67 results about "Verification problem" patented technology

The invention discloses an RAG query method, system, medium and product capable of inhibiting a hallucination problem of a generative large model, and relates to the field of RAG query, the method comprises the following steps: obtaining a preliminarily generated text obtained by answering a user initial problem by a preset generative large model; extracting the preliminarily generated text to obtain a key verification problem; performing multiple rounds of questions and answers on the key verification question by using the generative large model to obtain multiple rounds of answers; performing consistency analysis on the multiple rounds of answers to obtain a consistency analysis result; a round of questions and answers with the highest comprehensive score in multiple rounds of answers are directly output at the high grade; when the key verification question is in the middle level, performing multiple rounds of question answering on the key verification question again; when the level is low, RAG retrieval and knowledge fusion are carried out on key verification questions, and questions are asked again; and carrying out illusion identification, correction and text optimization on the answers obtained by asking the questions again to obtain high-quality answers and outputting the high-quality answers. According to the invention, illusion can be accurately positioned, and high-quality answers can be obtained.

The invention discloses a multi-device behavior interoperation verification system and method based on an Internet of Things platform, and relates to the technical field of Internet of Things, and the method comprises the steps: collecting the operation state data and interaction behavior data of each device in real time according to the device interoperation demand of the Internet of Things platform; establishing an equipment behavior prediction model, and constructing a state transition equation of the equipment; according to the equipment operation state data and the interaction behavior data, verifying an interoperation consistency index of the multi-equipment behavior, and calculating an equipment behavior deviation coefficient; when the behavior deviation coefficient exceeds a preset threshold value, generating an equipment interoperation abnormal alarm, and positioning an abnormal interaction node; a three-dimensional dynamic model of the Internet of Things equipment is established through a digital twinning technology, and a multi-equipment behavior interoperation verification result and abnormal early warning information are displayed in real time. The method has the advantages that the problem of state transition consistency verification in multi-device behavior collaboration is mainly solved by constructing a dynamic matching mechanism of the device state transition equation and the operation state feature library.

The invention relates to the technical field of block chain technology and artificial intelligence collaboration, and discloses a verifiable multi-agent collaboration method based on smart contract driving, comprising the following steps: task planning generates a directed acyclic graph and a zero-knowledge proof circuit template, submits the directed acyclic graph and the zero-knowledge proof circuit template to a planning contract, and deploys an exclusive zero-knowledge proof contract; the AI agent collaboratively executes the subtasks, and zero-knowledge proof is generated and submitted through a Veckrii auction tool; zero-knowledge proof contract chain verification is carried out, and after verification is passed, a signature is received and multi-layer gating verification is executed; and the governance contract carries out pledge management and cost distribution on the agents, the providers and the experts. According to the method, zero-knowledge proof and a trusted execution environment are combined, and an under-chain execution behavior is converted into an encryption certificate capable of being verified on a chain, so that the problem that the execution process cannot be verified due to the lack of distributed trusted evidence storage in a traditional centralized framework is effectively solved; and after each subtask is completed, double signature and multi-layer gating verification are forcibly executed.

The invention discloses a large model fine-tuning training system, and belongs to the technical field of education and teaching model training, in the system, a data acquisition subsystem is responsible for collecting a data set containing training and verification question-answer pairs related to the humanity and community field, and the data is from professional textbooks, academic papers and practice cases; the model fine tuning subsystem adopts a small sample learning method to carry out training and fine tuning on the selected general large model, and outputs the fine-tuned model; and the fine-tuned model verification subsystem evaluates the performance of the fine-tuned model, and determines whether the model needs to be further optimized according to an evaluation result so as to ensure the accuracy and practicability of the model. According to the method, the precision and practicability of the model in the field of education and teaching are improved, and the quality and efficiency of the final model are improved by performing screening and difficulty grading on the data, so that the model better adapts to specific education and teaching application scenes and requirements.

