24results about How to "Easy to audit" patented technology

The invention discloses an event causal reasoning method for intelligence analysis and electronic equipment. The method comprises the following steps: acquiring a plurality of evidence fragments related to a query text from a plurality of sources, and taking the plurality of evidence fragments as an evidence fragment set; extracting entities from the evidence fragments, and linking the entities into a knowledge graph; event elements are extracted from the evidence fragments, event nodes and causal edges are generated according to the event elements, and one or more candidate causal edges are generated based on the event pairs; retrieving from the evidence fragment set to obtain one or more binding evidences of each candidate causal edge; and calling a large language model for each candidate causal edge to generate a causal assertion, performing consistency verification on the binding evidence of the causal assertion and the same candidate causal edge, combining into a traceable causal evidence chain, and performing event causal reasoning analysis. According to the method, the event-level representation is constructed, and the event-event candidate causal relationship is generated, so that a reasoning object is improved from entity co-occurrence to a reasonable and maintainable event causal structure.

The application provides an intelligent processing method and device for auditing problem qualitative analysis and rectification tracking based on a large model and a medium. It belongs to the technical field of auditing. The system comprises a model construction module: an intelligent auditing knowledge cloud brain is constructed, and knowledge base data is generated; based on the knowledge base data, key auditing element extraction is performed on the auditing business scene, and auditing element extraction result data is generated; auditing problem matching model training is performed according to the auditing element extraction result data, and an auditing problem accurate matching model is constructed; relying on the powerful knowledge base and intelligent matching capability of the large language model, knowledge content closely related to the problem is rapidly and automatically filled in, covering key information such as problem name, manifestation form, problem qualitative analysis, qualitative analysis basis and treatment and punishment basis, which greatly saves the time of auditing personnel for manually searching and sorting data, more ensures the accuracy and integrity of the problem description, and lays a solid foundation for subsequent auditing analysis.

This invention discloses an individual motion behavior analysis system based on dynamic modal configuration and rule-guided fusion, belonging to the fields of intelligent perception, sports science, and human-computer interaction. It includes a data input module, a dynamic modal organization and configuration module, a multimodal feature modeling and fusion module, an indicator-level evaluation and behavior state prediction module, a statistical rule-guided analysis and result correction module, and a structured result generation and output module. These modules work together to form a complete closed loop for motion behavior analysis technology. This invention aims to address the problems in existing motion behavior analysis technologies, such as limited dimensionality of single-modal information, rigid multimodal fusion methods, insufficient stability and interpretability of analysis results, and strong dependence on large-scale manually labeled data.

The invention discloses an unmanned express delivery vehicle control system and method based on operation state perception, and relates to the field of unmanned express delivery vehicle control systems. The method comprises the following steps: acquiring vehicle-mounted data in a driving period through a vehicle-mounted sensor array, performing time alignment and validity marking on the vehicle-mounted data to obtain a data quality label, constructing an evidence item set based on the data quality label, generating a sub-portrait label set according to the evidence item set, and displaying the sub-portrait label set in the driving period. And generating a portrait parameter set based on the sub-portrait label set through a preset priority rule and a persistence judgment strategy. Although the scheme has excellent performance in the aspects of improving system adaptability, safety, dynamic decision-making capability, reviewable performance and resource utilization efficiency, the problems of sensor faults, data loss, real-time data processing bottleneck, label interpretation errors and the like may influence the real-time response and decision-making precision of the system in a complex or extreme environment. Therefore, the overall performance and stability are reduced, especially under the condition of high load or low signal.

The embodiment of the invention discloses a multi-language entry management method, electronic equipment and a storage medium. The method comprises the steps that a structured form file is constructed to serve as a unified data source of multi-language entries, wherein the structured form file comprises a key column used for storing a unique key name and a plurality of language columns used for storing translation texts of all languages; reading the structured table file, constructing a mapping relationship between a key name and a translated text for each language type, and generating a corresponding language entry file; and reading the plurality of language entry files, extracting and combining mapping relationships between key names and translated texts in the files, and generating the structured form file. By means of the mode, centralized management and bidirectional reversible conversion of the multi-language entries can be achieved, development and maintenance cost is reduced, multi-language consistency and quality are improved, and proofreading and auditing of the entries are facilitated.

