62results about How to "Efficient retrieval" patented technology

The invention relates to the technical field of image processing, and discloses an industrial article appearance detection method and system based on machine vision. The method comprises the steps that a workpiece image is collected, and a standardized image is obtained through Gaussian filtering, histogram equalization and Otsu algorithm preprocessing; feature vectors are extracted through edge detection, Hough transform and a gray-level co-occurrence matrix; training a defect identification model by adopting a CNN of an attention mechanism; inputting an image for real-time detection; fusing the detection data and the environmental parameters by using a D-S theory; and establishing a database based on a distributed architecture to generate a quality report. According to the method, the detection accuracy is improved through multi-source information fusion, and efficient management and analysis of mass detection data are realized by adopting a distributed architecture.

The invention belongs to the field of database enhancement, and particularly relates to an automatic RAG knowledge base construction method and system.The method comprises the steps that firstly, an RAG knowledge base is constructed, a user can customize an icon and a name of the RAG knowledge base, meanwhile, a data embedding model and a parallel retrieval mode function module are built in, and a text recall test tool is configured; the data embedding model can convert unstructured data into vectors and store the vectors in the classification vector database; a to-be-retrieved document is uploaded to a knowledge base to be preprocessed, a document text demand space is obtained, an optimal effective recall information fragment set is obtained from parallel retrieval modes of vector retrieval, keyword retrieval, mixed retrieval and the like of the knowledge base by utilizing a segmented TopK recall strategy and a text recall test tool, and the optimal effective recall information fragment set is stored in a database; and finally, obtaining an effective question and answer result according to the document text demand space and the optimal effective recall information fragment set, thereby improving the construction efficiency of the knowledge base and the question and answer accuracy.

The invention provides a customer service interaction method, an interaction device, equipment, a storage medium and a program product, and relates to the technical field of computers. The customer service interaction method based on the large language model comprises the following steps: performing intention scene recognition on a received user question; identifying that the user question belongs to a specified intention scene, and determining a corresponding domain knowledge base matched with the user question; performing enhanced retrieval on the knowledge base in the corresponding field based on the user question to obtain a business knowledge fragment for enhancing the user question; and inputting the business knowledge fragments into a large language model, so that the large language model performs filtering type question reply processing based on the business knowledge fragments, and outputs reply information of the user question based on a processing result. According to the technical scheme, the real intention of the user in continuous interaction can be deeply analyzed, the content which the user wants to consult can be more accurately grasped, and the consulting requirement of the user is better met.

The invention discloses a data management system based on a monitoring video stream, belongs to the field of data analysis, and aims to solve the problems of camera connection acquisition, image quality optimization, pedestrian related information analysis and data storage retrieval and backup in video monitoring. The system comprises a data acquisition module, a target detection module and a data storage module. The data acquisition module is used for identifying and connecting a video monitoring camera, performing video stream data acquisition, adjusting an acquisition frame rate by estimating a data size, and evaluating and optimizing image quality to obtain a detection image data set; the target detection module is used for carrying out feature extraction, classification and marking on pedestrians in the detection image data set, recording pedestrian trajectory information, carrying out real-time tracking and updating, identifying a behavior mode according to the trajectory information, dividing a gathering area and obtaining the pedestrian flow; and the data storage module is used for classifying and labeling the data, constructing an index structure to guarantee efficient retrieval, dividing data grades according to multi-dimensional features, and dynamically adjusting a backup time interval.

According to the construction scheme multi-modal retrieval enhancement generation method and system based on the large model, intelligent scheme writing support is provided for constructors through a natural language processing technology by means of mass building industry data and knowledge, and the construction efficiency is improved. Comprising but not limited to the steps of assisting constructors in compiling a construction scheme, providing real-time answers to problems possibly encountered in the construction process, providing optimization suggestions according to project characteristics and the like; through the application of the large model, the working efficiency of constructors can be greatly improved, the waste of human resources is reduced, and the accuracy and feasibility of a construction scheme are ensured. Besides, the model can be continuously learned and updated, and the ability of assisting in compiling question answering is continuously optimized along with the accumulation of project experience, so that the model is more widely applied to each link in the building industry in the future.

