34 results about "Business domain" patented technology

The application discloses a tax intelligent operation and maintenance management system based on a digital and intelligent base, and particularly relates to the technical field of tax business operation and maintenance; the system is constructed by collecting tax operation and maintenance related data of a sharing platform, a tax and bank channel and internal multi-business domains, constructing a cross-domain operation and maintenance object correlation graph, and identifying the relationship change of version solidification and adjacent version difference recognition caliber coding parameter data based on the graph; the system compares and matches to identify the drift candidate set of the cross-domain operation and maintenance object correlation edge, performs the presequence constraint and postsequence landing point check along the caliber coding parameter mapping transmission path to form an influence fragment set, and extracts a semantic abnormal fragment set through semantic consistency review of running representation data; finally, the relationship drift source object is determined according to the co-occurrence convergence characteristics of the relationship change item, so that the accurate identification and rapid positioning of the tax business semantic level anomaly are realized.

The application relates to the technical field of blockchains, and discloses a mobile product whole-life-cycle tracing method based on a blockchain, which comprises the following steps: a relay chain extracts compliance judgment indexes in administrative rule update data, generates compliance check salt values in combination with time stamps, and distributes the compliance check salt values to parallel chains of each business domain; the parallel chains of the business domain map material tracing data and the compliance check salt values, generate business state vectors containing local details and embedding time-effect features of the compliance judgment indexes; and the relay chain implements logical check on the collected business state vectors by using an orthogonal state mask, anchors to a global tracing ledger after verifying a compliance time sequence, and establishes a strong time sequence constraint benchmark of management logic in a heterogeneous chain by the compliance check salt values, eliminates compliance risks caused by rule replacement lag of local nodes, and ensures response efficiency of cross-chain tracing under a mass transaction condition by using a bit operation mechanism.

The present application relates to the technical field of artificial intelligence large model, specifically to a report generation agent autonomous construction method based on knowledge enhanced large model, which comprises: knowledge extraction on business domain structured documents and unstructured texts and construction of domain knowledge graph, adoption of an improved progressive knowledge injection algorithm to integrate knowledge into general large language model parameters to form a knowledge enhanced base model. Extraction of task planning sequences and operation trajectories generated by historical reports to construct a demonstration sample library, and fine tuning of a primary report generation agent through a behavior cloning strategy. A reinforcement learning training field containing multiple rounds of complex tasks and diversified environmental disturbances is built, interactive data is collected to form a behavior experience pool, decision parameters are iteratively updated based on proximal policy optimization, and autonomous construction of the agent is completed. This method can improve the effect of domain knowledge fusion and enhance the task planning and autonomous decision-making ability of the agent in complex scenarios.

The application belongs to the technical field of sensitive data management and control, and particularly relates to a business sensitive data management and control system based on a block chain, wherein the system comprises an architecture definition module, a ledger segmentation design module, a sub-ledger consensus construction module, a cross-business domain data interaction module, a sensitive data node allocation module and a ledger fault tolerance module. The scheme realizes local encrypted storage of sensitive data original text and operation logs of the sub-ledger by constructing a double-layer segmented architecture of the main ledger and the sub-ledger, combining cloud network fusion technology, completing sensitivity evaluation through feature selection and a machine learning model, preferentially allocating high-frequency data to near-main-ledger nodes according to data heat, preferentially allocating new data to low-load nodes according to node load, obtaining an optimal node load allocation scheme through a genetic algorithm when there is no low-load node, realizing dynamic expansion and merging of the sub-ledger, reducing data migration cost during business domain expansion, and guaranteeing uninterrupted sensitive data access services.

This application relates to the field of security auditing technology, and provides a method and system for cross-domain business behavior auditing based on clock synchronization. The method includes: performing clock calibration and dynamic compensation on slave nodes of each business domain, assigning a unique time sequence identifier to each slave node of the business domain to bind domain attributes and node identity information; encrypting and collecting operation logs and / or data flow records of each business domain to obtain core time sequence features; constructing a benchmark fitting based on the core time sequence features and calculating residuals, quantifying the residuals to obtain time sequence deviation quantization values, and performing cross-domain business abnormal behavior detection based on the time sequence deviation quantization values ​​and multi-level abnormal thresholds; when abnormal behavior is detected, performing time-series full-link tracing based on the calibrated timestamp and unique time sequence identifier, and reliably storing the abnormal behavior and corresponding tracing link data, thereby achieving accurate link reconstruction of abnormal behavior and enhancing the integrity of the evidence chain, and improving the accuracy and reliability of cross-domain business compliance auditing.

