graph theory and topology; the theory of mathematical coding is based on combinatorial mathematics, linear algebra, number theory, set theory, algebra, computer science and other subjects. Giving a coding mirror image from the odd topology to the odd topology, and a mirror image or inverse substitution operation from one odd topology (private key) to another odd topology (public key); establishing a odd extension graph group and a topological graph group signature thereof; a set ordered odd topology graph W-operation graph lattice is constructed by using topological coding point superposition operation and edge addition operation, and a topological graph lattice signature and infinite'one-time pad 'keys are obtained; a generic topological coding matrix and a Fibonacci key sequence are defined so as to be suitable for wider information security application scenarios.

The application discloses a multi-field knowledge fusion method based on semantic trees, and mainly relates to the technical field of knowledge graph in multi-source and multi-field knowledge cross fusion. The application is based on the establishment of semantic trees of multiple fields of knowledge, proposes a corresponding association rule mining method, performs association analysis between the semantic trees, finds the association rules of different fields, and performs mining and calculation of associated elements based on graph operation theory, so that organic fusion of multi-field knowledge can be realized. The application can solve the problems of not easy alignment, association and merging of semantic trees of different fields due to heterogeneity, and can fuse multi-source and multi-field knowledge into an organic whole, so that more comprehensive, unified and cooperative knowledge sharing can be provided, and the application is especially suitable for theoretical and systematic design of a unified technical architecture in a 'knowledge-intensive' field.

The invention relates to a small molecule collision induced dissociation mass spectrum prediction method based on end-to-end graph reinforcement learning, and the method comprises the following steps: step 1, molecule representation and graph operation, step 2, Markov decision process definition, step 3, hydrogen migration module realization, step 4, strategy network architecture and value network architecture construction, and step 5, prediction of a small molecule collision induced dissociation mass spectrum based on end-to-end graph reinforcement learning. And 5, reinforcing a learning algorithm and a training process. According to the method, the whole prediction process is subjected to end-to-end optimization under a unified graph reinforcement learning framework, the model interpretability is enhanced, chemical understanding is promoted, manual rule limitation is overcome, and the training efficiency and the model robustness are improved.

The application discloses a kind of non-imaging semiconductor overlay error measurement device and method, including building bright field single-pixel imaging light path;Based on single-pixel imaging light path under the premise of non-imaging Fourier frequency domain acquisition;Two-dimensional cross-correlation coefficient distribution is obtained by the calculation of Fourier frequency domain;Profile center subpixel positioning is carried out using two-dimensional cross-correlation coefficient distribution;Overlay error is calculated by the difference calculation of the subpixel positioning of two profiles.The application realizes the premise without imaging, only through light intensity measurement direct positioning detection target subpixel profile center and calculating overlay error, greatly reduce the data amount generated by graph operation, improve measurement efficiency, expand the application of single-pixel imaging technology in the field of semiconductor overlay error measurement.

The invention discloses a writing assistance method and device, electronic equipment and a storage medium, and the method comprises the steps: outputting a current guide problem related to a to-be-written requirement under a current interaction round with a target object; receiving the current response content of the target object to the current guide problem; analyzing based on the current response content and historical response content under a historical interaction round before the current interaction round to obtain a to-be-executed graph operation after the current interaction round; on the basis of the atlas operation, updating a target atlas related to the to-be-written requirement in an interaction interface with the target object, and returning to execute the step of outputting the current guide problem related to the to-be-written requirement under the current interaction round with the target object; and outputting a writing template for guiding the target object based on the target map. According to the scheme, the guiding continuity and effectiveness in the writing assisting process can be improved, and generation of the personalized writing template for the target object is facilitated.

An approach and corresponding computer system adapted for combining relational and non-relational data storage and indexing techniques is proposed. The approach utilizes graph data structures that are specifically configured for supporting performance improvements in relational querying while maintaining the advantages from a non-relational database. The graph data structures can be used for graph traversals and graph operations, among others, which can improve an overall computational complexity level associated with various operations at the cost of storage.