This application discloses a data-aware intelligent analysis method, system, storage medium, and computer device. The method includes: receiving a target problem and initializing state information, wherein the state information includes key sub-conclusions, a reasoning state logic diagram, and key data descriptions; a language model-based agent iteratively performs the following operations: obtaining input information for the current step; updating the state information of the current step based on the action and environmental feedback information of the previous step, and making decisions based on the state information of the current step to obtain the action and tools to be invoked for the current step; invoking the tools to execute the action of the current step, and obtaining environmental feedback information for the current step after the action is completed; checking whether the target problem has been analyzed; if not, proceeding to the next loop; if completed, generating analysis results. This method can improve the reasoning ability, response consistency, and interpretability of deep reasoning models.

The invention discloses a lightweight virtualization isolation method and system based on a Rust token mechanism, the system comprises a token management module and a trap management module, and the method comprises the following steps: obtaining a resource demand configuration file, casting a corresponding token by using the token management module to obtain a token structural body, and transferring the ownership of the token to a VM object; the Cpu token is consumed, initialization of a CPU context is completed, and a Guest OS starts execution; when the synchronization exception is triggered to be trapped, the trapped management module obtains a corresponding exception token and processes the exception token by matching with a strategy; and when the hyper call HVC is triggered to be trapped and is a management operation, the trapped management module obtains the corresponding HVC token and the management token, verifies the HVC token and the management token, and executes the corresponding management instruction after the verification is passed. Through a token-based resource isolation method, the overhead during operation is reduced while high security is ensured, and the method has good verifiability, portability and expansibility.

The application discloses a kind of based on multi-strategy hybrid mode under insurance product process management method, it is related to computer technology field, the multi-strategy hybrid architecture of event-driven is as core in the present application, through lease main node and majority confirmation guarantee control plane single commit semantic, substantially reduce the risk of out-of-order and repeated execution in concurrent scene;With version unique identifier as anchor point, the environment and product domain information are included in the abstract context, combined with heartbeat reconciliation, standardized serialization and content definition block, the consistency check between channels is upgraded from coarse-grained version number comparison to fine-grained content check, and the missing commit can be replayed in order or the differential synchronization can be performed after deviation is found, less dependent on full coverage, shorten abnormal window;Under the condition of network jitter and node heterogeneity, timeout optimization strategy is based on robust statistical quantile threshold and hierarchical backoff, dynamically matches the delay distribution of each channel, reduces long tail waiting and cascading timeout.

The invention discloses an intelligent report generation system and method based on knowledge graph enhancement, and belongs to the technical field of natural language processing and knowledge graph fusion. In order to solve the problems of insufficient accuracy, low efficiency and opaque reasoning process in the existing structured report generation only depending on large model intrinsic knowledge, the method comprises the following steps of: constructing a knowledge graph containing entities, attributes and relationships, and converting a user natural language problem into a structured query statement by utilizing a first large model; after an entity relationship result is obtained in the knowledge graph, a report with a clear basis is generated by the second large model; and adaptation can also be realized through query languages. By means of structured knowledge and efficient query of the knowledge graph, the accuracy and generation efficiency of the report conclusion are improved, the reasoning process is traceable, meanwhile, the flexibility and universality of the scheme are enhanced, and the method is suitable for structured report generation scenes such as enterprise data analysis and industry research.

This invention proposes an auxiliary review prompting method, device, and medium based on leukocyte population data, relating to the field of clinical laboratory and in vitro diagnostic data processing technology. The method acquires leukocyte population data from an isolated blood sample obtained through a blood analyzer. This data includes at least location parameters reflecting the position of the leukocyte population and dispersion parameters reflecting the dispersion of the leukocyte population. A score is calculated according to a preset scoring rule; the score is compared with a stratification threshold; and laboratory review priority prompts are output based on the comparison results. This invention utilizes leukocyte population data generated synchronously during routine blood tests, eliminating the need for additional blood collection or reagents. It achieves early auxiliary triage of acute lymphoblastic leukemia and infectious mononucleosis in children through interpretable linear scoring, and is suitable for deployment in laboratory information systems or blood analyzer middleware.

