50 results about "Semantic relevance" patented technology

The disclosure provides a relevance prediction model training method and device, and an abstract generation method and device. The relevance prediction model training method comprises: extracting a first semantic feature vector of a first sentence sample and a second semantic feature vector of a second sentence sample; generating a training sample, wherein the training sample comprises the first semantic feature vector and the second semantic feature vector, and a preset semantic relevance label of the first sentence sample and the second sentence sample; inputting the training sample into a machine learning model to obtain a semantic relevance prediction result of the first sentence sample and the second sentence sample; determining a loss function according to the semantic relevance label and the semantic relevance prediction result; and training the machine learning model by using the loss function to obtain a relevance prediction model.

This invention discloses a large-scale reliable fault diagnosis method for aviation maintenance scenarios, comprising: constructing a domain model for aviation troubleshooting tasks; standardizing and retrieving the input fault descriptions, normalizing and weighting the retrieval results to generate a unified-ranked candidate document list; selecting several high-scoring documents to perform dynamic context expansion and semantic relevance verification to obtain a structured evidence set; using this as input to the domain model for aviation troubleshooting tasks to output a troubleshooting solution; structurally decomposing the troubleshooting solution into several sub-steps and quantifying the uncertainty of each sub-step; combining fuzzy representation recognition and uncertainty classification to label the risk level of each sub-step, and forming a credible aviation maintenance diagnosis solution after manual review. This invention also discloses a large-scale reliable fault diagnosis system for aviation maintenance scenarios. This method and system can achieve stable generation, reliable verification, and risk-controlled output of aviation maintenance diagnostic content.

This invention discloses a resume screening and display method and system for human resource management. It constructs a keyword-dimension mapping table based on job description text using a natural language processing engine; analyzes keyword frequency, syntactic position, modification strength, and semantic relevance to generate a job requirement intensity vector; retrieves historical recruitment data based on job meta-information to generate a historical weight vector; adaptively calculates a smoothing coefficient based on the amount of historical data; and integrates dual-source weights to generate a dynamic weight vector; standardizes the scoring of each dimension of the resume, using a normalized formula based only on valid dimensions to calculate the matching degree, avoiding score distortion caused by missing fields; and finally displays the results sorted by matching degree. This invention abandons the preset weight mode, achieving dynamic screening with a tailored approach for each job, improving matching accuracy, operational efficiency, and the quality of human-machine collaborative decision-making, providing a scientific, efficient, and transparent intelligent solution for human resource management.

This invention discloses a method and system for coal mine safety risk assessment based on causal subgraph enhancement and LLM dynamic completion. The method includes: 1. Acquiring textual data in the field of coal mine safety and constructing a knowledge graph; 2. Extracting entities based on user queries and selecting initial triples in the knowledge graph through semantic relevance scores; 3. Expanding reachable triples using breadth-first search to construct an initial causal subgraph; 4. Performing causal reasoning and dynamic completion on the subgraph using a Large Language Model (LLM) to generate an enhanced complete causal subgraph; 5. Finally, generating an interpretable coal mine safety risk assessment report using LLM. This invention aims to solve the problems of missing causal logic and knowledge illusion in existing coal mine risk assessments by improving the accuracy of risk assessment through causal subgraph enhancement technology.

This invention discloses a method, device, and medium for querying engineering project data based on an AI large-scale model, relating to the field of data query technology. The method includes: calculating the business matching degree between the metadata of each data segment in the candidate dataset and the query intent object to obtain a business matching degree score; after fusing the semantic relevance score and the business matching score, re-ranking the candidate dataset, selecting the top K data segments, extracting engineering entities from the K data segments, linking them with nodes in the engineering project knowledge graph, extracting subgraphs associated with the query intent object, aggregating the K data segments according to the subgraphs, expanding them according to the result organization rules, and generating a query report. This invention enhances the professionalism and decision-making reference value of the query report by generating a set of associated evidence.

An electronic device, which determines semantic correlations between structured and unstructured data and data retrieval in response to natural language queries, may include circuits, such as a data-acquisition circuit, a data-division circuit, and other circuits. Together, these circuits may acquire tabular data organized into rows and a query. The circuits may also divide the tabular data, construct various graphs including the tabular data, and determine similarity coefficients for the tabular data and the query. Based on these graphs and similarity coefficients, the circuits may construct an expanded graph by modifying the modified graph based on the modified graph, as well as an expanded result graph based on relevance between the expanded graph and the query. After constructing the expanded graph and expanded result graph, the circuits may store or transmit one or more of the expanded graph and the expanded result.