This invention relates to a method, system, and medium for intelligent identification of data meta-objects based on metadata. The method includes: acquiring raw metadata from multiple heterogeneous data sources and mapping it to a unified Meta-Entity structure; generating technical tags and business semantic tags for the attributes of each Meta-Entity structure using a multimodal semantic annotation model; fusing the business semantic tags with an enterprise domain knowledge graph using a graph attention network to obtain statistical feature vectors, structural feature vectors, and semantic features; fusing the statistical feature vectors and structural feature vectors using an MLP-GNN Hybrid model to obtain a unified feature vector; dynamically adjusting the weight coefficients of the statistical features, structural features, and semantic features based on reinforcement learning, so that the weighted identification score maximizes the comprehensive reward function; and writing the technical tags, business semantic tags, and weighting strategy into a data directory. This invention can complete model transfer in new business domains with a small number of labeled samples, achieving high identification accuracy.

This specification provides a risk detection method, apparatus, system, and device. The method includes: acquiring structured configuration parameters of a target resource operation project to be detected, and target business knowledge of the business domain to which the target resource operation project belongs; using a preset large language model, performing semantic analysis processing on the structured configuration parameters based on the target business knowledge to obtain project configuration information corresponding to the target resource operation project; acquiring multi-dimensional configuration risk detection rules corresponding to the target resource operation project; using the preset large language model, performing multi-dimensional configuration risk detection processing on the target resource operation project based on the project configuration information and the multi-dimensional configuration risk detection rules; and determining whether there is a risk in executing the target resource operation project based on the multi-dimensional configuration risk detection processing results.

The application provides a method and device for identifying interface release dependency, a storage medium and a computer device. The application determines the incremental calling interfaces of the to-be-tested business domain code comprehensively and accurately through the combination of static code scanning and interface calling monitoring, so that the interface release dependency of the to-be-tested business domain code can be accurately identified according to the deployment of the incremental calling interfaces in the production environment. Moreover, the application can identify the parameter adjustment of the internal interfaces and the calling of the internal interfaces in the to-be-tested business domain code to consider the dependency of the external business domain code on the to-be-tested business domain code. It can be seen that the application can fully consider the bidirectional dependency relationship between the to-be-tested business domain code and the external business domain code, accurately and comprehensively identify the interface dependency of the to-be-tested business domain code in the release process. Moreover, the application can greatly improve the release dependency identification efficiency through automatic identification, and balance the high efficiency and high identification accuracy.

This application discloses an intent recognition method and electronic device. In response to a natural language query input by a user, the method identifies the target intent and target object contained in the natural language query. Based on prior knowledge of the attributes of the business domain category to which the target object belongs, the method extracts candidate attributes of the target object from the natural language query. Based on the natural language query and the attribute prior knowledge, the method performs semantic reasoning on the candidate attributes to generate corresponding attribute constraints. The attribute constraints include the target attributes corresponding to the candidate attributes under the business domain category, as well as the constraint logic operators on the target attributes. Based on the target intent, target object, and attribute constraints, the method generates a structured intent representation corresponding to the natural language query.

The embodiment of the invention provides a knowledge segmentation retrieval method and device, equipment, a medium and a product, and relates to the field of data processing.The method comprises the steps that a knowledge base is divided into a tree structure according to hierarchies, semantic routing is completed in cooperation with structured slot position information extracted from user query, and the semantic routing efficiency is improved. According to the method, accurate reduction of the retrieval range from the full knowledge base to the files in the target node is realized, traversal retrieval of the content of the full base is not needed any more, the calculation complexity of retrieval is reduced, meanwhile, the retrieval process can accurately match the actual business intention of user query by relying on constraints such as the business field and the classification level of slot position information, and the user experience is improved. According to the method, cross-domain irrelevant retrieval and mistaken recall conditions are reduced, and then mixed retrieval and reordering are only carried out for candidate files of which the range is reduced, so that efficiency loss caused by massive data calculation is reduced, the problem of precision reduction caused by semantic fuzzy matching in full-amount retrieval is avoided, and the retrieval efficiency is improved. The problems of precision degradation and insufficient efficiency caused by data scale increase are solved.

