192 results about "Match algorithms" patented technology

The invention relates to the technical field of power distribution automation of a power system, and particularly discloses a general operation and maintenance method and system supporting multiple types of power distribution terminals. According to the method, millisecond-level automatic accurate identification of the power distribution terminal is realized through a multi-layer protocol feature fingerprint extraction technology and a weighted dynamic similarity matching algorithm. On the basis of a double-library cooperation mechanism, after equipment identification succeeds, communication parameters, point table mapping relations and operation and maintenance instructions which conform to international / industrial general standards are automatically triggered, loaded and issued. The plug-and-play and zero-configuration deployment of the power distribution terminal is realized, and the point table configuration error rate is reduced to below 1%. And an unidentified equipment self-learning closed-loop process is designed, so that the system can quickly adapt to a power distribution terminal of a new manufacturer, a new number or a new firmware version. And an automatic parameter read-back and verification comparison mechanism after configuration execution forms an operation and maintenance operation closed loop, so that the accuracy of configuration issuing each time and the controllability of the equipment operation state are ensured.

The invention discloses a defect segmentation loss evaluation method based on space consistency optimization, and relates to the field of computer vision and defect segmentation. The problem that an existing loss calculation method is difficult to give consideration to defect space consistency and fair modeling of defects of different scales, and consequently loss evaluation accuracy is poor is solved. According to the method, a prediction result and a defect area in a real label are subjected to instantiation modeling, a prediction and real defect set is constructed, and one-to-one optimal matching of the prediction result and the real defect set is realized through a Hungary matching algorithm; in order to enhance the modeling capability for defects of different sizes, a Wasserstein distance is introduced to measure the spatial similarity of matched defect pairs, defect segmentation loss is calculated for successfully matched defect pairs based on the distance, a fixed loss value is directly applied to predicted defects which are not successfully matched, and the sum of the defect segmentation loss and the fixed loss value is used as a total loss value of a sample. The method is mainly applied to an industrial product surface defect segmentation task.

The embodiment of the invention provides an address data matching method and related equipment, and belongs to the technical field of geographic information services. The method comprises the following steps: constructing an address annotation corpus according to input address information data and a preset address database; the method comprises the following steps: generating a geographic information embedding vector according to a preset geographic information knowledge graph, performing address element analysis in combination with an address annotation corpus to obtain an address element sequence so as to construct a dictionary tree, and performing similarity screening through a spatial hierarchical matching algorithm to obtain a similar address set; generating an address embedding vector matrix through a preset word embedding vector model, and performing feature extraction through a preset semantic feature extraction model to obtain semantic-level similar features; according to input address information data, multi-dimensional character similarity matching is carried out to obtain character-level similar features, then weighted fusion is carried out in combination with semantic-level similar features, and target matching address data is determined according to a weighted fusion result. According to the embodiment of the invention, the address data matching accuracy and efficiency can be improved.

The invention discloses an interactive terminal system for artificial intelligence recruitment, and relates to the technical field of artificial intelligence and human resources. The system comprises an intelligent interaction module, a demand analysis module, a data storage module, an intelligent matching engine, a post issuing module, an intelligent evaluation module, a process management module and a user feedback and model optimization module. The system receives an unstructured recruitment demand through a multi-modal interface, analyzes and generates structured post data based on a natural language processing technology, realizes accurate man-post matching through a multi-dimensional matching algorithm, and automatically completes post release, skill evaluation, video interview analysis and process management. According to the method, the problems of inaccurate demand understanding, low matching efficiency and tedious process in traditional recruitment are effectively solved, intelligent and automatic management of the whole recruitment process is realized, and the recruitment efficiency and talent matching accuracy are remarkably improved.

The invention provides a UI automatic test system and method, computer equipment and a storage medium, a manual case structuring module receives a manual case and converts the manual case into structured data in a preset format; the page element obtaining module calls a browser driving engine according to the target URL to obtain full page elements; the AI element matching module is used for scoring full elements of a page through a pre-matching algorithm based on operation types and interface element descriptions in the structured data, screening out first K elements, and carrying out secondary screening through a large model to determine target elements; and the automatic use case execution module generates an automatic test use case according to the target element and the operation data in the structured data, executes the automatic test use case and verifies whether an execution result conforms to expectation or not to realize UI automatic testing. By the adoption of the scheme, the labor cost of UI testing is reduced, the efficiency and adaptability of UI testing are improved, and then the automation level and the large-scale capacity of UI testing are improved.

