104 results about "Semantic translation" patented technology

The invention discloses an intelligent building safety management system based on big data, and belongs to the technical field of intelligent buildings. Comprising the following modules: a multi-source data fusion and preprocessing module, which adopts intelligent data standardization and feature extraction technologies to realize efficient integration and semantic conversion of cross-source heterogeneous data and ensure data quality and consistency; the security risk dynamic knowledge graph construction module is used for constructing a semantic association network of building security elements through a graph neural network so as to continuously self-learn and dynamically update a risk feature relationship; the intelligent risk assessment and early warning module is used for carrying out comprehensive modeling, accurate positioning, layered assessment and intelligent early warning on building safety risks; the safety decision support module is used for providing an executable risk assessment report and an emergency decision suggestion; and the safety management visualization module adopts an interactive multi-dimensional visualization technology to intuitively present the dynamic evolution process of the building safety risk, so that the risk perception and management efficiency is remarkably improved.

The invention discloses an SQL (Structured Query Language) statement generation method and equipment based on a natural language and a medium, and the method comprises the following steps: constructing an RAG knowledge base and a data table knowledge graph of a project, then obtaining a to-be-converted natural language of the project, and performing keyword retrieval on the natural language through the RAG knowledge base to obtain an association table set; performing graph traversal on the natural language through the data table knowledge graph to obtain an associated node set; packaging the screening condition of the natural language to be converted, the association table set and the association node set to generate a context knowledge base of the natural language; and finally, according to the context knowledge base and the semantic conversion model, obtaining an SQL statement of a natural language. The problem of low efficiency of multi-table association path discovery is solved by constructing a dynamic extensible database relationship graph; a unified retrieval enhancement mechanism for the RAG knowledge base, the extensible knowledge graph and the historical query features is established, and the problem that target context retrieval is low in efficiency is solved.

The invention discloses a Tibetan language multi-dialect real-time semantic conversion method based on a cross-language BERT model, and the method comprises the following steps: S1, collecting original text corpora of each dialect of the Tibetan language, and constructing a standardized training corpus set; s2, performing parameter initialization on the mBERT model, and preliminarily training the mBERT model; s3, constructing a semantic modeling model, and performing fine adjustment on the semantic modeling model; s4, receiving to-be-converted text input, and encoding; s5, obtaining an intermediate semantic representation vector of the text through a semantic coding sub-module; s6, inputting the intermediate semantic representation vector into a semantic generation sub-module, and generating target text output; and S7, executing syntactic consistency correction and language fluency correction. According to the method, mBERT modeling and an adversarial optimization mechanism are fused, real-time semantic consistency conversion of multiple dialects of the Tibetan language is achieved, and the method has the advantages of being high in accuracy, high in robustness and low in response delay.

The invention relates to a multi-agent search enhancement generation method and system. The method comprises the following steps: analyzing a task instruction through a first agent to generate a search feature set; matching a plurality of corresponding second intelligent agents and performing directional semantic conversion on the original semantic vector to generate a query statement set; screening a plurality of third intelligent agents, and executing query operation through the third intelligent agents to obtain a preliminary retrieval set; inputting the preliminary retrieval set and the retrieval feature set into a fourth agent for verification, and generating an optimized retrieval result; in conclusion, according to the multi-agent search enhancement generation method and system provided by the invention, intelligent analysis, cross-modal semantic conversion, dynamic database partition scheduling and search result optimization verification of task instructions are realized through a multi-agent cooperation mechanism; the method has the effects of improving the retrieval precision and efficiency, enhancing the adaptability to complex query intentions and optimizing the collaborative retrieval capability of the heterogeneous database.

The invention discloses a touch feedback enhanced multi-mode robot grabbing control system and method. The touch feedback enhanced multi-mode robot grabbing control system comprises a touch feedback enhancement module, a visual touch language large model semantic generation module and a touch enhancement diffusion control optimization module. The tactile feedback enhancement module is based on a multi-agent cooperation system, performs semantic level and signal level processing on tactile signals, and realizes upper and lower layer coordination through a feedback routing mechanism. And the visual touch language large model module adopts a touch encoder based on CLIP to realize multi-mode semantic conversion and material attribute identification. And the tactile enhancement diffusion control module adopts a BRIDGeR style interpolation diffusion technology, and combines visual embedding and tactile constraint to realize high-frequency optimization of an action track. According to the method, the material identification accuracy, the precision operation success rate and the complex scene self-adaptive capability are remarkably improved.

