50 results about "Data understanding" patented technology

The invention provides a cross-domain data integration and fusion method based on a large model, a terminal and a storage medium, and the method comprises the steps: carrying out the format conversion and cleaning of cross-domain original data through a unified data access adaptation layer, and outputting a standardized data set; carrying out semantic understanding integration by utilizing a large model, analyzing field names, generating field semantic tags in combination with a business term knowledge base, judging business relevance by calculating similarity, and generating an inter-domain field mapping relationship; and performing technical standard conversion and data fusion processing based on the mapping relationship, and performing intelligent quality treatment by adopting a mode of combining a quality treatment standard library and a large model. According to the method, a large model is introduced to assist data understanding and treatment, so that the technical problems of insufficient semantic understanding, non-uniform standards and the like in existing cross-domain data integration are effectively solved, and the accuracy and automation level of data integration are remarkably improved.

The invention relates to the technical field of electronic digital data processing, provides a financial multi-source data processing system and method, and aims to flexibly schedule tasks, dynamically adjust the acquisition frequency, ensure timely and accurate data acquisition and save resources in the aspect of data acquisition according to financial market rules, realize multi-module collaborative operation in the aspect of data processing and improve the data processing efficiency. The data quality is effectively improved from preliminary cleaning to standardized processing, a rule engine can update rules in real time and adapt to market and supervision changes, a semantic mapping module accurately processes financial terms and improves data understanding and integration, and an intelligent strategy adjustment module optimizes a processing strategy, reasonably allocates resources and improves processing efficiency. And the real-time monitoring and feedback module is convenient for management personnel to control the system state and timely adjust the strategy, an efficient and intelligent data processing scheme is integrally provided for the financial industry, and financial decision making and business development are assisted.

The invention provides a data backup method based on a large model, and the method comprises the steps: S1, building a data model, and carrying out the data understanding and analysis and model training and optimization; s2, making a backup strategy: performing intelligent analysis on a current to-be-backed-up data set by utilizing the trained large model; dynamically adjusting and optimizing a data backup strategy according to an analysis result of the large model, respectively adopting incremental backup or full backup or differential backup, and reasonably distributing backup resources and priorities according to importance and recovery requirements of data; s3, data backup execution: executing data backup operation according to the backup strategy formulated in S2; and S4, backup management and optimization: managing and optimizing the backup data regularly.

The invention discloses an agricultural decision reasoning large model based on a large language model and a construction method thereof. The method comprises the following steps: constructing a space-ground-air integrated multi-modal agricultural data acquisition and fusion module, integrating data of a ground sensor, unmanned aerial vehicle remote sensing, meteorological sensing and Internet of Things, solving the problem of data isomerism through reinforcement learning and a Transform architecture, and ensuring real-time integrity of the data. In combination with crop simulation model mechanisms such as DSSAT and the like, the inference rule base is dynamically updated to realize rapid iteration of agricultural knowledge; an agricultural feature extraction module and an adaptation layer are integrated, so that professional data understanding is enhanced; a multi-stage training strategy is adopted, and large-scale agricultural text fine tuning is carried out; reinforcement learning is introduced and expert feedback is combined to optimize decision generation, and a causal inference module is innovatively added to improve the inference precision. The model fuses the language ability of a large language model and agricultural knowledge, can accurately predict and diagnose agricultural problems and provide decision suggestions, and significantly improves the intelligence, precision and high efficiency level of agricultural production.

Disclosed is an explainable method for monitoring a state of a generator of a wind turbine generator system on the basis of a spatio-temporal graph. The method includes: S1: acquiring data collected by a supervisory control and data acquisition (SCADA) system; S2: carrying out data understanding on the SCADA data, selecting features associated with the generator, and carrying out data preparation on the selected feature data, and obtaining valid data; S3: embedding the SCADA data, and forming a directed spatio-temporal graph data sequence; and S4: carrying out modeling of a normal behavior model of the generator on the constructed directed spatio-temporal graph data sequence, computing a full-graph-level residual and a node-level residual, computing a residual through an exponentially weighted moving average (EWMA) control chart method, carrying out full-graph-level state monitoring on the generator, forming a fault information transmission chain relation, and enhancing explainability and robustness of a monitoring result.