The invention discloses a large-model-based campus work order transfer method, device and equipment and a medium, relates to the field of work order processing, and is applied to a preset campus work order transfer system. Performing knowledge graph construction by using the obtained campus work order knowledge graph template, the historical campus work order information and the campus department responsibility information to obtain a target campus work order knowledge graph; performing vector conversion and historical campus work order retrieval on the to-be-processed campus work order through a preset deep learning large model and a vector database to obtain a retrieval result corresponding to the to-be-processed campus work order; and performing information analysis and department responsibility matching on the to-be-processed campus work order based on a preset deep learning large model, the retrieval result and the target campus work order knowledge graph, and completing work order transfer operation corresponding to the to-be-processed campus work order according to an obtained matching result. The work order processing quality is effectively improved, and the manual intervention cost is reduced.

The embodiment of the invention provides a power business document processing method and device, equipment, a computer storage medium and a computer program product. The method comprises the following steps: acquiring an unstructured document corresponding to an electric power business hall service system; identifying basic content blocks contained in the unstructured document based on a preset document structure rule; based on the complexity of each basic content block, determining a target server to which each basic content block is allocated from a server cluster; extracting key service information in the corresponding basic content block through the target server; and fusing the key business information into a power business knowledge graph corresponding to the power business hall service system to obtain a power business document database. According to the method, the structured processing efficiency of the power business document and the management efficiency of the power business information in the power business hall are improved.

The invention discloses an intelligent question and answer method and system based on a mineral knowledge base, relates to the technical field of artificial intelligence and data mining, and is used for solving the technical problems that an existing mineral knowledge base is poor in reasoning ability and low in mineral knowledge acquisition efficiency and accuracy. The intelligent question and answer method based on the mineral knowledge base comprises the steps of obtaining mineral knowledge; distinguishing mineral names, mineral attribute names and mineral attribute values in the mineral knowledge; establishing a mineral knowledge base data table structure and a mineral knowledge graph triple structure; the mineral knowledge base data table structure and the mineral knowledge graph triple structure are stored in a knowledge database; building a webpage interface, wherein the webpage interface is provided with a search engine and an intelligent question and answer mechanism relying on a large model; the intelligent question-answering mechanism realizes intelligent question-answering of the mineral knowledge base based on the knowledge database. The invention is used for providing the intelligent question answering method and system based on the mineral knowledge base.

The invention belongs to the technical field of Lustre file management, and particularly relates to a method and system for improving Lustre small file management and performance, and the system comprises a nested file system, a three-level index module, a cache index module and a self-adaptive cache module; the nested file system realizes management and storage of small files; the three-level index module comprises a small file global metadata index, a container file internal index and a cache index and is used for optimizing retrieval and storage management of small files; the cache index module is used for managing small files accessed at high frequency and reducing repeated I / O requests; and according to the self-adaptive caching module, the client dynamically adjusts a caching strategy of a small file according to the file access frequency and mode, so that the reading speed of a high-frequency access file is improved, and network I / O (Input / Output) operation is reduced. The performance bottleneck of a Lustre file system during small file processing is solved through a nested file system, and the access burden of a metadata server and an object storage server is reduced through a three-level index architecture.

The invention relates to the technical field of environmental monitoring, and discloses an exclusive encrypted file format of a universal water-gas-sound-residue pollutant data acquisition instrument. File structures including a file header setting section, a data storage format definition section, a self-checking section and an encryption algorithm section are sequentially stored in a file to form an exclusive encryption file format of the universal water, gas, sound and residue pollutant data acquisition instrument, various encryption and checking measures are adopted, and the safety and integrity of data in the transmission and storage process are ensured. The method has the advantages that data are stored in the database and prevented from being tampered and stolen, data storage efficiency and retrieval speed are increased due to reasonable data storage formats and index designs, data management and analysis are facilitated, file damage or data tampering can be timely discovered through CRC32 check and data integrity check codes, data reliability is guaranteed, and data storage efficiency and retrieval speed are increased. And on the basis of an XML structured storage mode and an open file format design, subsequent addition of new pollutant parameters or function extension is facilitated, and good expandability is achieved.