The invention provides an intelligent course matching and promotion path recommendation method and device based on post portraits. The method comprises the steps of obtaining internal multi-source heterogeneous data of an enterprise and performing preprocessing to obtain a standardized multi-dimensional feature data set; constructing a dynamic post capability portrait by adopting a semi-definite programming method of an asymmetric random block model; analyzing a causal relationship between capability dimensions by applying a linear time primitive algorithm to obtain an employee capability assessment report; a constraint solution algorithm meeting excessive dissipation is adopted to calculate the capability gap matrix to obtain a personalized capability improvement requirement; screening and combining the learning resources through a weighted matching algorithm to obtain a personalized learning recommendation plan; and adopting a promotion path planning algorithm to construct multi-dimensional promotion possibility evaluation to obtain personalized promotion path suggestions. According to the invention, a closed-loop feedback mechanism of learning, evaluation and promotion is realized, and a systematic solution is provided for enterprise talent development.

The invention discloses a shaping template positioning method based on a BIM matching algorithm and pre-assembly optimization, and the method comprises the steps: reading a corrected BIM and a field point cloud, and constructing a stage freezing boundary; generating candidate hole positions by using three-point reference and one-circle two-slot holes, and completing reachable, interference and processing constraint checking; according to alignment of the drifting main shaft and the normal direction, the long hole is orthogonally provided with a limit and a backstop in the same direction and distributes slippage; an assembly chain for round hole initial insertion, pushing insertion along the main shaft and final hole locking is made; and a template processing drawing, an operation card, positioning parameters and deviation statistics are output, and one-time positioning under the complex working condition is achieved.

The invention relates to the field of automobile financial consumption, and provides a method, a system, equipment and a medium for intelligently and automatically recommending an automobile loan scheme. According to the method, a two-dimensional portrait is constructed by obtaining user data and fund party data, the user data is processed by feature crossing and a portrait algorithm to generate a user portrait, and the fund party data is processed by normalization and feature engineering to generate a fund party portrait. The method is advantaged in that the user and fund party portraits are matched based on the machine learning model, the matching degree is calculated, the personalized car loan scheme is generated according to matching degree sorting, interest rate, deadline and approval requirements are considered, accurate matching of the user and the fund party is realized, and recommendation success rate and user satisfaction are substantially improved. Meanwhile, the method supports dynamic adjustment of the model, a matching algorithm is optimized in real time through performance comparison of the challenger model and the champion model, efficient extension and multi-fund-party access are achieved, and the method has industrial popularization value.

The invention discloses a lossless compression storage method and system based on heterogeneous data in a credential environment, and the method comprises the steps: firstly, extracting a mapping relation between a file type and a structural feature through scanning a target file, carrying out the preliminary lossless compression after classification and grouping, and further employing a sliding window matching and sequence matching algorithm to optimize compressed data for a redundancy mode, and then generating a metadata structure containing a check code and a compression parameter, packaging the metadata structure into a data packet supporting cross-platform transmission, finally transmitting the data packet to a target storage device through a secure write-in protocol, and verifying data consistency and recovering path feasibility by using the check code. According to the method, the data compression efficiency is remarkably improved, the compatibility of cross-platform transmission and the integrity of stored data are ensured, and efficient and reliable technical support is provided for complex electronic file processing.

The invention discloses a reference multi-target tracking method based on language decoupling and fine-grained multi-modal feature alignment, and relates to a computer vision technology. The method comprises the following steps: A, giving a training data set containing a video sequence and language description; b, inputting the video sequence in the step A into a backbone network to extract visual features, and inputting language description into a language model to extract text features; and C, performing multi-modal alignment and fusion through a cross attention mechanism according to the visual features and the language features extracted in the step B. And D, decoupling the language features extracted in the step B into local description and a motion state. And E, inputting the refined features extracted in the step C and the local description extracted in the step D into a static semantic enhancement module to extract target information. And F, associating the current frame target obtained in the step E with the existing trajectory by using a Hungary matching algorithm. G, inputting the matched target features in the step F and the motion state in the step D into a motion perception alignment module to enhance the target recognition capability; the tracking performance of the method is improved.