This invention relates to the field of defect detection technology, specifically to a PCB defect detection method for industrial incremental scenarios. It includes a feature fusion method based on a lightweight segmentation strategy that combines random pruning with separate processing of strong and weak features. This strategy simplifies redundant computation by using a feature selector, and differentiates and recombines defect features of varying saliency during the feature fusion stage. This significantly improves inference speed while maintaining the precision of segmentation, solving the problems of high computational cost and difficulty in capturing minute defects in existing methods. The method also includes an incremental learning approach employing a background classifier adaptation mechanism and local semantic distillation. Class-specific regularization and spatially weighted logical alignment distillation work synergistically. By dynamically calibrating the background prediction logic and constructing a pixel-level semantic relevance matrix, it achieves deep alignment between new and old knowledge and background distribution. This effectively solves the catastrophic forgetting problem caused by the evolution of PCB background texture in existing incremental learning methods, significantly enhancing detection stability and the adaptability of the enhanced model.

The application discloses a voice question and answer method based on audio retrieval enhancement generation and double-layer rearrangement, which comprises the following steps: dividing a plurality of original documents contained in a knowledge base document according to a plurality of blocking strategies to obtain a plurality of types of document blocks divided based on different blocking strategies; converting a voice question input by a user end into a user question text and generating a user query vector based on the user question text; determining the relevance ranking results of the plurality of types of document blocks through a plurality of retrieval methods according to the user query vector; reordering the plurality of document blocks ranked higher than a preset ranking threshold in the relevance ranking results according to the semantic relevance between the user question text and the plurality of document blocks ranked higher than the preset ranking threshold, to obtain a reordering result; generating a question answer text according to the user question text and the corresponding document blocks in the reordering result, and generating a voice answer based on the question answer text and returning the voice answer to the user end.

This invention discloses a method and system for reviewing sensitive content in images, comprising: obtaining the original description of the image to be reviewed, the original description including a textual summary of the visual content of the image and the text content present in the image; extracting entity regions from the image to be reviewed, retrieving relevant external background information based on the image to be reviewed and the entity regions, and using the labels of the entity regions to exclude interference information to generate enhanced search results; performing semantic relevance filtering on the enhanced search results based on sensitive topics, and iteratively fusing the filtered enhanced search results into the original description to generate a final fused description; constructing a training dataset containing positive and negative judgment sample pairs based on the fused description, fine-tuning the visual language model, and using the fine-tuned visual language model to review sensitive content in new images. This invention can effectively improve the accuracy and reliability of sensitive content recognition in images.

The application discloses a large-scale data query method fusing deep semantic representation and explainable retrieval, and belongs to the technical field of information retrieval. The method comprises the following steps: adopting a hierarchical Transformer encoder to perform word-level, phrase-level and sentence-level multi-granularity semantic feature extraction on a query text, and dynamically fusing to generate multi-scale semantic embedding through a gating mechanism; constructing a semantic-symbol dual-channel layered index structure, and dynamically switching a retrieval channel according to a semantic correlation threshold; identifying a query type through an intention classifier and selecting an optimal retrieval strategy; extracting attention weights of each layer of the encoder to generate a semantic correlation strength matrix through cross-layer aggregation, and outputting an explainable retrieval result description; and adopting a write-once-copy and block incremental reconstruction strategy to realize incremental index maintenance. The application realizes efficient, accurate and explainable large-scale data query.

This invention relates to a retrieval enhancement generation method based on multi-level knowledge fusion, belonging to the field of natural language processing technology. It aims to solve two core problems in existing retrieval enhancement generation technologies: fragmented knowledge levels and interference from noise information. The invention includes the following steps: Graph index construction: Segmenting documents into fragments and extracting entities and relations to generate a knowledge graph with key-value pairs, followed by deduplication and optimization; Multi-level knowledge retrieval: Acquiring local and global knowledge separately, and constructing cross-level semantic associations by connecting knowledge modules through an intermediate layer; Dual-dimensional filtering: Filtering core knowledge based on a weighted score of semantic relevance and knowledge importance, combined with a maximum boundary relevance strategy to remove redundant noise; Response generation: Inputting the filtered knowledge into a large-scale language model to generate accurate and coherent responses. This invention significantly improves the generation quality of knowledge-intensive tasks through cross-level knowledge bridging and precise noise filtering, possessing significant theoretical and practical application value.