The invention discloses a module and a method applied to FPGA (Field Programmable Gate Array) remote upgrade configuration data integrity verification. The module and the method are realized by FPGA logic resources. The configuration data integrity verification module comprises a message filling module, a message expansion function module, a compression iteration module, a constant parameter matrix and a top layer control module. The message filling module is used for receiving original configuration data; the message expansion function module is used for expanding the message blocks output by the message filling module; the compression iteration module is used for hash compression operation; the constant parameter matrix is used for storing constant values required by compression operation; and the top layer control module is used for completing the connection of the sub-modules. Compared with the existing technical scheme, the scheme can realize the integrity verification problem of the configuration data generated by random errors and malicious tampering; by adopting the mode of'pipeline + iteration ', the real-time performance and the low-delay characteristic are ensured, the consumption of the FPGA logic resources by the module is reduced, and the FPGA logic resources can be flexibly deployed on the small-capacity FPGA.

The invention discloses a page table pointer checking method and a system on chip, and relates to the technical field of on-chip storage, and the page table pointer checking method comprises the steps that a first check code is added to a target page table pointer of a target storage fragment, the first check code is used for performing cyclic redundancy check on an address pointer of the target storage fragment, and the first check code is used for performing cyclic redundancy check on the address pointer of the target storage fragment; the first data is obtained based on the first data of the target position in the first verification result obtained through verification. And when a target page table pointer is obtained, performing cyclic redundancy check on an address pointer in the target page table pointer, and obtaining a second check code based on second data of a target position in a second check result obtained through check. Furthermore, whether the target page table pointer is tampered or not is determined based on the matching condition of the first check code and the second check code, so that the check problem of the page table pointer is solved, wrong data address access caused by the fact that the target page table pointer points to wrong address space is avoided, and the security and stability of data access of the system on chip are ensured.

The present application belongs to the technical field of large language models, and relates to a method and system for solving the hallucination problem of large language models, the method comprising: making a basic response to an initial question in a first large language model; judging and verifying the type of the basic response based on prompt word engineering and context learning in a second large language model, generating a verification question and a verification result; based on the verification result, analyzing and correcting the basic response, and generating an accurate result to solve the hallucination problem. The present application connects two large language models and an output model, implements question type judgment and verification based on prompt word engineering and context learning, and analyzes and corrects the basic response, effectively solving the hallucination problem of large language models and improving the accuracy of the results.

The invention belongs to the technical field of aviation simulation, and particularly relates to a simulation test design and evaluation method for opposite task planning verification, which comprises the following steps: establishing an evaluation index system, performing task planning evaluation on a test result according to the evaluation index system to obtain an evaluation result, and then optimizing a test process according to the evaluation result. According to a task scene, a test object participating in simulation is determined, after an initial situation and a test process are determined, test data in a simulation test are evaluated by establishing an evaluation index system, and whether task planning content is effective and feasible is verified through a calculation result value. The advantages and disadvantages of different task planning contents under the same task can be compared through the calculation result; a universal simulation test evaluation method is formed, the correctness and feasibility verification problem of task planning content is solved, and the situation that task planning achieves task requirements can be systematically evaluated.

The invention relates to the technical field of artificial intelligence and network security, in particular to a dynamic verification login method based on a large model, a login service component, equipment and a medium, and the method comprises the steps: receiving a user login request, and extracting user login information; calculating a user risk entropy value based on the extracted user login information; a user risk level is determined according to the user risk entropy value, corresponding problem difficulty configuration is selected, a dynamic verification cue word containing risk level parameters is generated and sent to the large model to generate dynamic verification information, and the verification information comprises a text problem, an image verification problem or a logical reasoning problem; dynamic verification information generated by the large model is rendered and then returned to the login page; receiving user response information, generating a verification prompt word containing a user answer and a standard answer, and sending the verification prompt word to the large model for semantic matching verification; and controlling the user login permission according to a verification result returned by the large model. Compared with a traditional static rule, the risk assessment can identify abnormal login behaviors more accurately.