The invention discloses a multi-level standardized splitting method for detailed addresses, which adopts BERT + CRF technology to accurately split address named entities, utilizes a five-level administrative division address library to check and complement superior and subordinate relationships to ensure the correctness of an address hierarchical structure, and combines a road name library and an address fence technology to improve the accuracy of the address hierarchical structure. According to the method, the geographic range is accurate through regular matching and mixed similarity matching algorithms, the addresses are further verified and complemented by using multi-weight regular segmentation logic and a place name library, and detailed address information which is not limited to house numbers and floor final levels is reserved, so that the complemented address information is ensured to be complete. According to the method, a multi-algorithm fused high-speed matching strategy is combined with an exquisite bidirectional verification and feedback closed-loop mechanism, so that high efficiency is ensured, and meanwhile, the matching accuracy, robustness and logic consistency of address elements (administrative division, roads, interest points and fences) at all levels are remarkably improved.

The invention discloses a satellite spatio-temporal trajectory data association analysis method based on PNTA, and particularly relates to the technical field of data association analysis. Aggregation and crossing behaviors of trajectories in a sliding window are sensitively captured by introducing trajectory crossing and overlapping risk coefficients; secondly, the modal feature degradation perception coefficient dynamically measures the feature degradation degree caused by shielding, interference or sensor failure of the multi-modal information carried by each track from two dimensions of time sequence stability and spectrum distribution, and the feature degradation degree of the multi-modal information carried by each track is evaluated through joint evaluation. The method can actively trigger a confidence suppression and modal enhancement mechanism when a system enters a high-pseudo-consistency mismatch risk area, avoids the damage of false trajectory combination to subsequent clustering and scheduling path deduction, and can maintain the high efficiency and stability of a matching algorithm in a low-risk state. Therefore, fine control over the misjudgment risk in the trajectory recognition link is achieved, and the accuracy, robustness and system response capability of multi-target trajectory association in a complex environment are remarkably improved.

The invention discloses a vehicle target tracking method and system based on an outdoor scene. The method comprises the following steps: matching each to-be-tracked target detection frame and each tracked target tracking frame through a Hungary tracking matching algorithm to obtain a first tracking matching result and target detection frames and target tracking frames which are not successfully matched for the first time; matching the target detection frame and the target tracking frame which are not successfully matched for the first time through a Hungary tracking matching algorithm so as to obtain a second tracking matching result; and generating a tracking queue according to the first tracking matching result and the second tracking matching result. According to the method, the detection frame and the tracking frame which are successfully tracked and matched do not participate in next tracking and matching, and the error matching rate of the second time can be effectively reduced through screening of the first matching; and taking a result after twice matching as a final tracking matching result, and updating the successfully matched detection frame to a corresponding tracking target list.

The invention belongs to the technical field of large language model application, and discloses a large model streaming output-oriented intelligent content analysis and multi-modal rendering method, which comprises a configuration analysis module, a JSON configuration file, a multi-modal rendering module and a multi-modal rendering module, the configuration file defines a unique identifier, a start signal, a paragraph start / end mark, a data stream block type and a front-end rendering strategy of each content type, a content type identification rule base and a start mark-content type rapid lookup table are constructed, and millisecond-level mark matching is realized; the state machine control module is used for processing the streaming data based on an idle-preparation-processing three-stage finite-state machine: waiting for a new content starting signal or mark in an idle state, recording a content type after detecting the new content starting signal or mark, and turning to a preparation state; according to the invention, a cross-transport packet identifier repair technology is adopted; the problem that the identifier is segmented in streaming transmission is solved; through a tentative matching algorithm, the system can automatically detect, cache and recombine segmented content identifiers, and ensure the accuracy of content segmentation.