The invention relates to the technical field of data processing, in particular to an order data processing method and system based on a block chain. The method comprises the following steps: acquiring a multi-source order data set; performing transaction address analysis on the multi-source order data set to generate order transaction address data; order processing server construction is carried out on the multi-source order data through the order transaction address data, and order processing distribution server data is generated; order processing access frequency analysis is carried out on the multi-source order data set through order processing distribution server data, and order processing access frequency data is generated; and performing order screening on the multi-source order data set based on the order processing access frequency data to generate high-frequency order access data and low-frequency order access data. According to the invention, through optimization of multi-source data integration, order access frequency analysis, transaction synchronization and channel optimization, the reliability and efficiency of order transaction are improved.

The invention relates to the technical field of computers, and discloses a data persistence method and device, equipment, a medium and a program product, and the method comprises the steps: defining model information corresponding to a single business demand for each business demand; determining a model calibrator corresponding to the single service demand according to the model information corresponding to the single service demand; storing the model verifier and the model information corresponding to each business demand in a target database and a cache according to a corresponding relationship; obtaining a target model calibrator from the cache according to the to-be-persisted data; verifying the to-be-persisted data through the target model verifier; and when the verification is passed, storing the to-be-persisted data in the target database according to the target model information in the metadata warehouse. According to the method, the corresponding model information is defined for each business requirement, different business scenes can be flexibly adapted, the model verifier and the model information are stored in the cache, and the data retrieval and verification speed and the data persistence efficiency can be improved.

The invention discloses a video content retrieval method and device based on television system voice interaction and a terminal, and relates to the technical field of video processing, and the method comprises the steps: when a video is played for the first time, extracting a picture frame from the video at a preset frequency, converting the picture frame into a multi-dimensional image feature vector and a corresponding video timestamp, and carrying out hierarchical storage in a database, constructing vectorized data containing visual semantic information; obtaining a voice retrieval instruction, performing intention recognition and semantic understanding, extracting a detection keyword, and generating a multi-dimensional retrieval feature vector; calculating a matching degree between the multi-dimensional retrieval feature vector and a multi-dimensional image feature vector of a video picture frame stored in a database, and screening out picture frames of which the similarity is higher than a preset similarity threshold to form a retrieval candidate matching set; and determining matched picture playing. The video content retrieval method is efficient, accurate and high in interactivity, and retrieval experience and operation efficiency of the user in the video watching process are remarkably improved.

The invention relates to a class C liquid material storage system which comprises an automatic packaging line area, the automatic packaging line area comprises an automatic packaging line system and a warehousing mechanism, and the warehousing mechanism comprises a storage tank supply station, a canning station, a labeling station, a cover sealing station and a tray assembling station which are sequentially connected in the upstream and downstream direction; the automatic packaging line system is in signal connection with the labeling station; the automatic warehouse area comprises a warehouse management system, an automatic warehouse and a pallet supply station, the warehouse management system is in signal connection with the automatic packaging line system and the automatic warehouse, the warehouse management system is used for sending the pallet requirements of the corresponding number to the automatic warehouse, and the automatic warehouse is connected with the pallet supply station; the automatic bin is used for sending pallets to the pallet supply station; and the pallet supply station is connected with the tray assembling station. The automatic packaging line area and the automatic warehouse area can be connected in series, and efficient goods taking can be achieved by means of the automatic packaging line system and the warehouse management system.