The invention discloses an electronic scale control method and device based on image recognition and reverse payment, and a medium, and relates to the technical field of electronic scales. The method comprises the following steps: acquiring an image of an article placed on a scale pan, and obtaining weight data of the article; performing forward reasoning on the article image, extracting image features, and outputting a preliminary identification result of the article and a corresponding confidence coefficient; fusing the preliminary recognition result and the weight data on a feature level to generate an article identifier; in response to a reverse payment instruction initiated by a merchant, a dynamic collection two-dimensional code is rendered and generated on a terminal display screen, and a user is prompted to show a payment code image; a payment code image shown by a user is collected, and the payment code image is analyzed to obtain an encrypted collection account identifier. According to the method, the object type is automatically obtained through image recognition, fusion verification is carried out on the object type and the weighing data, the recognition accuracy and reliability are greatly improved, and a foundation is laid for automatic pricing.

The invention relates to the technical field of computers, and discloses an intelligent behavior data analysis method based on artificial intelligence. Collecting data of at least two data sources, generating an original record identifier for the original behavior record, and associatively storing the original record identifier as an original evidence record set; preprocessing the record set to obtain a behavior event sequence, segmenting the behavior event sequence into behavior segments, and constructing a behavior primitive library; and extracting semantic features and time sequence features from the fragments, determining a candidate primitive set, inputting the time sequence features into a first artificial intelligence model to determine a target primitive, and fitting parameters to obtain a primitive mapping result. And performing consistency verification based on the constraint set, and sequentially replacing the target primitive and adjusting the boundary of the fragment to perform fallback recalculation when the verification is not passed. And generating an evidence tuple for a verification passing result, linking the evidence tuple into an evidence packet, recording a feature snapshot and a version identifier, inputting a primitive time sequence chain into a second artificial intelligence model to output a behavior analysis result, and performing recomputable playback based on the evidence packet to reproduce the analysis result.

The invention discloses an aquatic product processing security management and control system based on blockchain traceability, and relates to the technical field of network security and data traceability, and the system comprises the steps that a multi-source data collection module collects processing whole-link parameters and related information, encrypts and uploads the parameters and related information, and associates an identifier; the data preprocessing and encryption module cleans standardized data, encrypts sensitive information and generates a unique identifier; the block chain traceability core module constructs an alliance chain to realize data chaining and non-tampering traceability; the intelligent safety management and control module evaluates compliance in real time and links equipment to dispose risks; the distributed data storage module adopts a hybrid architecture to guarantee data reliability; and the visual interaction and early warning module presents information and pushes early warning. According to the invention, full-chain data credible traceability is realized based on the block chain, risks are intelligently controlled and accurately identified and disposed, multi-agent collaboration is efficient, and data security is guaranteed; the method adapts to multi-scene requirements, improves processing compliance and emergency response capability, and assists in industrial safety management and control upgrading.

The application relates to an AI generated code judgment method and system based on submitted fact anchoring, which comprises the following steps: collecting AI behavior events generated by an AI coding tool and converting the AI behavior events into uniform event objects as AI behavior event evidence; obtaining difference data of a parent submission corresponding to a current submission of a target code as Git submission fact evidence; performing one-to-one verification and cancellation of the Git submission fact evidence and the AI behavior events; generating an AI code contribution proof object when the verification and cancellation are successful; constructing a three-element evidence chain composed of the AI behavior event evidence, the Git submission fact evidence and the verification and cancellation proof object; and generating an AI code contribution statistical report based on the AI code contribution proof object. The application takes the final submission content as a fact boundary, avoids repeated statistics of the same AI behavior event or the same submission difference line through verification and cancellation, and improves the accuracy of AI generated code source judgment and source-by-source statistics.