A system (100) for adaptive data blending across heterogeneous BI platforms for real-time decision making, comprising: (a) a data source abstraction module configured to establish standardised and secure connections with several different Business Intelligence (BI) platforms and convert their native data structures into a unified intermediate schema; (b) a real-time data ingestion and synchronisation module operable to continuously monitor and retrieve data updates from the BI platforms using event-driven mechanisms and configurable synchronisation policies; c) an adaptive data mapping and transformation module designed to intelligently harmonize, adapt, and semantically translate disparate data formats using rule-based logic and AI-assisted schema mapping; (d) a semantic context and metadata management module configured to maintain a unified data dictionary, track data sequence and maintain semantic consistency across platforms; (e) a cross-platform query orchestration engine designed to decompose common user queries, translate them into platform-specific query languages, execute them across the relevant BI platforms, and aggregate the results into a coherent output; (f) a decision intelligence and recommendation module that analyses the mixed data to generate real-time insights, alerts and contextual recommendations for decision support; and (g) a security, governance and compliance module configured to enforce access controls, maintain audit trails and ensure compliance with data protection regulations; h) the system provides seamless, real-time and secure data blending to enable consistent analytics and decision-making across heterogeneous BI environments.

The invention discloses a financial knowledge graph construction method and system based on artificial intelligence, and relates to the field of artificial intelligence data processing. The method comprises the following steps: performing multi-dimensional semantic analysis on a heterogeneous financial data source, and extracting a structured semantic fragment; constructing a financial entity perception unit, identifying a multi-granularity entity and generating a unique code; generating a preliminary relation graph based on the event cascade relation and the attachment structure, and injecting a semantic translation label; normalizing the atlas relationship through semantic separation and a label reconstruction mechanism to form a financial relationship network with consistent semantics; executing evolution increment iteration in combination with the newly added corpus, and dynamically updating nodes and edge sets; and performing semantic consistency and structural integrity evaluation on an iteration result, and outputting a stable financial knowledge graph structural body. By introducing a multi-factor semantic analysis model, a causal relationship modeling mechanism and a graph evolution iteration strategy, systematic improvement of the financial knowledge graph in the aspects of structural expression precision, semantic reasoning ability and dynamic adaptability is achieved.

The invention discloses a territorial space planning method based on big data, and belongs to the technical field of territorial space planning. The method comprises the following steps of: 1, integrating territorial space comprehensive data and realizing standardized and normalized preprocessing and semantic association of the data; 2, precise expression and cross-scale conversion of spatial element semantics are achieved; 3, dynamically evaluating and quantifying the information distortion degree and uncertainty in the cross-scale semantic conversion process, and performing intelligent error correction; 4, realizing high-fidelity restoration and semantic consistency reconstruction of the information; 5, realizing an accurate and dynamic territorial space planning decision; 6, converting the cross-scale spatial semantic data into a visual and interactive visual form, and realizing intelligent presentation and interactive analysis of the spatial data; and 7, ensuring real-time performance, safety and openness of planning data, and realizing intelligent and dynamic management of territorial space planning.

The invention provides a Text-to-SQL (Structured Query Language) conversion method based on semantic modeling, which relates to the technical field of natural language processing and data analysis, and comprises the following steps: designing a structured semantic model comprising a fact table and a dimension table by adopting a dimension modeling method, and defining business field division, data model definition and model dictionary construction; performing word segmentation processing on the natural language question input by the user, matching the natural language question with dimensions, indexes and dimension values in the structured semantic model based on the model dictionary library, and generating Schema information containing database types, table names, field definitions and filtering conditions; calling the large model, and converting the natural language problem into a preliminary semantic SQL in combination with Schema information and a preset rule; calling the large model again, performing logic verification and grammar optimization on the preliminary semantic SQL, and generating a corrected semantic SQL; based on Schema information and data model definition, the corrected semantic SQL is converted into an executable database SQL statement.