This invention discloses a substation secondary circuit design method based on equipment functional areas and template matching. The method includes assigning descriptions to secondary equipment based on voltage level or bay; constructing secondary circuit data templates; matching the data template library based on the connection relationships of local equipment functional areas within a bay; extracting functional area connection information through a typical template information flow table; and converting the functional area connection information into actual equipment terminal connections to complete the substation secondary circuit design. This invention mutually converts the functional areas of secondary equipment with the design of secondary circuits, constructing data templates for substation secondary circuits. By improving the VF2 graphical matching algorithm, it achieves digital storage of data templates and rapid matching of complete bay information using local information from data templates. After matching, the functional area connections of the data templates are reverse-converted into equipment terminal connections, realizing automatic secondary circuit design and improving the digital design level of secondary systems.

The invention discloses an intelligent equipment management system integrating charging and data dump, and relates to the technical field of equipment management, and the system comprises an identity authentication module which is used for carrying out the dual authentication of a user identity through a biological feature and an authority token, extracting the three-dimensional feature of a task from a production system, and storing the extracted three-dimensional feature; calculating the adaptation degree of the task and the equipment through a feature matching algorithm to generate a candidate equipment pool; the working condition sensing module is used for generating working condition fingerprints, and comparing historical fingerprints through a dynamic time warping algorithm to judge an abnormal level; the charging transmission and storage module is used for constructing an energy data balance model according to the priority score and the task characteristics, dynamically adjusting the charging power and the dump bandwidth, and scheduling equipment through a predictive migration mechanism; and the resource management module is used for constructing a full file of the equipment, dynamically adjusting an equipment matching rule and a user permission based on an association rule algorithm, and generating a predictive maintenance work order, so that the use safety, the matching accuracy and the maintenance efficiency of the equipment are remarkably improved.

This invention discloses an improved DBSCAN algorithm and a many-to-many matching algorithm for 6G vehicle-to-everything (V2X) networks, belonging to the field of V2X technology. It designs an improved density-based spatial application noise clustering algorithm to cluster more dispersed vehicles, employs a many-to-many matching algorithm to pair vehicles with auxiliary MEC servers, and designs a multi-agent-based task offloading mechanism to reduce latency and improve resource utilization efficiency. The improved DBSCAN algorithm proposed in this invention overcomes the sensitivity of parameter selection, does not require particularly precise parameter settings, performs well on datasets with uneven density, overcomes the influence of vehicle density on clustering results, and the proposed many-to-many matching mechanism improves the utilization of remaining vehicle computing resources and greatly alleviates the pressure on edge servers, while ensuring that high-priority tasks are computed first.

The invention discloses a product recommendation method and device, computer equipment and a storage medium, belongs to the technical field of artificial intelligence, and is applied to health management and health insurance product recommendation scenes. The method comprises the following steps: firstly, carrying out precise modeling on dynamic changes of user physiological indexes by utilizing time sequence analysis, and obtaining an interpretable trend and an abnormal mode, so as to realize early recognition and quantitative evaluation of health risks; secondly, carrying out traceable and explainable multi-modal fusion on an abnormal index quantification result and individualized multi-dimensional information (family medical history, medical behaviors and the like), and outputting a comprehensive health risk score; thirdly, through efficient clustering analysis of historical risk data, a clear group portrait of similar risk groups is formed, and robustness and expandability of grouping intervention and personalized matching are improved; and finally, based on a matching algorithm of similar risk groups and a rich product library, personalized recommendation of health insurance products is realized, and the user satisfaction and the conversion rate are improved.

An apparatus and method for training a machine learning model to generate a match score using unstructured profile data and unstructured reference data. The apparatus includes at least a processor and a memory communicatively connected to the at least a processor. The memory instructs the processor to receive a plurality of unstructured profile data, receive a plurality of unstructured reference data, filter, for each unstructured reference datum of the plurality of unstructured reference data, the plurality of unstructured profile data, generate, for each unstructured profile datum of the filtered plurality of unstructured profile data, a match score by calculating, using a matching algorithm, a degree of similarity between the plurality of metrics and the plurality of criteria, and display, an output based on the generated match score.