The embodiment of the invention relates to the technical field of image retrieval, and provides a secure outsourcing image retrieval method and device, electronic equipment and a medium, and the method comprises the following steps: obtaining an image retrieval request sent by a query user, including retrieving an encrypted image; in response to the image retrieval request, performing ciphertext feature extraction on the retrieved encrypted image to obtain a first orthogonal moment feature vector, and calculating a first ciphertext hash value corresponding to the first orthogonal moment feature vector; determining a target hash bucket in a hash bucket set based on the first ciphertext hash value; screening a preset number of target encrypted images with the minimum distance from the first orthogonal moment feature vector in the target hash bucket; and converting the encryption form of the target encrypted image, and returning the converted target encrypted image to the query user. Therefore, efficient retrieval of the secure outsourcing image is realized, and the security of the scheme is improved.

The application provides an information retrieval prompting method and device, electronic equipment and a storage medium, the method comprising: preprocessing device retrieval information input by a user to determine a retrieval keyword; sequentially matching the retrieval keyword with a prompt word library, a synonym library and a pinyin word library to obtain multiple prompt information corresponding to the device retrieval information; the prompt word library is constructed based on attribute information of various devices, the synonym library is constructed based on synonymous words corresponding to the attribute information of various devices, and the pinyin word library is determined based on the prompt word library; and the multiple prompt information is sent to a front end for display. The application can provide accurate retrieval prompt information for a user, help the user quickly and effectively retrieve attribute information of an Internet of Things device of interest, improve the multi-dimensional prompting and accurate retrieval capability of the Internet of Things device, and provide a good user experience.

The application discloses a method for extracting keywords information of bidding and opening bidding based on regular expression and scrapy, comprising the following steps: S1, determining the analysis field of the bidding and opening bidding project; S2, collecting and sorting the keyword list of the field, wherein the keyword list is created as an ordered keyword list or an unordered keyword list according to the field requirement; and S3, selecting the subsequent processing method according to the specific field condition. By combining the regular expression and the scrapy, the application can extract multiple keywords information of the bidding and opening bidding project, so as to avoid the problem of missing search. The application respectively extracts the keywords information of the bidding, the winning bidding and the opening bidding project, can carry out targeted query extraction according to the different keyword extraction modes of the opening bidding project, realizes the quick extraction of important information in the text, removes the useless data, solves the problems of information redundancy, reading difficulty and the like.

The invention discloses a method and system for efficiently storing and retrieving cross-language text data and a storage medium, and relates to the technical field of data management, and the method comprises the following steps: dividing different sentences according to the structures of the sentences, and constructing a structure tree; classifying different part-of-speech characters according to a grammatical structure to which the characters belong, and forming a grammatical linear linked list; storing the same-part-of-speech characters in a same-dimensional linear linked list, and forming a multi-dimensional part-of-speech linear linked list with the part-of-speech category as the dimension by performing part-of-speech analogy in sequence; the method comprises the following steps: establishing an N-dimensional linear linked list by taking Chinese as a retrieval reference, and putting different language representations of the same character into a same-dimensional language linear linked list; and based on the established grammar linear linked list, the part-of-speech linear linked list and the language linear linked list, obtaining the position of the to-be-retrieved character and outputting the position information to finish the retrieval operation. According to the method, the effects of efficient storage and retrieval can be realized, and the problems of large text data volume, large occupied storage space and relatively low expandability and improvement caused by emotional intention change of an expressor are solved.

The application relates to the technical field of artificial intelligence, and discloses a large model expert library construction based on a knowledge graph, which realizes real-time acquisition of multi-source data, marks an update point through a hash fingerprint algorithm and semantic similarity calculation, and stores the update point in a message queue in combination with metadata determination; a low-dimensional semantic vector is generated by extracting features and constructing a fusion network, and cross-modal mapping is established; an attribute graph is built with the semantic vector as a node, and the attribute graph contains basic facts, reasoning rules and time sequence edges; a user request is parsed as a graph traversal task, and a result is returned through counterfactual reasoning; sensitive data and relationships are encrypted and indexed; source field graph features are migrated to optimize target field graph, and the construction of the expert library is accelerated. The application realizes efficient data processing, semantic correlation analysis, safe query and rapid construction of the expert library, and improves data utilization and knowledge reasoning efficiency.