The invention discloses an unmanned aerial vehicle rescue scheduling optimization method and device based on semantic translation modeling, and relates to the technical field of unmanned aerial vehicle emergency rescue intelligent scheduling, and the method comprises the steps: obtaining emergency rescue task description information; a natural language rule in the task description information is converted into a standardized constraint expression, and a response time target, a coverage range target, resource parameters and network structure parameters are recognized; constructing an initial scheduling model, and solving the initial scheduling model through an improved multi-objective optimization algorithm to obtain an initial scheduling scheme of the unmanned aerial vehicle; and when an interference event in the execution process of the initial scheduling scheme is detected, constructing a dynamic re-optimization model by taking the initial scheduling scheme as a disturbance reference, solving the dynamic re-optimization model by taking the scheme disturbance cost and the arrival late time as optimization targets, and outputting a repair scheduling scheme and an interpretable decision result. And the accuracy, the real-time performance and the decision credibility of emergency rescue scheduling are obviously improved.

The invention discloses a form content autonomous efficient filling method and device based on a multi-modal large model. The method comprises the following steps: receiving input data in a text, image, document or voice form; performing cross-modal information deep fusion and feature extraction on the input data through the multi-modal large model; a knowledge base and historical data are called, and form filling content is generated through intelligent reasoning and semantic conversion; and performing compliance decision judgment on the generated content, if the generated content is normal, finishing filling and recording a log, and if the generated content is illegal, finishing filling and recording the log after exception processing. Wherein cosine similarity calculation is adopted for cross-modal feature correlation analysis, weighted summation is adopted for feature fusion, corresponding formulas are adopted for semantic understanding result optimization and weight updating, and a scoring formula is adopted for compliance quantification. According to the method, multi-modal input can be efficiently processed, and the accuracy and compliance of form filling are improved.

The invention relates to the technical field of network deployment, in particular to a method and system for interactively configuring parameters through an LLM and a router gateway. The method comprises the following steps: arranging corresponding data of a natural language, a router command and a command execution result, and forming Quest / Answer training data required by a large model for providing a semantic mapping basis of a router configuration field for the large model; and training the general model through a large model fine tuning technology, so that the general model translates a router configuration related natural language into a specific router command and translates a command execution result into a natural language. According to the method, corresponding data pairs of a natural language, a router command and an execution result are constructed, and targeted fine adjustment is performed on a large model, so that the model has a bidirectional semantic translation capability in the field of router configuration, and the technical bottleneck that a traditional interaction scheme depends on independent interface development is broken through.

The invention discloses an inspection agent collaborative perception system based on semantic driving, and relates to the technical field of intelligent inspection, and the system comprises a prototype mapping module which collects inspection target category information and inspection target monitoring indexes, and generates an inspection task semantic prototype set and an inspection task semantic mapping table through semantic conversion; the semantic map module is used for collecting inspection image data and inspection video data through an inspection agent, performing pixel-level visual feature matching based on an inspection task semantic prototype set, and generating a semantic confidence map and a semantic request map; the coupling mutual sending module is used for executing sparse selection and directional mutual sending of semantic supply and demand coupling under the common constraint of the semantic confidence graph and the semantic request graph, and generating a multi-source sparse semantic feature queue; and the fusion remarking module is used for executing position-level semantic attention fusion and measurable sensitivity recalibration on the multi-source sparse semantic feature queue to generate a fusion probability graph and a fusion instance table.

The invention relates to a pet language translation method and system based on audio learning, and belongs to the technical field of animal training. The method comprises the following steps: constructing a pet standardized sound library; when the preset scene is triggered, playing the target sound signal; the target sound signal is a sound signal related to a preset scene in a pet standardized sound library; when it is detected that the first sound signal sent by the pet is matched with the sound signal in the pet standardized sound library, triggering a feedback operation corresponding to the matched sound signal; collecting pet sound in the current environment in real time, responding to the matching of the pet sound and the sound signal in the pet standardized sound library, and outputting a pet demand of the matched sound signal; the pet demand corresponds to a semantic translation result of the pet sound. By means of the mode, the training logic which can be stably recognized and can be copied and executed can be provided, accurate and reasonable pet language translation can be achieved, and the probability of mistranslation or wrong translation is reduced.