The application provides an AI-based station building automatic early warning and control method and gateway. The gateway side performs AI preliminary judgment on station building video stream and station building state data to obtain a candidate event and a preliminary judgment probability; when a verification trigger condition is met, a verification action is adaptively selected from a verification action library and executed to change an observation condition and obtain secondary observation data; wherein the verification action selection adopts an information gain-cost ratio objective function, the expected information gain and cost of each candidate verification action are calculated, and the action with the maximum information gain-cost ratio is selected for execution; and then the gain and consistency evidence of the probabilities before and after the verification in the logarithmic probability space are subjected to secondary judgment, the early warning and linkage control are triggered only when the verification is passed, and an explainable evidence chain is generated and reported. The application has the beneficial effects of significantly reducing false alarms and invalid linkages, and improving the early warning credibility and disposal efficiency in the station building unattended scenario.

The invention discloses reimbursement robot equipment based on AI large model bill discrimination, and belongs to the technical field of financial reimbursement automation equipment. The equipment integrates deep linkage of a printing execution module, a reimbursement robot hardware system and an AI recognition auditing system, and realizes integrated operation of paper bill scanning recognition, electronic bill generation, auditing result associated printing and filing through cooperative control of a mechanical arm and closed-loop connection of a printing process and a reimbursement full link. Closed-loop circulation of bills in the bill delivery bin, the scanning position, the temporary storage bin and the filing bin is achieved through the multi-axis mechanical arm; adopting a ViT + Qwen multi-mode large model to realize end-to-end structured extraction of the bill; and in combination with a background AI auditing engine and a multi-dimensional analysis module, full-chain automation from bill physical storage, intelligent identification and compliance auditing to prediction and archiving is completed.

This invention proposes an access authentication method that binds switch ports to hosts, addressing the challenges of server-side port verification and high management complexity when multiple clients access the network. This method aggregates switch port information (switch identifier + port number) into logical users and binds them to client certificates for unified authentication. Clients can obtain current port information via a link-layer discovery protocol and submit authentication requests in the form of "switch identifier - port number." The authentication server verifies the connection based on stored port information, logical users, and certificate public keys. If the connection port does not match the binding relationship, access is rejected. The authentication request includes the MAC address of the first connected switch to prevent switch spoofing. This technology effectively improves authentication security and flexibility. When network configuration changes, only the binding relationship needs to be updated on the server side, without modifying client configurations, thus significantly reducing the management burden.

This invention discloses a multidimensional attention scheduling and constraint anchoring method based on cognitive rhythms. The method uses deterministic scheduling formulas D(n)=((n-1)mod A)+1 and G(n)=((n-1)mod B)+1 to drive an AI model to switch its attention focus at each step n within an A×B dimensional space consisting of A attention domains and B granularity levels. Each step specifies the currently operable domain and granularity, while the remaining domains are frozen as anchoring constraints. The method achieves ordered traversal of the generation process, deterministic reproducibility, and configurable parameters, simulating the cognitive rhythms in human creation. It is applicable to attention scheduling for various AI generation tasks such as music generation, text generation, and image generation.

The application discloses a kind of big model inference front input data credible verification method and system for power internet of things.The method comprises the following steps: receiving power internet of things multi-source heterogeneous input data;Input data is authenticated, format is parsed and standardized;Based on PKI certificate system and CRL+OCSP mechanism, authenticity is verified;Based on hash chain, local trusted timestamp and periodic blockchain anchoring mechanism, integrity is verified;Extract data features and cross-context features, build multidimensional data graph structure;Based on GraphSAGE, node embedding vector is generated and structural anomaly is identified;Based on Bayesian causal network, posterior probability calculation and abnormal link inference are carried out;Based on random forest fusion score, whether input data is allowed to enter big model inference according to risk level is judged.The application can form input data credible verification closed loop before big model inference, improve the credibility, robustness and explainability of big model inference in power internet of things scene.