The application discloses an enterprise collaborative office method and system based on intelligent AI, and the method comprises the following steps: collecting and preprocessing the data of each end of the enterprise collaborative office; constructing an AI collaborative engine containing multiple AI models, and training and optimizing each model by using the preprocessed data; the AI collaborative engine analyzes the semantics and intention of the input information of each end, automatically triggers the corresponding process and allocates tasks, adjusts the allocation strategy according to the execution situation; automatically identifying the relevant departments and personnel of cooperation, establishing a temporary collaborative team and generating a cooperation report; performing hierarchical permission management on the data of each end, monitoring abnormal behavior and generating a security audit log; combining the task execution and process operation data, continuously iterating and optimizing each model of the AI collaborative engine. The enterprise collaborative office method and system based on intelligent AI solve the problems of existing collaborative office data islands and insufficient intelligence, improve the collaborative efficiency and management level, and adapt to the needs of enterprise digital transformation.

The present invention discloses a video question-answering method, system and medium based on multimodal information fusion, which belongs to the field of computer vision technology and is used to solve the technical problems that the video question-answering technology relies on single modal information, cannot make full use of the multimodal information contained in the video, and the efficiency and accuracy of video retrieval are insufficient. It includes: extracting basic information of video files to obtain video summaries and text subtitles; segmenting text subtitle sentences to obtain subtitle segments, and obtaining corresponding video segments and audio segments; clustering each audio segment to obtain several voiceprint type clusters; obtaining the video lens set and subtitle set corresponding to each voiceprint type cluster, and obtaining the character label corresponding to each type of voiceprint; obtaining key frame vectors and subtitle segment vectors based on video segments and subtitle segments; storing all information in a vector database; performing intent analysis on questions input by users based on video summaries, and calling the vector database for video retrieval and question answering.

The application belongs to the technical field of energy management, and particularly relates to an energy consumption monitoring and optimization control method and system of an energy management platform, which comprises the following steps: dividing a data area and a strategy area in a shared memory, establishing a lock-free ring buffer, a reliability bit mask and a state decision prefix tree; collecting processes write energy consumption data, system state parameters and time stamps into the lock-free ring buffer and set the reliability bit mask; an analysis process extracts the energy consumption data as energy consumption sampling values and calculates kurtosis values of the energy consumption sampling values, and encapsulates a composite data packet of the state vector and the kurtosis values; an optimization process traverses the state decision prefix tree to obtain a state duration expectation value, and obtains a predicted wake-up time; a control process scheduling priority is updated by using the kurtosis values, a kernel timer is configured to send a predicted control signal at the predicted wake-up time and perform an operation, and the application realizes active prediction wake-up of the control process, ensures real-time issuance of control instructions, and reduces system energy consumption.

The invention discloses an intelligent medical literature search question-answering system, which belongs to the technical field of artificial intelligence, and realizes accurate understanding of a user query intention, efficient retrieval of medical literatures, automatic summarization of literature contents and real-time generation of a dialogue outline through natural language processing and deep learning technologies. The system interacts with the user through a multi-round dialogue mechanism to gradually refine and confirm the requirements of the user and provide more accurate literature search results and professional suggestions.

The invention provides a construction process data retrieval method and system, and relates to the field of railway tunnel engineering management.The method comprises the steps that multi-source heterogeneous data related to the construction process is recognized based on a large model, and recognition content is output; performing entity extraction on the construction data, and outputting a space-time dynamic map of the construction process data according to a local relationship and a global relationship between entities; generating fusion data of the text information and the LaTeX formula sequence according to the technical specification knowledge; generating a business scene logic graph according to the business scene document; constructing a multimode information database according to the spatio-temporal dynamic map, the fusion data of the text information and the LaTeX formula sequence and the business scene logic map; and judging a service scene based on the trained expert model, selecting different mixed retrieval modes according to the service scene to carry out data search in the multimode information database, and outputting a retrieval result. According to the method, construction process data retrieval which is high in query speed, comprehensive in query result and high in credibility is realized.