This invention relates to the field of software development technology, specifically providing a method and system for code generation and isolated execution based on a multi-level model. The method includes: acquiring business domain data containing business objects, field attributes, object relationships, and semantic metadata, and constructing a business ontology model; generating code based on the model, dividing the code into a model layer, a read-only logic layer, and a data operation layer, and configuring differentiated code regeneration rules; embedding the business domain data into the model layer code in a structured manner, enabling direct runtime access; performing layer validity verification at runtime through an import interception mechanism, prohibiting the import of read-only logic layer code into data operation layer code; repeating the aforementioned operations when the model changes, and ensuring idempotency of the code generation process according to the differentiated rules. This invention can standardize the code layering architecture, achieve mandatory runtime call isolation, balance model synchronization and custom logic retention, and adapt to automated development and deployment processes.

The application discloses a data processing method and device, equipment, storage medium and program product, and relates to the computer technical field. The method comprises the following steps: acquiring a domain model constructed based on a business requirement, the domain model being used for graphically representing entities and attributes in a business domain to which the business requirement belongs, the domain model comprising M entities, and each entity comprising at least one attribute; determining a first mapping relationship and a second mapping relationship according to the domain model, the first mapping relationship being used for mapping each entity to a table corresponding to the entity, and the second mapping relationship being used for mapping the attribute to a field corresponding to the attribute, the field corresponding to the attribute being a field in the table corresponding to the entity in which the attribute is located; and storing, in a form of a table, business data in a business domain into a data warehouse based on the first mapping relationship and the second mapping relationship, the business data being data related to at least one entity in the M entities. The method improves the data processing efficiency.

This invention relates to the field of intelligent task processing technology, and more particularly to a method and system for autonomous multi-agent task processing in enterprises based on a large model. The method receives business instructions from an enterprise management system, generates a structured task tree with business domain labels, hierarchical structure, and dependencies using a natural language target parsing algorithm, maps sub-task nodes to candidate enterprise agents to obtain a first allocation result, and performs a second dynamic reallocation based on the load difference between the task load index and a preset load threshold to obtain a second allocation result. Each candidate enterprise agent asynchronously executes sub-task nodes according to dependency edges, achieving state persistence across steps, agents, and days. An exception self-healing algorithm automatically handles execution anomalies, and a compliance audit algorithm provides full-process traceability and access control. This invention realizes natural language-driven, load-balanced autonomous execution and full-process compliance auditing for complex enterprise business processes.

The present application provides an Internet security self-monitoring disposal system and method, comprising a hit event detection step: collecting system behavior data, calculating real-time anomaly score through a trained anomaly detection model, and judging whether to trigger an attack event based on a dynamic threshold; a threat intelligence sharing step: feature extraction and desensitization processing are performed on the attack event, a feature abstract is generated, and cross-business domain sharing is realized through a block chain layered architecture; a disaster recovery disposal step: a disaster recovery priority is calculated based on the threat parameters of the attack event, and when the disaster recovery priority exceeds a preset threshold, the corresponding disaster recovery protection action is triggered; a dynamic response step: a layered reinforcement learning model is constructed, a disposal strategy is generated in combination with security targets, business constraints and compliance rules, and the strategy network is optimized through a reward function, and the present application solves the technical problems of traditional systems relying on fixed rules, low intelligence sharing efficiency, passive lag of disaster recovery and rigid compliance strategy through a whole-process closed-loop design.