The invention relates to the technical field of visual reasoning, and provides a safety monitoring multi-modal model reasoning method and device. According to a user problem and a user image in a security monitoring scene, an object position in a visual scene is converted into text information, and the text information, the user problem and the corresponding user image serve as input information; obtaining visual features according to the input information through a cross-modal semantic converter; constructing a visual scene of the user image into hierarchical description comprising scene description and object description; modeling the context of the user image according to the user question, and generating a text prompt of a visual scene; reasoning is carried out through a large language model according to the visual features, the hierarchical description and the text prompt, reasoning output is obtained, and the problems that in a multi-modal scene, the complex scene perception ability is insufficient, and the large language model reasoning ability is insufficient in utilization in the prior art are solved.

The invention discloses an AI agent calling method for safe operation. The AI agent calling method comprises the following steps: 1) setting a corresponding intention category for each function of an AI agent; modeling the execution interface parameter of each function into an intention parameter; 2) taking each intention category and the corresponding intention parameter as a dictionary item in an intention dictionary to obtain an intention dictionary; 3) establishing mapping between the intention parameter and the execution interface of the function, wherein the mapping is used for converting the semantics of each dictionary item into the semantics of the corresponding execution interface; 4) the AI agent receives an instruction input by a user, analyzes an intention category corresponding to the instruction, and queries the intention dictionary according to the intention category of the instruction to obtain a matched dictionary item; and 5) converting the semantics of the matched dictionary item into the semantics of the corresponding execution interface by the AI agent, matching and calling the corresponding execution interface to execute the instruction according to the semantics of the execution interface, and returning the operation result of the instruction to the user.

The invention discloses a transformer area differentiated resource integrated communication method based on an intelligent fusion terminal, and the method comprises the following steps: S1, building a three-dimensional resource description model which comprises an equipment type, a communication protocol and data features, and carrying out the unified modeling of a photovoltaic inverter, an energy storage device and an intelligent electric meter; s2, dynamically calculating a communication priority weight according to the real-time state of the equipment and the operation requirement of the power grid; S3, selecting an optimal communication channel through a hybrid decision model; and S4, realizing protocol non-sensitive interaction by adopting an AST driving architecture which comprises a grammar analysis layer and a semantic conversion layer. According to the method, deep coupling of a communication technology and power system control is achieved, and normal form transformation from passive adaptation to active optimization is achieved through breakthrough of the three technologies. Firstly, the multi-dimensional cooperative communication capability is improved, a dynamic priority mechanism is introduced, a three-dimensional weight model is constructed based on an equipment type, a power grid state and a service demand, and accurate distribution of communication resources is realized.

The invention provides an integrated edge box with multi-protocol bidirectional conversion and intelligent reasoning, in the edge box, a multi-protocol access and bidirectional semantic conversion module analyzes protocol messages from different protocol devices through a plurality of pre-configured protocol adapters, and an analysis result is written into a unified data cache and message bus module; converting the content of the protocol message into a semantic object; according to a mapping rule defined by the semantic object, the semantic object is converted into a message format required by a target protocol, and semantic-driven cross-protocol bidirectional conversion is realized; the unified data caching and message bus module is responsible for data caching, format unification, inter-module communication and task scheduling; the edge AI reasoning and dynamic decision module obtains equipment data from the unified data cache and message bus module for AI reasoning; according to the scheme of the invention, heterogeneous equipment is enabled to cooperatively operate in a unified intelligent edge system, and cross-protocol semantic intercommunication and intelligent self-adaptive operation in a real sense are achieved.

The invention discloses a construction element detection method and a construction management method, and the method comprises the steps: obtaining construction data, building standard data and building log data of a target building, the construction data comprising at least one of text data, voice data, image data and video data; modal feature extraction is carried out on the construction data based on the building standard data, structural features of the target building are obtained, and the structural features comprise quality element features, safety element features, progress element features and cost element features; semantic conversion is conducted on the structural features according to the building log data, construction detection logs of the target building are obtained, and the construction detection logs comprise a quality log, a safety log, a progress log and a cost log; by automatically generating the classified construction detection log from the structured element features, the automatic and standardized output of the detection result is realized, and the efficiency and intelligent level of construction detection are greatly improved.