The invention provides an illusion recognition method and device for big language model data and a storage medium, and belongs to the technical field of computers, and the method comprises the steps: generating a plurality of first verification questions based on segmented texts corresponding to all knowledge themes in a knowledge base of a target big model; generating a plurality of answers for each first verification question; based on the text segment and the plurality of answers, calling a preset contradiction evaluation large model to perform consistency evaluation on each answer of the first verification question to obtain a respective corresponding first evaluation result; based on each first evaluation result, determining a target verification question from the plurality of first verification questions, and generating a second verification question for an answer corresponding to the target verification question; performing contradictory evaluation based on the second verification question and the answer to obtain respective corresponding second evaluation results; and determining the illusion recognition result of the knowledge base based on the first evaluation result or the second evaluation result. The illusion recognition accuracy and recognition efficiency of the large model data are improved.

This application discloses a puzzle generation method, apparatus, device, and readable storage medium. The method obtains a structured specification representation and seed parameters of a seed puzzle; generates a new problem using the seed parameters and the structured specification representation; verifies the accuracy of the structured specification representation using the new problem; if accurate, uses a puzzle generator to load the structured specification representation, reorganize the logical formula, and randomly sample specified data according to defined constraints to obtain a SMT problem; after verifying the SMT problem's logical reliability and solvability using a solver, obtains new configuration parameters based on the SMT problem; fills the new configuration parameters into the structured specification representation and renders it using a natural language template to obtain a target puzzle with a problem instance and a standard answer. This application, through pre-coding, reproducing, and verifying puzzles, can automatically generate accurate and reliable new puzzles in batches.

The invention relates to a question and answer generation model training method and device, computer equipment, a storage medium and a computer program product. The method comprises the steps that problem elements under multiple dimensions in a business system are determined, and the problem elements are obtained by dividing business data in the business system according to the multiple dimensions; generating verification data based on the question elements through a large language model, wherein the verification data comprises a verification question and a verification answer for the verification question; constructing sample data based on the question elements and the verification data, and training a question and answer generation model through the sample data to obtain a trained question and answer generation model; and the trained question and answer generation model is used for outputting target verification data for the construction based on the input target question elements, a target verification question included in the target verification data and a target verification answer for the target verification question, and is used for performing model verification for a question and answer model in the service system. By adopting the method, the accuracy of model verification can be improved.

The invention relates to the technical field of blockchain technology application, and discloses a technology transfer information blockchain management system based on a smart contract. The system comprises a local verification node, a cloud policy engine, a partition permission controller, an encryption transmission channel, a policy divergence analysis module and a hierarchical execution module. The local verification node triggers a preset intelligent contract rule in the fixed verification partition to perform initial transaction compliance verification; if the initial verification fails, the cloud policy engine extracts verification problem features and stores the verification problem features in the temporary cache region, and activates the remote verification partition take-over permission; the partition permission controller locks and fixedly verifies the partition data write-in permission and reads related data; the encryption transmission channel uploads data to the cloud through a security protocol; the strategy divergence analysis module generates a multi-stage verification strategy; and the hierarchical execution module loads a multi-level strategy hierarchical verification risk, and issues a block chain binding permission according to a result. The system enhances the flexibility and security of technology transfer transaction verification, and improves the efficiency.

The invention discloses a credibility evaluation method for artificial intelligence software. The credibility evaluation method comprises the steps of 1, full-life-cycle data acquisition and preprocessing, 2, multi-dimensional collaborative evaluation, 3, scenarized rule adaptation, 4, comprehensive credibility analysis and early warning and 5, visual output of evaluation results. According to the method, traditional dimensions, a large model and multi-mode special evaluation are integrated, the credibility verification problem of an emerging AI technology is solved, and real-time monitoring and collaborative verification of credibility are achieved through full-life-cycle data collection and cross-dimension linkage logic.

The application discloses a kind of mathematical application question solving method and system based on multi-path search, belong to problem solving technical field, the method utilizes mathematical solving model to solve mathematical application question, after determining initial answer;Using initial answer replaces key numerical condition in mathematical application question, obtains verification problem;Using mathematical solving model solves verification problem, determines verification problem answer;The consistency between key numerical condition and verification problem answer is calculated, to determine whether initial answer is accurate;When determining that initial answer is not accurate, according to initial answer to construct negative prompt information;Utilize negative prompt information to revise erroneous reasoning path, until the initial answer that mathematical solving model solves is determined to be accurate;The initial answer that is determined to be accurate is used as the final answer of mathematical application question.Accurate solution to mathematical application question is realized, and the problem that current method is not accurate to solve mathematical application question is solved.