The embodiment of the invention discloses a method and device for constructing a field consanguinity tree, a storage medium and a terminal. Firstly, a target field, a database query statement of the target field and upstream blood relationship information are obtained, the target field serves as a root node, a first executable calculation function packaging calculation logic of the target field is generated, and the calculation process is dominant and modularized. By analyzing input parameters of the function, an upstream field which directly depends on is automatically identified and determined to serve as a child node. And then recursively processing each sub-node, generating a corresponding calculation function, analyzing the dependency of the calculation function, tracing layer by layer until all leaf nodes (namely original bottom table fields), outputting a complete blood relationship tree containing all the nodes and the corresponding calculation function, and displaying the complete blood relationship tree through a visual interface. According to the method, automatic and accurate tracing from a target field to an original data source is realized, a computable and reusable blood relationship knowledge framework is constructed, and a solid foundation is provided for data understanding, problem investigation and intelligent data service.

The invention discloses an ultra-high-speed optical module digital manufacturing scene knowledge reasoning method, device and equipment based on a large model technology and a medium, and belongs to the technical field of artificial intelligence, and the method comprises the steps: constructing a knowledge graph of an ultra-high-speed optical module digital manufacturing scene; converting the multi-modal data into vector representation; a natural language input by a user is converted into a high-dimensional vector, and related knowledge points and reference data are matched; performing labeling and data enhancement on historical reasoning question and answer pairs and contextual dialogues, determining a contextual relationship by referring to a knowledge graph, and forming a fine tuning data set; performing fine tuning on the large language model by utilizing the fine tuning data set; and performing deep understanding on the matched related knowledge points and reference data by the fine-tuned large language model to generate a reasoning question and answer result. According to the method, professional knowledge and multi-modal data in the manufacturing process of the ultra-high-speed optical module are considered, the natural language processing technology and the multi-modal data understanding capacity are combined, and the understanding accuracy of the system on complex queries and the answering accuracy are improved.

The invention relates to the technical field of data integration and analysis, and discloses a digital twinning-oriented multi-dimensional data integration and analysis system, which comprises a data collection module, a data integration module, a data analysis module and a result output unit, in the digital twinning-oriented multi-dimensional data integration and analysis system, the same semantic elements in different data sources are accurately identified through a comprehensive algorithm in pattern matching, and the accuracy is improved; for data fusion, a weighted average method and a voting method are adopted for numeric data and classified data, reasonable fusion is achieved, and it is guaranteed that the result is reliable and effective; feature extraction can fully mine feature information from structured and unstructured data; according to attribute correlation analysis, the correlation relation between matching attributes is accurately grasped by means of a Pearson's correlation coefficient formula, and data understanding is deepened; results are output and displayed clearly, integration and analysis results can be presented visually, and subsequent decision-making and other operations are facilitated; the multi-dimensional data utilization efficiency and the analysis value in the digital twinning scene are effectively improved, and powerful data support is provided for development of related fields.

The embodiment of the invention provides a multi-modal data processing method and device, and according to a multi-modal data processing scheme provided under the technical concept of the specification, on a model architecture, each modal data related to a service problem can be subjected to discrete coding, embedded vectors of a plurality of tokens are obtained on each corresponding modal, and the embedded vectors of the plurality of tokens are subjected to data processing, so that the embedded vectors of the plurality of tokens are obtained. And outputting each representation vector formed by further coding the embedded vector of each token when the large language model generates the text as the input of a subsequent multi-modal generation module. Thus, multi-modal perception and generation can be integrated through a multi-modal large model, cross-modal data understanding and generation are achieved, and large language model implementation performance and more scene extension are facilitated.