The invention provides a multi-word expression-oriented machine translation differential test method aiming at the problem of inaccurate multi-word expression semantic translation in a mainstream machine translation system. The method comprises the following steps that a word segmentation tool based on deep learning is adopted to divide words into vocabulary units, syntactic labels are distributed in combination with a pre-training sequence marking model, and a dependency analysis tool spaCy is utilized to mark the syntactic relation between the words; converting the tagged corpus into a standard CoNLL format, extracting a multi-word expression of a sentence through an automatic tool, and establishing a test data set of a sentence-level and phrase-level corresponding relationship; inputting the test set into a multi-translation system to generate a translation, and using an alignment tool AWESOME to accurately locate a corresponding relationship between a source language and a target language MWEs; the translation similarity is calculated based on BERTScore, mistranslation, translation omission and non-translation are recognized through an intra-group and inter-group dual check mechanism in combination with a dynamic threshold value, and evaluation of the translation accuracy of machine translation on multi-word expression is completed. According to the method provided by the invention, multi-word expression translation errors can be accurately recognized, and the accuracy of phrase-level semantic translation of a machine translation system is finely evaluated through a differential test method.

This disclosure relates to the field of data governance technology, and provides a method, system, apparatus, medium, and device for constructing data indicator services. The method includes: acquiring data source information and extracting indicator metadata and underlying basic data from it; inputting the indicator metadata into an indicator governance framework and the underlying basic data into an indicator calculation framework; performing semantic transformation on the indicator metadata in the indicator governance framework to obtain an indicator semantic knowledge base and generate indicator definition information; establishing a bidirectional communication channel between indicator governance and indicator calculation, and sending the indicator definition information to the indicator calculation framework through this channel; receiving indicator query requests; performing indicator calculations in the indicator calculation framework based on the request and the indicator definition information, combined with the underlying basic data, to generate indicator calculation results; outputting the calculation results to a user terminal and feeding back the calculation results to the indicator governance framework through the bidirectional channel. This embodiment improves the consistency and response efficiency of indicator services.

The invention provides an academic knowledge base construction method and system based on distributed crawlers and GraphRAG, and belongs to the technical field of computer information processing and knowledge management. The method comprises the following steps: constructing a distributed crawler architecture based on Selenium + Scrapy, completing character recognition and document logic structure reconstruction, storing a document as a Markdown format file, uploading the Markdown format file to a Dify knowledge base, segmenting the Markdown format file into knowledge blocks, and generating vector representation; extracting a semantic relationship between the entities, and constructing a knowledge graph according to the determined reasoning relationship; synchronously integrating the constructed knowledge graph and the segmented knowledge blocks into an intelligent agent, and constructing an intelligent agent process; by inputting a natural language question, an intelligent agent is triggered to execute semantic analysis, vector representation recall and knowledge graph reasoning, and an interpretable answer is returned. The problems that the data collection coverage rate is insufficient, text semantic conversion is distorted, deep knowledge association is missing, and knowledge base deployment is complex are solved.

The application discloses a video shadow detection method and device based on a dynamic prompt memory network, which can process confused shadows with the help of a projected shadow object, can aggregate time information without accumulating errors, and comprises a dynamic prompt module and a prompt-based memory module; the dynamic prompt module converts semantic information provided by a visual base model DINOv2 into a shadow mask when direct encoding of a shadow is difficult, local prompting focuses on semantic conversion and identifies a shadow position based on semantic information, and global prompting identifies a shadow boundary based on color and texture information; the memory module is used for solving deformation and long-term time consistency problems, preventing error accumulation, using local prompting as a time matching agent, reducing memory usage, and minimizing dependence on historical shadow masks; and the method can improve detection performance in a complex scene and maintain time consistency of results during long-term detection.

The application discloses a teaching aid robot multi-modal semantic cooperative scheduling method and system, and relates to the technical field of machine semantic translation processing. The teaching aid robot multi-modal semantic cooperative scheduling method and system comprises the following steps: S1, collecting and preprocessing the board area image and the teacher hand key point trajectory, and constructing a standardized teaching scene data set; S2, analyzing the semantic propagation convergence efficiency of the board area, and completing the sorting selection of the target board area; S3, evaluating the semantic binding stability degree, and adjusting the voice expression rhythm space pointing trajectory; S4, evaluating the space positioning execution effect and performing intensity evaluation, and correcting the voice emphasis time length and action dwell time. In the multi-role cooperative teaching scene, the existing robot cannot jointly analyze the teacher's ambiguous reference semantic and the real-time visual focus, gesture trajectory and historical semantic context, resulting in the drift between the semantic pointing and the actual board space position, and it is difficult to realize accurate semantic positioning.