The present application belongs to the field of aviation simulation technology, and particularly relates to a simulation test design and evaluation method for verification of mission planning. An evaluation index system is established, and the mission planning is evaluated based on the test results according to the evaluation index system to obtain the evaluation results, and then the test process is optimized based on the evaluation results. The test objects participating in the simulation are determined according to the mission scenario. After the initial situation and the test process are determined, the test data in the simulation test are evaluated by establishing an evaluation index system. The effectiveness and feasibility of the mission planning content are verified by calculating the result values. The advantages and disadvantages of different mission planning contents under the same mission can also be compared through the calculation results. A universal simulation test evaluation method is formed, which solves the problem of verifying the correctness and feasibility of the mission planning content, and can systematically evaluate the situation in which the mission planning achieves the mission requirements.

The invention relates to a three-dimensional model verification method, and belongs to the technical field of building design, the method is applied to a server, the server cooperatively works with a first web end, a third web end and a BIMBase end, and the method comprises the following steps: obtaining a verification task input by the first web end; determining a to-be-calibrated design model file based on the calibration task, and converting the to-be-calibrated design model file into a lightweight model file; sending the lightweight model file to a third web end for display, and obtaining a verification problem input by the third web end; sending the verification problem to a BIMBase end for display, and obtaining a modified to-be-verified design model file input by the BIMBase end; and rechecking the modified design model file to be rechecked until the rechecking is passed. According to the application, the server cooperates with the first web end, the third web end and the BIMBase end to form whole-process verification problem closed-loop tracking management of the whole verification link.

The invention provides a digital parallel system and method for key algorithm verification, and the method comprises the steps: achieving the construction of an algorithm verification platform through the servitization design thought of algorithm resources, and constructing an algorithm slot and a goods shelf type data system under the constraint of algorithm specifications; an object-oriented, componentization and parameterization thought is adopted, a model system containing different levels of red, blue and white parties is built, task elements are flexibly configured, model granularity is matched according to task requirements, a physical world is mapped, and an algorithm credibility verification environment structure is built; aiming at the evaluation and efficiency verification problem of the algorithm in a changeable scene, completing the establishment of an evaluation system; comprising the steps of multi-means evaluation data acquisition service, construction of a configurable evaluation system, evaluation implementation and algorithm verification, and implementation of an algorithm performance evaluation closed loop. In order to cope with different business scenes, the modularized and parameterized model system can quickly establish corresponding deduction scenes through different combinations and parameter configurations.

An embodiment of the present invention provides an information verification method, device, electronic device and readable medium, the method comprising: a terminal can display a verification interface in response to an information access instruction, the verification interface can include a verification control, a verification channel, a verification question and several answer cards for the verification question that move in real time in the verification channel, etc., the answer card includes at least two answer information, then during the process of the answer card moving in real time in the verification channel, the terminal can respond to the touch operation of dragging the verification control to move in the verification interface, obtain the target answer information pointed to by the pointing identifier of the verification control from the answer card, during the continuous touch operation, if the target answer information is successfully matched with the verification question for at least two consecutive times, then perform an unlocking operation on the verification interface.

The invention discloses a confidence-based verification method for the steering angle safety of an unmanned driving system, and belongs to the field of verification of the safety of the unmanned driving system. The method comprises the following steps: constructing a product network by using a self-combination technology, converting global robustness verification of an unmanned neural network into a standard verification problem, then linearizing an activation function of the neural network, and executing safety verification of an unmanned steering angle by using a neural network safety analysis tool Marbou. According to the technology, super attribute verification of the unmanned driving system can be realized, and the reliability is high; calculation is efficient, feedback can be quickly given out for the situation that the safety attribute is not met, and the method is easy to achieve in an existing unmanned driving system.