System for automating job matching comprises a user interface 102 for uploading job descriptions 114 and resumes 116, and displaying ranked results of candidate suitability of each job requirement. A database directory 104 stores the job descriptions and resumes. A document extension module 106 examines the file extensions of the stored documents to verify a file type or format. A processing module 108 comprises a large language model (LLM) 110 for processing and interpreting resume content through machine-readable instructions (e.g. natural language processing). This may include generating embeddings (by tokenising the documents, semantically analysing them, and assessing a textual context) and retrievers for each document, merging retrievers for respective resume and job description into one document, and creating filters for embedding and compressing the merged document. A master retriever and prompt template may be created. The LLM is coupled to vector database 118 for performing similarity searching with an embedding vector. A scoring module 112 (e.g. having rubric-based scoring templates and matching algorithm) scores candidates based on suitability for each job requirement. The scoring module may include computing scores based on relevant experience (112a, fig.1A), projects (112b, fig.1A), qualifications (112c, fig.1A), and skills (112d, fig.1A) in the candidate resumes.

The invention discloses an intelligent report generation method, system and device based on a multi-agent workflow and a medium, belongs to the technical field of artificial intelligence and natural language processing, and aims to solve the technical problem of how to improve report generation efficiency while ensuring professional report content. According to the technical scheme, the method comprises the steps of intelligent template analysis, wherein a template is analyzed, and paragraph structures and data category demand information are extracted; intelligent data matching: integrating a fuzzy character string matching algorithm and a context semantic understanding technology, and performing intelligent semantic mapping on the CSV structured data and a template placeholder; workflow arrangement: constructing a multi-agent cooperation framework based on LangGraph, adopting a TypeDict type to restrain and manage a workflow state, defining a complete state space including input information, an intermediate processing result and final output, and realizing an end-to-end automatic process of report generation; wherein the workflow arrangement comprises a template analysis node, a data processing node, a paragraph generation node and a report formatting node.

This application proposes a multi-agent boundary game method based on priority path planning, comprising: calculating the relative attack time of the invading agent to the boundary attack point and the relative defense time of the defending agent to the boundary attack point; calculating the priority parameters of each defending agent based on the relative attack and defense times, optimizing the maximum matching algorithm using the priority parameters to determine the defense strategy combination; planning a path for each defending agent according to the defense strategy combination, and optimizing the global path planning cost function using the particle swarm optimization algorithm to generate the globally optimal path for the defending agent; and adjusting the motion state of the defending agent through a linear feedback control algorithm to make it move according to the globally optimal path, thereby achieving path tracking. This application solves the problem of multi-agent defense task allocation and path planning in complex obstacle environments by combining priority path planning and the maximum matching method, effectively improving the defense success rate and adaptability.

The invention discloses a multilevel harmful text classification method, and relates to the technical field of natural language processing and text classification, and the method comprises the steps: firstly collecting harmful and normal texts to construct a positive and negative sample library; establishing a harmful category hierarchical tree; blocking the seed data, generating harmful examples by using a cue word template, and screening the harmful examples; then, a model is built for comparative learning training; finally, dynamic threshold adjustment is performed, a threshold is determined by combining real-time data, and whether the sample belongs to a specific harmful category or not is judged accordingly. Through a dynamic harmful category hierarchical tree and hierarchical activeness matching algorithm, it is ensured that a classification system is comprehensive in timeliness and accurate in positioning; meanwhile, seed text partitioning, sample expansion and the like are utilized to generate high-quality training data and flexibly adjust a classification threshold, so that the accuracy, adaptability and practicability of harmful text detection are remarkably improved.

The invention provides an automatic right and interest conversion and combination method and device based on a smart contract. The method comprises the following steps: receiving a right and interest conversion request initiated by a user through a client; based on the right and interest conversion request, decomposing the right and interest conversion request into a plurality of sub-requests by using a right and interest conversion request decomposition algorithm; carrying out value evaluation on rights and interests involved in any sub-request by utilizing a rights and interests value evaluation algorithm; based on the plurality of sub-requests subjected to value evaluation, searching matching parties in a plurality of markets by using a cross-market right matching algorithm, and generating a matching scheme; and the matching scheme is automatically executed through the intelligent contract, and conversion of rights and interests and on-chain recording are realized. According to the method, the efficiency and flexibility of right circulation are improved.