A system and method for automating and improving tabular and list-based data extraction from a variety of document types is disclosed. The system and method detect and sort which documents include tables or lists, and performs row and column segmentation. In addition, the system and method apply Conditional Random Fields models to localize each table and semantic data understanding to map and export the extracted data to the desired format and arrangement.

A nuclear power scene-oriented multi-source heterogeneous data acquisition and processing system realizes thorough separation of equipment access and data understanding through a decoupling architecture of acquisition side plug-in and service side normalization, acquisition plug-in only transmits original data, and semantic analysis is centrally processed by a unified physical model engine; the problem of hard coding coupling of the plug-in and the business logic is avoided; by defining a standardized equipment abstract model and shielding bottom layer manufacturer, protocol and format differences, various kinds of heterogeneous data are standardized on the semantic level, unified modeling and semantic alignment of the multi-source heterogeneous data are achieved, data of various internet of things equipment sources are effectively integrated, and the data processing efficiency is improved. The problems of unclear data semantics and difficulty in unified use caused by diverse equipment manufacturers, different communication protocols and heterogeneous data formats are solved, cross-system linkage is promoted, semantic understanding, structured analysis and standardized presentation of cross-protocol and cross-platform equipment data are realized, and the safety, efficiency and management level of a nuclear power station are improved.

The invention relates to the technical field of Internet of Things data processing, and discloses a large-scale Internet of Things data processing method based on an object model and a rule engine, and the method comprises the following steps: equipment access and data collection: firstly, enabling a large number of Internet of Things equipment and sensors to access an edge gateway, and enabling the edge gateway to be responsible for the preprocessing of equipment data, equipment data formats are unified, then Internet of Things data are collected through Internet of Things message middleware, and large-scale data analysis is carried out subsequently based on an object model. According to the method, the object model is adopted as a unified data analysis standard, unstructured data generated by the Internet of Things is converted into meaningful structured data, semantic information is given to the data, the data understanding difficulty of a user is lowered, the data analysis speed is greatly increased, the data analysis difficulty and the equipment access threshold are lowered, and the user experience is improved. Therefore, the problems that the collection standards of heterogeneous devices of the Internet of Things are different, and collected data are difficult to analyze and standardize are solved.

The invention provides a shield construction site automatic layout method based on artificial intelligence. The method specifically comprises the following steps: S1, carrying out basic data acquisition and standardization processing on a construction site; s2, constructing a fine-grained field layout rule knowledge base; s3, inputting the basic data of the construction site after standardization processing, and based on the logic guidance and constraint of the fine-grained site layout rule knowledge base, intelligently generating functional region division, facility arrangement and traffic flow organization schemes of the shield construction site by utilizing an artificial intelligence large model to fuse data understanding and logic deduction; and S4, completely converting the site layout reasoning result in the step S3 from data to a graph, generating a construction site layout scheme, and visually displaying the construction site layout scheme. According to the method, standardized data input and knowledge-driven reasoning are constructed, and an artificial intelligence large model is combined, so that automation and standardization of functional area division, facility arrangement and traffic flow organization of the shield construction site are realized, and the design efficiency and consistency are improved.

The invention relates to the technical field of artificial intelligence, can be applied to business scenes of financial science and technology, medical health and the like, and discloses a data pipeline construction method, device and equipment and a medium, and the method comprises the steps: receiving a target data set and a task processing target, and generating a data general situation report through a data perception intelligent agent; generating and updating a solution strategy by utilizing a strategy planning agent based on the data general situation report, constructing an executable code script by utilizing a code generation and execution agent, and running in a sandbox environment to obtain a processing result; and executing grammar and logic verification by using a quality verification agent, submitting staged review after the verification is passed, and promoting the next stage to complete assembly line construction. According to the method, a closed-loop process of data understanding, strategy planning, code generation, execution and verification is realized through cooperation of multiple agents, and the automation degree and the processing reliability of data pipeline construction are improved.