The application relates to the technical field of machine learning modeling, and provides a satisfaction prediction model construction method and device, electronic equipment and a storage medium, the method comprising the following steps: classifying business domain index data according to data characteristics to obtain original numerical characteristic data and original discrete characteristic data; screening the original numerical characteristic data through a pre-screening unary regression model group to obtain numerical characteristic modeling index data; screening the original discrete characteristic data through a pre-screening unary classification model group to obtain discrete characteristic modeling index data; and constructing a satisfaction prediction model according to a model construction algorithm in combination with the numerical characteristic modeling index data and the discrete characteristic modeling index data. The satisfaction prediction model construction method provided in the application embodiment accurately selects characteristic information through a pre-screening model group, reduces the interference of redundant characteristic information, constructs a satisfaction prediction model with high prediction accuracy, and accurately predicts the satisfaction of customers.

The present application relates to the technical field of artificial intelligence large model, specifically to a report generation agent autonomous construction method based on knowledge enhanced large model, which comprises: knowledge extraction on business domain structured documents and unstructured texts and construction of domain knowledge graph, adoption of an improved progressive knowledge injection algorithm to integrate knowledge into general large language model parameters to form a knowledge enhanced base model. Extraction of task planning sequences and operation trajectories generated by historical reports to construct a demonstration sample library, and fine tuning of a primary report generation agent through a behavior cloning strategy. A reinforcement learning training field containing multiple rounds of complex tasks and diversified environmental disturbances is built, interactive data is collected to form a behavior experience pool, decision parameters are iteratively updated based on proximal policy optimization, and autonomous construction of the agent is completed. This method can improve the effect of domain knowledge fusion and enhance the task planning and autonomous decision-making ability of the agent in complex scenarios.

This invention discloses a Text2SQL generation method based on multi-source knowledge fusion and adaptive progressive processing. The method includes constructing an enhanced schema semantic library, a dynamic experience vector knowledge base, and a business domain knowledge graph to obtain a multi-source knowledge repository; constructing an adaptive progressive decision-making strategy based on the multi-source knowledge repository to obtain query processing paths from natural language query text; generating SQL candidate query statements by invoking a prompt word assembler and fusing it with a preset LLM large language model based on the query processing paths; performing SQL verification on the SQL candidate query statements; if the SQL verification passes, the final Text2SQL is used as a successful interaction case to dynamically update the dynamic experience vector knowledge base, and the confidence attribute is adjusted based on the experience value to update the business knowledge graph; if the SQL verification fails, an error message is sent and the natural language query text is re-entered. This invention solves the problem that existing methods cannot effectively achieve both accuracy and efficiency in obtaining SQL query statements.

This application provides a customer service routing method, apparatus, device, storage medium, and program product, which can be applied to the field of artificial intelligence technology. The method includes: responding to a user initiating a dialogue, routing the conversation to a first artificial intelligence model, whereby the first artificial intelligence model performs cross-business domain semantic understanding on the user's query and generates a response; determining the business domain based on the interaction content between the first artificial intelligence model and the user, and routing the conversation to a second artificial intelligence model matching the business domain, whereby the second artificial intelligence model continues to interact with the user based on the existing interaction context; performing multi-dimensional semantic analysis on the interaction content between the user and the second artificial intelligence model, evaluating the interaction quality level from different dimensions, and obtaining an interaction quality evaluation result; if the interaction quality evaluation result does not meet a preset interaction quality level, routing the conversation to a human customer service representative, whereby the human customer service representative continues to interact with the user based on the existing interaction context.

This application relates to a method for generating a data indicator system based on a large model. The method includes constructing large model prompts for the content of the current business system; using the large model to obtain a list of business domains within the current business system; constructing corresponding large model prompts for each business domain in the business domain list; and using the large model to obtain data indicator-related information for each business domain; importing the data indicator-related information into a data indicator platform to complete the construction of the data indicator system; and managing the data on the data indicator platform. This method can quickly generate a data indicator system using large model knowledge, which business experts can then modify. On the one hand, it changes the traditional reliance on business experts, significantly reducing workload. On the other hand, this method also generates SQL statements on how to extract data indicators from the database, greatly reducing the workload of data engineers in developing data indicators.