The application provides a security communication method for agent interaction, belongs to the technical field of artificial intelligence and network communication, and is used for solving the problems that the agent identity authentication is unreliable, vulnerable to prompt word injection and flooding attack in the related art. The method comprises the following steps: a verification party generates a semantic challenge and issues it to a request party; the request party generates a natural language semantic response by using a large language model, and embeds API information in the response in a semantic steganography manner; the verification party receives the semantic response and performs legality verification, which comprises whether the semantic correlation and the generation cost are higher than a threshold value; after the verification is passed, the API information is extracted and an invocation is performed, and a structured result is output. The application takes semantic capability as a trust root, realizes integrated identity authentication and secure API invocation through asymmetric cost semantic challenge and semantic steganography, is immune to injection attack from the protocol level, and naturally resists flooding attack.

This invention relates to the field of database construction technology, and discloses a knowledge base construction method and system based on AI text parsing and hybrid retrieval. By integrating semantic relevance and document layout features through adaptive block segmentation, it achieves accurate identification of core semantic units in the text and maintains logical integrity in the division, fundamentally avoiding information fragmentation and laying the foundation for high-quality knowledge organization. Secondly, through a hybrid retrieval mechanism combining keyword, semantic vector retrieval, and multi-dimensional re-ranking, it effectively balances retrieval response speed with the depth and accuracy of results, meeting users' multi-level query needs from rapid location to in-depth correlation mining. Finally, through dynamic updates and automatic association mapping functions, it achieves real-time synchronization and intelligent association of newly added knowledge, not only ensuring the timeliness of the knowledge system but also proactively building a cross-document knowledge network, thereby breaking down information silos and enhancing the overall utilization value and discovery capability of knowledge.

This application relates to the field of natural language processing technology, and in particular to a biomedical information extraction method based on a large language model. The method includes: acquiring the biomedical dataset to be processed and performing standardized preprocessing; converting the relation extraction data into high-dimensional vectors and constructing a local vector library; acquiring the medical information text to be processed as the query text, and performing a two-stage example retrieval and filtering in the local vector library to obtain a high-quality example set; generating context examples and performing context learning to understand the current task requirements and generate the information extraction results of the query text; and parsing the information extraction results. Based on the reordering capability of the cross-encoder model, this application designs a two-stage retrieval and reordering mechanism. By ensuring that the ICL examples provided to the large language model have both high semantic relevance and high task guidance, it significantly improves the accuracy and robustness of the model in biomedical named entity recognition and relation extraction tasks.

Systems and methods for adaptive transformer aided-semantic communication with multi-resolution encoding. The systems and methods include encoding (1002) patches of an image by flattening the patches to one-dimensional (1D) vectors to form encoded patches and determining an attention score of each of the patches using a vision transformer (ViT) and determining (1004) a semantic relevance of each of the patches to a user query using the respective attention score. The systems and methods further include adaptively transmitting (1010) the encoded patches with different resolutions based upon an amount of the semantic relevance.

The present application relates to the technical field of fraud short message interception, in particular to a kind of variant fraud short message interception method driven by deep learning.The present application first matches the word segmentation array based on the latest rule base, determines whether to intercept;Further, according to the semantic correlation between the word segmentation in the word segmentation array, the intention normality is obtained;Based on the intention normality, it is determined whether to intercept;Further, if the short message to be analyzed contains a link or a two-dimensional code, upload and simulate, statistical analysis and correlation analysis of the behavior of all newly created processes in the simulation environment, combined with the similarity between the webpage opened after the link or the two-dimensional code and the short message content, obtain the system-level behavior normal coefficient;Finally, the network-level behavior, memory-level behavior and system-level behavior normal coefficient are input into the pre-trained neural network to determine whether to intercept, which enhances the recognition ability of homophonic substitution, semantic reconstruction and other hidden techniques, improves the variant fraud short message interception accuracy and reduces the false positive rate.

This invention relates to the field of embodied intelligence and robotic task planning technology, and discloses a method for reordering agent actions based on dynamic semantic graphs. The method acquires natural language objectives, current environmental observations, dynamic semantic scene graphs, and historical interaction information to construct or update the dynamic semantic scene graph; it uses a language model to output candidate high-level actions and obtains semantic prior information; it filters executable candidate actions based on environmental constraints and node states; it calculates scoring signals such as semantic relevance score, exploration gain score, historical redundancy penalty, path cost penalty, and semantic prior score, comprehensively scores and reorders the candidate actions, selects the action with the highest score for execution, and updates the scene graph and historical interaction information based on feedback. This invention can reduce the instability of direct decision-making by the language model, reduce redundant exploration and ineffective interactions, and improve task execution efficiency and decision interpretability.