The application discloses a resource dynamic allocation method, system, device and product in a high concurrency scene, and relates to the field of resource allocation. First, a resource request initiated by a user is received, and it is judged whether the user has resource application qualifications. If the user has resource application qualifications, it is judged whether the resource request meets resource allocation probability conditions based on a dynamic probability interval matching algorithm. If the resource request meets the resource allocation probability conditions, the resource request is intercepted and verified through a five-layer risk control mechanism, otherwise the user is prompted that the resource allocation fails. If the resource request passes the interception and verification of the five-layer risk control mechanism, it is judged whether the current resource remaining quantity meets the resource request based on a three-level inventory architecture of total inventory-time period inventory-Redis real-time inventory. If yes, a double distributed lock is used to control a resource allocation concurrency process, target resources are issued to the user, and the three-level inventory is updated, so that the resource dynamic allocation is completed, and the resource allocation efficiency, stability and security are effectively improved.

The invention provides an electronic purchase intelligent auxiliary bid evaluation method and device, and relates to the technical field of computers. The method comprises the following steps: analyzing the electronic bidding and tendering file, and performing text paragraph segmentation on the analyzed file; dividing the electronic bidding and tendering file into visual image blocks; establishing a spatial mapping relationship between the text paragraphs and the visual blocks; the method comprises the steps of obtaining review index items in a bid invitation file, screening out most relevant text paragraphs from a bidding file through a similarity matching algorithm, performing context enhancement, integrating industry relevant knowledge and bid evaluation rule constraints to construct a structured reasoning chain, generating optimized input information, inputting the optimized input information into a large language model, and outputting review suggestions. The method is used for assisting experts in bid evaluation. According to the method, bid evaluation experts can be assisted to accurately position the positions of the bid evaluation items in the evaluation file, preliminary scoring can be carried out according to a bid evaluation method, suggestions are given, the suggestions are used as references for scoring of the bid evaluation experts, the repeated work of the bid evaluation experts is reduced, and the bid evaluation time is shortened.

The application provides a kind of addressing matrix analysis method for multi-dimensional identification system, by the instruction of terminal input is disassembled into different components, combined into addressing information;Structured data based on multi-dimensional identification attribute, namely addressing matrix, is constructed;Terminal sends addressing request message to access routing device;The addressing request message is forwarded to regional server;K-d tree index is constructed;Using the pre-constructed k-d tree index, the terminal node coordinate C closest to the query point P in Euclidean distance is found;Match appropriate terminal node coordinate;After locking the target terminal through coordinate mapping and similarity matching algorithm, extract the key identifier as the analysis result of query;The method can significantly improve the positioning accuracy and query efficiency of terminal node, can significantly reduce the delay of data transmission, and provide faster response time.

The embodiment of the invention provides a hypergraph-based resource allocation method and system in a hybrid cloud management platform. The method is applied to the technical field of cloud computing, and comprises the following steps: performing virtual extension on a matched individual by adopting an F matching algorithm according to a service request, and performing one-to-one matching conversion; on the basis of the individuals subjected to one-to-one matching conversion, constructing an optimal distribution model by adopting a hypergraph theory; and solving the optimal allocation model by adopting a heuristic algorithm of local search to obtain a resource allocation scheme. In this way, virtual machine resources can be effectively allocated, and the cloud system can achieve the maximum performance.

This invention belongs to the field of intelligent question answering technology, and discloses a vertical domain intelligent question answering method and system based on hierarchical annotation and filtering ranking. The method includes: acquiring and preprocessing knowledge text data in the vertical domain; constructing a vector database and index based on the preprocessed knowledge text data; rewriting the question text data using a large language model; obtaining several candidate texts from the vector database based on the vector matching algorithm after vectorizing the rewritten question text data; evaluating the relevance and performing hierarchical annotation on the candidate texts based on a large language model, and filtering and ranking the candidate texts according to the annotation results to generate highly reliable question answering results. This invention, by combining large model technology, retrieval enhancement generation technology, and intelligent filtering and ranking mechanisms, significantly improves the accuracy of retrieval results, thereby enhancing the reliability and interpretability of question answering results, and effectively improving the user's interactive experience.