This invention relates to the field of artificial intelligence and can be applied to business system platforms in finance and healthcare. It discloses a method, apparatus, device, and medium for architecture reconstruction based on business domain isolation, including: acquiring full-volume job data and service association information of an initial shared cluster; grouping data by business domain to obtain job service mapping lists for multiple business domains; calculating the dedicated cluster requirements for each business domain based on the job service mapping lists and pre-acquired baseline carrying capacity thresholds, and deploying target dedicated clusters for each business domain; and migrating full-volume job data from the initial shared cluster to the corresponding target dedicated clusters in batches according to their respective business domains, resulting in a cluster architecture isolated by business domain. By splitting and reconstructing the shared cluster by business domain, the jobs of teams in different business domains are isolated through dedicated clusters to avoid mutual interference, ensuring the stability and availability of the job scheduling platform.

The application relates to the computer technical field, in particular to a micro-service domain algebra data type construction method based on functional programming, which comprises the following steps: obtaining a variant list of a target business domain; identifying the properties of each variant in the variant list, and determining the type of each variant according to the properties of each variant; combining all the variants in the variant list based on the type of each variant to obtain an algebra data type tree, compiling the algebra data type tree, verifying the compiled algebra data type tree, and performing business processing through the verified algebra data type tree. Therefore, the problems of scattered domain object definition, high coupling degree, lack of clear boundaries and complex and lengthy verification logic in the micro-service are solved, the readability and maintainability of the code are improved, and the overall reliability and development efficiency of the micro-service system are enhanced.

This application discloses a method, apparatus, device, and medium for generating GEO optimization strategies based on simulated web-based interactive behavior, relating to the field of model optimization technology. The method includes: constructing an evaluation prompt word matrix based on the business domain corresponding to the user terminal; inputting the question statements in the evaluation prompt word matrix into a preset web-based terminal constructed based on GEO to simulate user inquiry behavior on the preset web-based terminal and obtain corresponding text response information; using a text processing model to perform structured parsing of the text response information to extract target structured text; the target identifier name includes the target brand identifier and competitor identifiers; determining the evaluation dimensions for GEO, and evaluating the target structured text based on the evaluation dimensions to obtain the evaluation results corresponding to each evaluation dimension, so as to generate a list of optimization strategies corresponding to GEO based on the evaluation results. In this way, realistic GEO performance evaluation results can be obtained and more targeted optimization strategies can be generated.

This invention relates to the field of multidimensional data processing technology. It discloses a method and system for constructing a multidimensional data model for thermal power equipment. The method includes: collecting basic and indicator data of the thermal power equipment; using machine learning algorithms to identify data relationships between different business domains and constructing a multi-level data abstract model; defining the hierarchical structure and specific attributes in the data model, as well as the relationships between each layer; and dynamically updating the data relationships to adapt to changes in the data source, forming a data model that supports the entire lifecycle management of the equipment. This method combines advanced machine learning algorithms to identify data relationships between different business domains to support comprehensive equipment management.

This invention belongs to the interdisciplinary fields of enterprise digital supervision, graph data processing, and intelligent early warning, and relates to a penetrating real-time early warning method and system based on event projection influence domain pruning. The method includes: extracting standardized event objects; projecting the events onto a cross-business domain knowledge graph to filter for effective projection nodes; constructing an initial influence domain subgraph in the cross-business domain knowledge graph; pruning the initial influence domain subgraph according to the comprehensive influence scores of nodes and edges to obtain a target influence domain subgraph; selecting a corresponding early warning trigger mode based on the target influence domain subgraph and generating an early warning task; executing the early warning task on the target influence domain subgraph and outputting the early warning result. The system includes an event projection module, an influence domain construction module, an influence domain pruning module, a trigger control module, and a risk detection module. This invention can reduce the participation of irrelevant nodes in the calculation, reduce the resource consumption caused by full graph scanning, and improve the accuracy, real-time performance, and resource utilization efficiency of early warnings.

The application discloses a logic defect intelligent detection method and system based on business semantic enhancement, which comprises the following steps: constructing a historical business defect knowledge base by using multi-dimensional information of historical defects; extracting business domain, language, file path and business rules from the code to be detected; generating a code semantic vector according to the code to be detected; performing multi-modal representation learning on the code to be detected, using a multi-modal fusion mechanism to fuse the embedded representations of abstract syntax tree, control flow graph, program dependency graph and business semantic graph, and generating a code graph structure; adopting a hierarchical retrieval strategy to filter similar historical defects from the historical business defect knowledge base; inputting the code to be detected and the similar historical defects into an adaptive Agent reasoning engine, performing static and dynamic analysis on the code to be detected through a natural language instruction driven tool, and comprehensively generating a defect detection result; and verifying the result according to the confidence output by the adaptive Agent reasoning engine and generating a defect report.