The application discloses a plaintext knowledge base and vector database real-time automatic synchronization method and system. The method comprises the following steps: when the knowledge document of the plaintext knowledge management platform changes, triggering a change event; the knowledge synchronization service detects the change event, and acquires the changed knowledge document; the knowledge document is converted into a viking: / / format, and the file name, content and metadata are converted into a format meeting the requirements of a vector memory engine; according to the agent identifier associated with the document, the agent name is routed, and the corresponding tenant key is queried; the converted data is written into the file system of the vector memory engine according to the tenant directory, and automatic vectorization coding is triggered. The application realizes real-time, automatic and multi-tenant isolation synchronization from the plaintext knowledge base to the vector memory engine, solves the problem of knowledge update lag in large model application, and significantly improves the real-time performance and accuracy of knowledge retrieval.

This application provides a target re-identification method, apparatus, and medium, relating to the field of computer technology. The method includes: acquiring first multimodal features of candidate target pairs, the first multimodal features including physical color features, appearance vectorization features, specified structured features, and spatiotemporal context features; the specified structured features including specified structured information and its recognition credibility; and the spatiotemporal context features including the physical velocity of the candidate target pair; inputting the first multimodal features into an improved XGBoost model to determine whether the candidate target pair is the same target, including directly determining whether the candidate target pair is the same target based on the specified structured information whose recognition credibility is greater than a preset threshold and the physical velocity of the candidate target pair satisfying a preset short-circuit rule. This application extracts multimodal features of candidate target pairs in the first stage, uses gradient boosting trees for target re-identification in the second stage, and sets a short-circuit rule mechanism to save computational power and improve system robustness and interpretability.

This invention relates to a method, device, and medium for detecting hardware Trojans in RTL (Real-Time Linked Volume) based on one-dimensional byte streams. It belongs to the field of integrated circuit hardware security detection technology and addresses the problems of linear displacement sensitivity and hyperparameter dependence in RTL hardware Trojan detection. First, multiple RTL files are concatenated into a single file in a deterministic order, and a traceable boundary table is generated accordingly. Then, the RTL to be tested and the Golden reference are read into unsigned 8-bit integer sequences. Global sliding matching is performed on the test sequence over the entire reference sequence using a fixed-length window. The minimum absolute difference (SAD) is used as the similarity metric to find the best correspondence of each window in the reference sequence. Finally, the minimum SAD of all windows is concatenated into a one-dimensional difference curve. The Trojan's disturbance to the structure will appear as a high signal-to-noise ratio spike, fundamentally avoiding the problem of long step jumps that are difficult to locate in two-dimensional methods when encountering linear displacements caused by insertion / deletion.

The application discloses a kind of based on enterprise service whole scene business progress query hub system and method, it is related to enterprise service technical field, solve the scene coverage not comprehensive, data island is serious, query efficiency is low, authority control is not accurate, the degree of visualization is insufficient, lack of intelligent early warning and other technical problems existing in present enterprise business progress query.System includes data acquisition layer, data processing layer, hub core layer, application service layer, authority control layer and storage layer, each layer cooperates and works, realizes the unified acquisition of enterprise service whole scene (including customer management, project management, supply chain management, financial management, human resource management, operation and maintenance management etc.) business data, standardization processing, intelligent analysis and convenient query.Through multi-source data fusion technology, break data island, adopt intelligent search algorithm to improve query efficiency, in combination with fine-grained authority control to guarantee data security, intuitively present business progress by the aid of multidimensional visualization display.

The application discloses a machine learning-based dry-hot valley region soil organic matter and pH spatial distribution prediction method, relates to the technical field of soil spatial information and agricultural resource monitoring, and is based on a pixel and aligns multi-source covariates monthly, constructs machine learning and geostatistics three-line mutual evidence and quantile uncertainty; adopts a bidirectional circulation layer and time sequence self-attention coding, integrates a control layer to multi-head attention, and introduces landform unit, slope direction and altitude band condition coding for fusion and correction; generates a prediction interval with coverage guarantee under the condition of leaving one landform unit and one year, drives directional sampling and suitability confidence linkage; through a closed loop of sampling-retraining-revaluation, converges end error, extrapolation curve and coverage compliance, and outputs a high-precision prediction map, an uncertainty map, a sampling priority and a robust suitability partition.