The invention discloses a data conversion and integration method and system, and the method comprises the following steps: 1, extracting product research and development data from an IPD system through a standard interface, wherein the product research and development data comprises a product structure, technological process design and technological parameters; 2, mapping the extracted product research and development data to a unified data model, wherein the unified data model is subjected to hierarchical structured storage according to the product, the working procedure, the working step and the technological parameters; 3, based on a preset mapping rule and a semantic conversion function, converting the unified data model data into standard process flow data which can be identified by an MES system; and 4, synchronously pushing the converted data to an MES system through an adaptive interface, and recording a data version and an operation log. Through the implementation of the invention, the data model is unified, IPD and MES data semantics are unified, the problems of system isolation and data obstruction are solved, and the automatic conversion rate of research and development data is remarkably improved.

The invention discloses a document classification method, computer equipment and a storage medium. The document classification method comprises the following steps: receiving a target classification original document; specific information used for referring to a specific instance in the target classification original document is filtered out through a first generative model, common features of the category to which the target classification original document belongs are reserved, and standardized description of the target classification original document is generated; performing similarity retrieval in a rule database based on the standardized description, and recalling a plurality of candidate classification rules; and inputting the standardized description and the candidate classification rule into a second generative model to obtain a final classification result. According to the method, document expressions are aligned with classification rule expressions through semantic conversion, and the document classification accuracy is improved to a certain extent by adopting a mode of combining retrieval and reasoning.

The embodiment of the invention provides a semantic communication method and a related device, and the method comprises the steps: sending semantic conversion mode information to an intermediate device, the semantic conversion mode information being used for indicating a semantic conversion model and / or a semantic conversion knowledge base used by the intermediate device, the intermediate equipment is used for forwarding the coded data of the sending end to the receiving end; sending the semantic conversion mode information to the sending end; and sending the semantic conversion mode information to the receiving end. Therefore, under the condition that the semantic knowledge base or the semantic model between the sending end and the receiving end is not matched, the semantic communication quality between the sending end and the receiving end is improved through semantic conversion in the intermediate equipment.

The invention provides a man-machine interaction method and system based on cross-language bidirectional mapping. The method comprises the steps of obtaining first language original information data; analyzing the structured first language intention information to obtain a structured first language intention information set containing the first identification information; performing semantic conversion processing to generate second language expression data, establishing a cross-language bidirectional association relationship between the first identification information and a second language interface element identifier in the process, and generating an interactive second language information interface; receiving user interaction operation to generate behavior data; and according to the behavior data and the association relationship, directly obtaining the first identification information through query, and generating and outputting first language result information data. According to the invention, by establishing and querying the bidirectional association relationship, secondary translation of user operation is completely skipped, lossless and efficient closed-loop interaction from cross-language information identification to business instruction execution is realized, and the problems of insufficient interactivity and semantic distortion in the prior art are solved.

The application provides a silver government-enterprise multi-level permission control method and system, solves the problem of dynamic permission adaptation and control of heterogeneous systems by constructing an emergency permission object and a permission semantic adaptation gateway for semantic conversion and local execution feedback, can realize dynamic permission semantic adaptation and control of cross-institution heterogeneous systems in an emergency scene, and has the advantages of improving the flexibility and adaptability of permission changes.

The invention relates to the technical field of brain-computer interfaces, in particular to an imaginary voice classification method and system based on transfer learning, and the method comprises the steps: training a full-module model through health feature electroencephalogram signals, extracting spatial attention features through a perceptual coding module, capturing cross-brain region correlation features in combination with a sequence correlation module, and classifying the cross-brain region correlation features; semantic conversion from motion to language is realized; multi-modal noise is introduced through the injury simulation module, electroencephalogram signal characteristics in an injury scene are simulated, and the model in the first stage has higher robustness; by freezing the perceptual coding layer parameters of the first-stage model and only updating the subsequent module parameters, the obtained second-stage model improves the classification accuracy and generalization ability in the damage feature space; according to the method, the accuracy of the language intention recognition task is improved, and the method has higher individual adaptability and anti-noise capability.