The invention discloses a natural gas well engineering data management method and system and a medium. Relates to the technical field of data management. The natural gas well project is divided into multiple levels of service domains according to the overall service process of the natural gas well project, and the service objects of the service domains / sub-service domains and the data entities associated with the service objects are identified according to the service domains, so that the standardization and consistency of service domain definition and organization are ensured, the fuzziness and repeatability of service domain definition are avoided, and the user experience is improved. The definition of the data directory is improved, so that a user can quickly find related data resources; the data table is automatically associated with the business object, so that the inaccuracy in a manual annotation process is reduced; the efficiency and accuracy of data mapping are improved, and the business object modeling and data table mapping process is more objective and accurate.

This application proposes a method for synthesizing abnormal fact case data for the coal industry, relating to the field of data processing technology in smart mines. The method includes: acquiring a dictionary set for the coal industry; determining a first corpus set for the coal industry based on the operational constraint files within the dictionary set; classifying the first corpus set to obtain a second corpus set for different business domains within the coal industry; extracting key points from each corpus in the second corpus set for each business domain to obtain a key point set for the corpus; and simulating abnormal fact cases based on the second corpus set and the key point set of the corpus set to obtain a dataset of abnormal fact cases for the business domain. This application synthesizes simulated abnormal fact cases for the coal industry based on standardized item corpora from operational constraints, thereby providing data support for the training of large-scale models and improving the training effect of related large-scale models in the coal industry.

The application discloses an information service management method and system for cross-business domains of an e-commerce platform, and relates to the field of e-commerce. The method comprises the following steps: collecting multiple business domains of the e-commerce platform, performing full-business-process analysis to generate multiple intra-domain business operation sets; analyzing the consistency requirement strength of operation types, performing synchronous and asynchronous division to construct synchronous and asynchronous operation sets; building a twin resource pool, receiving multiple service requests initiated by multiple users, performing cross-business-domain resource pool calling simulation, judging whether to perform asynchronous blocking calling, and if yes, responding to the asynchronous arrangement with the synchronous and asynchronous operation sets. The application solves the technical problem that the existing cross-business-domain information service management cannot guarantee the strong consistency of core business data while taking into account the high concurrency performance and low response delay of the system, and achieves the technical effect that the cross-domain calling is adaptively balanced between consistency guarantee and performance optimization through intelligent identification and dynamic arrangement.

The application provides a method and device for identifying a calling chain, a storage medium and a computer device. According to class structure information of each code class in a to-be-identified business domain, a candidate class is preliminarily screened out. The candidate class refers to a subclass of an input / output serialization class, and the code class includes a code class with a public access modifier and a candidate code method with a preset parameter method field in a method name. For the candidate class screened out preliminarily, the application can determine whether the candidate class includes a candidate code method for implementing field reference in combination with annotation information of a class definition field. By skillfully using the inheritance relationship of the code class, the method characteristics of the code method and the class definition characteristics of the input / output parameter, the application can automatically identify the method related to the field reference and the related field name, and has the advantages of high efficiency and accuracy. By monitoring the calling conditions of the methods, the calling chain identification result of the field reference can be accurately recorded.

The application discloses a message content extraction method and device, computer equipment and a storage medium, and relates to the technical field of big data. First, the message to be audited is converted into a keyword vector through keyword extraction and vector conversion. Then, a pre-trained domain classification model is used to classify the keyword vector by domain, and the business domain to which the message belongs is determined. According to the domain classification result, a matching target business domain knowledge graph is selected, and a vector representation thereof is generated. Next, in combination with the vector representation, a preset long short-term memory network model is used to perform semantic coding on the message to generate a semantic representation vector. Finally, based on the semantic representation vector, a conditional random field model is used to accurately identify key entities, attributes and relationships in the message. The application also relates to the technical field of blockchains, and the message to be audited is stored on a blockchain node. The application improves the accuracy and efficiency of message content extraction, and can automatically identify key entities, attributes and relationships.