The invention discloses a secure storage system for internet data. The system comprises a data acquisition module, a data layering module, an encryption and decryption module and a database, wherein the data acquisition module is used for acquiring internet data generated in each network platform, and classifying the internet data according to data types; the data layering module is used for distributing storage spaces for different types of Internet data and grouping and merging the same type of Internet data; the encryption and decryption module is used for distributing corresponding secret keys and encryption algorithms for different data types of internet data; and the database is used for receiving and storing the encrypted internet data. Through the scheme, invalid data can be released, and the security of the data information is improved while the data information is accessed and deleted.

The embodiment of the invention provides a multi-node database cluster parameter detection method and a related device, and relates to the technical field of databases. The method comprises the following steps: constructing a multi-node parameter value mapping table so as to identify an association relationship between nodes and parameters in the multi-node parameter value mapping table through a unique key; predefining a parameter checking rule set, wherein the parameter checking rule set comprises a parameter consistency requirement and a compliance threshold value; and based on the multi-node parameter value mapping table and the parameter checking rule set, performing comparative analysis on the parameter values of the nodes in the multi-node database cluster, and outputting abnormal information. Based on the method, the detection efficiency and accuracy of multi-node database cluster parameter detection can be improved.

The invention discloses a metadata-based functional prototype modeling and naming method and system, and belongs to the technical field of system modeling, and the method specifically comprises the following steps: establishing a metadatabase, defining a keyword rule, intelligently extracting keywords in a user search request by using a natural language processing and deep learning model, and searching a matching record in the metadatabase; if the matching item is not found, triggering an intelligent naming process, and generating a new name according to an automatic naming rule, a machine learning algorithm and diversity evaluation; the system stores new names and related information by adopting a distributed storage mode, and introduces a data deduplication and consistency check mechanism to ensure real-time updating and data accuracy of a metadatabase; and the modeling and naming efficiency and accuracy of the function prototype are improved.

The present invention discloses a micro-enterprise marketing auxiliary method and system based on a large language model, which uses a large language model to improve the efficiency and effectiveness of marketing. The method integrates a vector database, an embedding model and contextual learning technology to achieve real-time understanding of user intentions, personalized responses and the ability to quickly adapt to business changes. The core advantages of this method lie in its accurate user demand analysis, personalized service provision and efficient knowledge retrieval system. By analyzing user behavior and conversation records, the present invention can provide product recommendations and marketing strategies that meet user preferences. In addition, the present invention also has the ability to update real-time speech to ensure the timeliness and accuracy of information; the automated order information extraction technology improves information processing efficiency and reduces the time and cost of manual operations. The present invention not only improves marketing efficiency, but also enhances the interactive experience between users and customer service, providing companies with a competitive advantage in a highly competitive market.

The invention discloses a process knowledge question-answering method based on a domain knowledge graph, which comprises the following steps of: identifying entity information and relation information in a document by adopting a natural language processing technology, and constructing a structured process knowledge graph in a triple form; by training a BiLSTM + CRF model, accurate recognition of subject entities and relationships in user questions is realized, and a plurality of candidate answers are acquired from a knowledge graph in combination with a predefined structured Cyber query statement template library; finally, semantic matching is conducted on the candidate answers and the user questions through a RoBERTa model, and then answers most relevant to the questions are generated. According to the method, the entities and the relations in the question statement can be accurately recognized, matching of a grammar level is considered, and the method is deeper to a semantic level, so that efficient retrieval and accurate question answering of complex process knowledge can be supported.

The application relates to the technical field of radar signal sorting, and provides a jitter PRI sequence retrieval method and device based on a full convolution network. The method introduces a PP sequence as a supplementary feature of a DTOA sequence in pulse sequence retrieval, uses a trained jitter PRI sequence retrieval model based on a full convolution network to retrieve the DTOA sequence and the PP sequence, effectively retrieves the jitter PRI sequence without needing an a priori jitter upper bound, and significantly improves sequence retrieval performance.