The invention provides an intelligent fishpond oxygenation monitoring system based on the Internet of Things. The intelligent fishpond oxygenation monitoring system based on the Internet of Things comprises a power supply module, an on-site monitoring module, a cloud service module and a user terminal power supply module. According to the intelligent fishpond oxygenation monitoring system based on the Internet of Things, the abnormal response speed is increased, the breeding risk is reduced, the video analysis algorithm unit reduces the manual patrol cost and prevents the property safety risk in time, and the instruction forwarding unit guarantees information circulation of all modules of the system and supports remote control and alarm pushing; the data visualization interface of the user terminal reduces the data understanding threshold and improves the management efficiency, the alarm management unit shortens the abnormal response time and reduces the breeding and property loss, the equipment control unit gives consideration to the intellectualization and manual intervention requirements and improves the management flexibility, and the historical data query unit provides a data basis for breeding scheme optimization and risk pre-judgment. And fine breeding is assisted.

The invention relates to the technical field of data analysis, and discloses a sports data intelligent analysis method, which comprises the following steps of: 1, acquiring and preprocessing data; the method comprises the following steps: S1, collecting training data of athletes; s2, cleaning the data; s3, performing normalization processing on the data; 2, constructing and training a training strategy prediction model; the method comprises the following steps: S1, constructing a training strategy prediction model based on a Transform model; s2, introducing a BERT model to encode training strategy data; and S3, generating personalized training suggestions by using the generative pre-training model. According to the intelligent sports data analysis method, by introducing the Transform model, the BERT model and the generative pre-training model, complex time sequence characteristics and multi-dimensional relevance in training data can be more accurately captured, so that the accuracy of training strategy prediction is improved, long sequence data can be better processed through advanced model architectures, and the accuracy of training strategy prediction is improved. Semantic information in the training strategy is understood, and personalized training suggestions more fitting the actual conditions of the athletes are generated.

A system is provided.SOLUTION: A system comprising: means for accepting inputs from a user; means for parsing the user's inputs; means for hosting a generative AI model that understands context, tone, and intent based on the parsed inputs and generates customized content; means for returning the customized content generated by the generative AI model to the user; and means for storing the customized content to aid in future generations.SELECTED DRAWING: Figure 1

The invention relates to the technical field of target detection, and discloses a target detection system in an aerial scene of an unmanned aerial vehicle. According to the system, in a target image acquisition process, an advanced sensor carried by an unmanned aerial vehicle is combined to ensure that a high-quality aerial image is acquired, in a target image preprocessing stage, median filtering and image enhancement technologies are adopted to effectively improve the definition and availability of the image, and a target detection unit loads a high-efficiency detection algorithm to improve the detection accuracy of the target. The target can be accurately identified and a bounding box can be generated, and low-confidence detection can be removed based on a confidence threshold screening result, so that the accuracy and reliability of overall detection are improved, and the target tracking unit records the motion trail of the target in real time, so that a user can comprehensively know the dynamic state of the target, and the user experience is improved. And the display unit displays detection and tracking results in a visual manner, so that the operation experience and the data understanding ability of the user are improved, and the practical value and the application effect of the aerial photography data of the unmanned aerial vehicle are further improved.

The present application provides a code value unification method, device, computer equipment and medium. According to the method, the code value unification device can obtain one or more source tables, and check whether the code value and field corresponding to the data in the source table exist in the standard code value library. If the code value and field exist in the standard code value library, the first classification method and the second classification method are determined, and the code value in the source table is adjusted. Through the above method, data from multiple data sources can be integrated through this method, avoiding problems such as data processing errors and data comprehension errors caused by the inability to integrate data. Furthermore, the method can realize the unified code value standardization process without using complex code logic, thereby improving data processing efficiency. In addition, through this method, the impact of different data code values ​​caused by different classification methods can also be reduced.