The present invention discloses a data management system and method for a supply chain finance platform, and specifically relates to the technical field of supply chain finance services. The present invention copies the historical operation records of the logged-in user as a verification question data set, generates a verification question containing multiple real and false historical operation behaviors when performing identity authentication, and successfully authenticates the identity if all real historical operation behaviors are correctly and completely selected, which can effectively reduce the number of people who are not logged into the platform and effectively improve the security of platform data. It retrieves the supply chain transaction data and financial transaction data of enterprises in the platform, analyzes the supply chain transaction risks and financial transaction risks of different types of enterprises, and then determines whether it is necessary to adjust the enterprise operation authority, which can effectively reduce the enterprise transaction risks and promote the overall improvement of the transaction quality of platform enterprises. It provides a platform enterprise supply chain transaction risk assessment method and a platform enterprise financial transaction risk assessment method, which can improve the accuracy of risk assessment.

This application provides a fixed solution verification method, system, electronic device, and storage medium for PPK (Programmable Point Kinematics) analysis. The method includes: acquiring GNSS observation data from multiple epochs at a user station; performing ambiguity fixing calculations based on the GNSS observation data for each epoch to obtain candidate fixed solutions, where each epoch corresponds one-to-one with a candidate fixed solution; calculating the carrier phase double-difference integer characteristic corresponding to each candidate fixed solution; verifying the correctness of each candidate fixed solution based on the carrier phase double-difference integer characteristic across multiple epochs; and determining the positioning result based on the candidate fixed solution if it is correct. This method fully utilizes the correlation of multi-epoch observation information in PPK post-processing mode, transforming the verification problem of a single-epoch fixed solution into a check of the consistency of correlated information across multiple epochs, significantly improving the accuracy and reliability of fixed solution verification.

The embodiment of the invention provides a method, equipment and device for chip logic verification, a medium and a program product, and relates to the field of chip design tools. The method includes: constructing an NAND graph (and-inverter graph, AIG) that characterizes a sub-logic circuit, the sub-logic circuit including at least a portion of a first circuit and at least a portion of a second circuit; a component graph is determined based on the NAND graph, the component graph including nodes corresponding to circuit function modules in the sub-logic circuits, and the circuit function modules including AND gates and inverters in the NAND graph. The method further includes determining a logical equivalence between the first circuit and the second circuit based on the component map. In this manner, by verifying logical equivalence using the component graph, a logical equivalence verification problem may be solved using circuit configuration information, and a scale of solving a model may be reduced, thereby improving efficiency of verifying logical equivalence.

This invention discloses an automated testing method, system, equipment, storage medium, and program product for microscopic imaging measurements. It addresses the problem of objective verification of measurement results through a log-driven automatic extraction and judgment mechanism; establishes a geometrically dependent linkage verification link to solve the correctness verification problem in dynamic linkage scenarios; uses the start timestamp of the test step as a filtering anchor point for pattern grouping and matching to solve the problem of accurate location of effective logs; and constructs a unified "log parsing—geometric calculation—threshold judgment" process to solve the integrated verification problem of multiple types of measurement scenarios, forming a traceable link of "operation action—log fragment—geometric reference value—judgment result," thus solving the problem of difficulty in tracing back the cause of failure.

The present invention provides an encryption and verification method for DIKWP resources for essential computing and reasoning, comprising the following steps: S1, establishing a DIKWP model of a user, wherein the DIKWP model comprises a data model, an information model, a knowledge model, a wisdom model and an intention model; S2, obtaining a verification request issued by a user, wherein the DIKWP model performs intention recognition on the verification request of the user and generates verification questions of different difficulties; S3, the user inputs an answer to the verification question, and the correctness of the answer is judged by the DIKWP model of the user. If the answer of the user is correct, the verification request of the user is passed, and corresponding operations are performed according to the needs of the user.