The present invention discloses a natural language query method and system based on spatiotemporal knowledge cube, including constructing a spatiotemporal data cube based on a relational data table, constructing a virtual knowledge graph describing the spatiotemporal knowledge cube according to the spatiotemporal data cube, parsing the user's natural language, using the domain ontology in the spatiotemporal knowledge cube, splicing a prompting project of a large language model, inputting the prompting project into the large language model, and automatically generating GeoSPARQL from the large language model; executing GeoSPARQL in an ontology-based data access system (OBDA) to obtain corresponding data. The method proposed by the present invention is simple and easy to operate, improves the availability and value of data, and non-professional users can also use the method proposed by the present invention to achieve fast and accurate data retrieval, which is particularly suitable for fields that require high data interoperability and deep data understanding.

Most of the existing production applications in different domains are still running on. Mainframe applications in production receive data from various resources and process these data within. Understanding the structure of input data and output data is extremely important. A method and system for machine learning based understanding of a plurality of data elements in a mainframe program code has been provided. The method discloses a machine learning model that understands the structure of data elements in a Mainframe program code. The model considered is a graph neural network based architecture model. The disclosed method replicates memory mapping happening in the application program environment. The method understands the structure of the data element and the impact created by each data element on other data elements in the application and interfacing applications. The disclosed solution serves as a building block in problems such as code translation, reverse engineering etc.

The invention discloses a customer screening method and system, a product, equipment and a storage medium, an insurance company is helped to generate a clear customer group hypothesis through a large model based on the requirement of a specific insurance activity, namely, a feature set which should be possessed by a customer suitable for the activity; and performing approximation degree calculation on the customer portrait text generated by the multi-modal information, and finally, accurately screening out target customers meeting customer group characteristics, thereby improving the marketing accuracy. According to the method, a complex process which originally depends on manual analysis is converted into an efficient automatic process, and a full-process automatic closed loop from data understanding to decision output is constructed, so that the accuracy and flexibility of customer screening are remarkably improved on the whole, and the marketing preparation period is greatly shortened.

The invention discloses a cross-modal association identification system oriented to heterogeneous data understanding, and belongs to the field of data processing, the cross-modal association identification system oriented to heterogeneous data understanding comprises a first data acquisition module used for receiving an input specific application scene of cross-modal association identification and related heterogeneous modals, compared with the prior art, the method has the advantages that the data is collected and processed by determining the specific application scene of cross-modal association recognition and the related heterogeneous modals, cross-modal association is achieved by sharing distance measurement in the semantic space, the information retrieval efficiency is high, and the method is easy to implement. A designed modal conflict detection and analysis module can effectively distinguish a conflict scene, the situation that in a multi-modal information conflict scene, a standard cross-modal association process fails due to semantic consistency between default modals is avoided, and a designed semantic calibration module further analyzes the conflict scene and determines whether the conflict scene is reasonable or not.

The invention discloses an operation and maintenance system data semantic automatic identification and entity construction method based on a large language model, and the method comprises the following specific steps: 1, compiling an automatic script, and traversing all menus, pages and function points of a system through recursion; 2, loading a system page by using a front-end automation tool, and extracting a UI text related to a data field; 3, a prompt word is constructed, and a structured Prompt template is designed; 4, according to the nested structure of the interface response data, the constituent parts of the entity are preliminarily determined, and the user entity comprises the attributes of user Id and user Name; step 5, visualizing and outputting a semantic recognition result; and outputting a final result including a field list, name and semantic interpretation, an entity list and an entity relationship in a graphical form. Original manual empirical work is converted into an automatic process, the system data understanding efficiency is greatly improved, and a large amount of manpower and time cost are saved.