The invention discloses an oil field GIS modularization application program interface verification method and system, and the method comprises the steps: firstly, setting a uniform interface specification, guaranteeing that all modules follow the same standard during the interface development, and reducing the verification problem caused by the interface difference; the access point of each interface is provided with interface metadata used for describing the basic information, the parameter type, the value range and the like of the interface. The interface metadata is the basis of interface verification and provides all information required during interface calling. When the interface is called, the transmitted parameters are dynamically verified according to the interface metadata to ensure that the parameters conform to the definition in the interface metadata. And after the parameter verification succeeds, obtaining a corresponding business logic processing object and a data format according to the interface metadata. This ensures that the system can correctly process the interface request, avoiding problems caused by mismatching of data formats or processing logic errors. Therefore, the method provided by the invention can solve the problems in the prior art.

The invention provides a ship power system whole system short circuit protection design and verification method and a test system, and the method comprises the steps: carrying out the modeling of short circuit current related equipment and cables in a real-time digital simulator according to actual equipment parameters, and forming a digital system of a power system according to an actual power system topological structure; the short-circuit current of each node in a short-circuit fault analysis network is simulated in the system, accurate data support is provided for short-circuit current analysis and short-circuit protection schemes of the power system, and the function of improving the accuracy of calculating the short-circuit current of the ship power system is achieved. A short-circuit protection device needing to be verified in the whole system is connected with the electric power digital system through the real-time digital simulator interface module, short-circuit faults are simulated through all nodes in the digital system, the short-circuit protection scheme of the whole system can be comprehensively verified, the engineering development risk is solved, and the engineering development efficiency is improved. The problems of fine design and test verification of short-circuit protection of a large ship power system are solved.

This invention discloses a multi-device behavior interoperability verification system and method based on an Internet of Things (IoT) platform, relating to the field of IoT technology. The system includes: real-time collection of operational status data and interaction behavior data of each device according to the device interoperability requirements of the IoT platform; establishment of a device behavior prediction model and construction of device state transition equations; verification of interoperability consistency indicators of multi-device behavior based on device operational status data and interaction behavior data, and calculation of device behavior deviation coefficients; generation of device interoperability anomaly alarms and location of abnormal interaction nodes when the behavior deviation coefficients exceed a preset threshold; and establishment of a three-dimensional dynamic model of the IoT devices using digital twin technology to display the multi-device behavior interoperability verification results and anomaly warning information in real time. The advantage of this invention is that by constructing a dynamic matching mechanism between device state transition equations and an operational status feature library, it effectively solves the problem of state transition consistency verification in multi-device behavior collaboration.

The invention discloses a trusted computing verification method for a complex equation in cloud computing, which comprises the following steps: a user submits a computing request to a computing service provider through a client, and the computing service provider executes computing operation corresponding to a computing type and returns a computing result; the client side executes multiple rounds of independent modular verification according to the preset safety parameters, if all verification rounds pass, it is considered that the calculation result is credible, and otherwise, the calculation result is refused. According to the method, the verification problem of large integers and complex operation is solved by utilizing different operation types and optimization technologies such as the Fermat's small theorem, the Hough law and the Maclaurin expansion, and the credibility of a calculation result is improved. Besides, the method not only can verify the accuracy of a single calculation result, but also supports the verification of the correctness of the calculation service or the overall logic of the program, has certain program correctness judgment capability, and expands the application scene of the method in trusted calculation and program behavior verification.

The present invention relates to a refined and agile construction method for a multi-aircraft collaborative operation simulation environment, which belongs to the technical field of aircraft semi-physical simulation, and solves the problem that the existing simulation environment construction method cannot realize simulation verification when multiple aircraft are in collaborative operation. The method includes: standardizing and encapsulating the original aircraft model to form a unified standardized aircraft model, and storing it in a model library; formulating application style assumption data according to the collaborative process of the aircraft to be simulated; the assumption data includes: model type, number of model instances, and aircraft collaborative simulation process; according to the application style assumption data, calling the corresponding standardized aircraft model from the model library to generate a model instance and solve it, and the model instances interact with each other in the form of a subject data packet; according to the data solved by the model instance, the operation status of each model instance is displayed in real time. The present application can quickly construct a multi-aircraft collaborative operation simulation environment, and can be widely used in the field of aircraft simulation.