The invention relates to the technical field of artificial intelligence, in particular to a training big data spatio-temporal distribution visual analysis system and method, and the system comprises a training data preprocessing module, a training data range setting module, a spatio-temporal distribution visual display module and a training data traceability analysis module. Through key information extraction identification, data range setting according to needs, visual display combining time series and geographical distribution and double-layer traceability, data understanding is improved, a spatial-temporal distribution visual analysis means is provided for visually and efficiently analyzing and training big data, and the analysis efficiency of training big data can be effectively improved.

The embodiment of the invention discloses a data processing method and device, electronic equipment and a storage medium, and can solve the problems of insufficient cognitive dimension, insufficient data understanding depth and low intelligent level in current mine operation. In response to the target task, collecting multi-modal data through a multi-modal sensor; performing feature extraction on the multi-modal data to obtain target modal features; obtaining a target detection result corresponding to the target task according to the target task, the target modal feature, the target mining detection model and the target mining knowledge graph; the target mining detection model is a model obtained through multi-scene sample data training in advance, and the target mining knowledge graph is created according to an operation specification file; the target detection result at least comprises a risk detection result and a job adjustment suggestion; and outputting a target detection result corresponding to the target task.

The invention discloses an interactive selection and exploration method oriented to table data understanding in an immersive environment. The method comprises the steps that S1, a user uploads table data; s2, rendering an interactive three-dimensional space data table in the virtual reality space; s3, a user uses a controller or a naked hand gesture to carry out six table selection modes including click selection, open path selection and closed path selection on the three-dimensional space table data; s4, feeding back a selection operation of the user and activating a selection area expansion function; and S5, the user uses a controller or a naked hand gesture to adjust the boundary of the selection area and finally determines the boundary. According to the method, the table operation efficiency of the user in the virtual environment is greatly improved, particularly in the data analysis and visualization process, the user can interact with complex table data more naturally and flexibly, and the depth and breadth of information exploration and data insight are promoted.

The invention discloses a data interaction method and related products thereof, and the method comprises the steps: carrying out the interpretation of obtained source data, and generating basic interpretation data; understanding the basic interpretation data by using the data understanding large model, and generating application understanding data expressing potential contents of the basic interpretation data; and in response to a user demand, processing the application understanding data and / or the basic interpretation data by using the data processing logic generated by the demand analysis large model to generate a data processing result. According to the scheme, the basic interpretation data can be understood by utilizing the large language model, and the application understanding data containing the potential content of the data is generated. The understanding data is used as the application link layer to establish a deepened link between the basic data layer and the demand response layer, so that accurate response to the user demand is realized, and the effective processing and generation speed of the data for the user demand is improved.

The invention belongs to the technical field of financial risk control, and particularly relates to a data understanding and extracting method for a financial data report, which comprises the following steps: acquiring an original file of the financial data report; wherein the financial data report is stored as a PDF (Portable Document Format) file of an editable version; traversing each page of table data of the original file, and constructing dynamic mapping between a header and a corresponding cell based on deep learning; identifying a corresponding header according to the to-be-analyzed target, and collecting content information recorded by the related cells in a cross-page manner by utilizing a mapping relationship between the header and the corresponding cells; and splicing all the collected content information, and generating standardized structural data and analysis quality reports according to the spliced content information, so that the large model can call the structural data and the corresponding analysis quality reports to perform financial risk control analysis.

This invention discloses an interactive selection and exploration method for understanding tabular data in immersive environments, comprising: S1, the user uploads tabular data; S2, an interactive 3D spatial data table is rendered in a virtual reality space; S3, the user uses a controller or bare-hand gestures to perform six table selection methods on the 3D spatial table data, including point selection, open path selection, and closed path selection; S4, feedback is provided on the user's selection operation and the selection area expansion function is activated; S5, the user uses a controller or bare-hand gestures to adjust the boundary of the selection area and finally confirm it. This invention greatly improves the efficiency of user table operations in virtual environments, especially in data analysis and visualization processes, allowing users to interact with complex tabular data more naturally and flexibly, promoting the depth and breadth of information exploration and data insight.