This application provides a dialogue method, device, storage medium, and program product. In this method, the current dialogue content between an intelligent customer service representative and a target user, as well as the target user's corresponding historical dialogue dataset, can be obtained. Target historical dialogue content with semantic relevance to the current dialogue content is selected from the historical dialogue dataset. Using a large language model, user preference information is inferred based on the target historical dialogue content and the current dialogue content to obtain current user preference information adapted to the current dialogue content. This current user preference information is used as the model interaction context for constructing the current dialogue, thereby using the model interaction context to drive the large language model to generate dialogue response content adapted to the current user preference information. The dialogue response content is then output to the target user to complete the current dialogue. This approach reduces context resource consumption and improves the large model's instruction compliance capability and the accuracy of the intelligent customer service representative's responses.

The application provides a memory operation processing method and system based on natural language instructions, which are applied to the memory management field of long-time task AI agents, and the method comprises the following steps: receiving a natural language instruction, converting the natural language instruction into a structured memory operation instruction according to an instruction format specification containing a priority adjustment operation and a structure adjustment operation; after the instruction is verified, parsed and adapted and translated, the back-end system is driven to execute; and the application further comprises automatically triggering a priority adjustment or a structure adjustment operation according to the semantic correlation of the instruction and a task target or the redundancy of a memory entry. Through the introduction of a standardized processing pipeline and an intelligent scheduling mechanism, the application solves the technical problems of memory priority drift and redundancy in long-time tasks, can improve operation consistency, significantly improves the performance of long-time tasks, and realizes cross-platform compatibility.

The application provides a visual positioning method, a visual positioning device and a computer storage medium. The visual positioning method comprises the following steps: associating an initial image feature and an initial text feature through mutual attention, to obtain a first image-text fusion feature; determining a text auxiliary attention score distribution according to the semantic correlation between the first image-text fusion feature and the initial image feature; selecting a visual feature according to the text auxiliary attention score distribution, to generate a first-stage query sequence feature; associating the first-stage query sequence feature and the initial text feature through mutual attention, to obtain a text-enhanced query sequence feature; associating the text-enhanced query sequence feature and the initial image feature through mutual attention, to obtain a second-stage query sequence feature; and inputting the second-stage query sequence feature and the initial image feature into a decoder, to generate prediction box information. Through the above method, the visual positioning accuracy of the query sequence is improved through two-stage enhanced query sequences.

This application provides a data retrieval method, apparatus, device, and storage medium for knowledge graphs, relating to the field of graph data retrieval technology. A data retrieval method for knowledge graphs includes: in response to a user's query, obtaining an initial candidate node set from a preset knowledge graph; calculating an initial semantic relevance score for each node in the initial candidate node set; constructing an initial weight distribution vector based on the initial semantic relevance score; performing a retrieval in the knowledge graph using a preset algorithm based on the initial weight distribution vector to obtain an iterative weight distribution for each node in the knowledge graph; and sorting each node in the knowledge graph according to the iterative weight distribution to obtain retrieval results corresponding to the query content. According to embodiments of this application, the accuracy and recall of graph retrieval in complex reasoning scenarios can be improved.

The present application belongs to the technical field of scientific and technological information retrieval, and particularly relates to a scientific and technological information retrieval method, system and device fusing a knowledge graph and semantic matching, aiming to solve the problems of low accuracy, poor relevance and lack of depth of retrieval results. The present application comprises: constructing a scientific and technological information knowledge graph containing entity nodes and relationship edges; generating a text-based semantic fingerprint and an adjacency subgraph-based structure fingerprint for each entity, and establishing a semantic-structure coupling index; performing semantic coding on a natural language query to obtain a query semantic vector, matching to obtain a query anchor point and determining a structure intention; performing multi-level path expansion in the knowledge graph according to the structure intention to form an expanded subgraph; calculating the semantic similarity between the nodes of the expanded subgraph and the query, selecting a core result and backtracking an associated path, and outputting a retrieval result graph. The present application takes into account both semantic relevance and graph structure constraints, and improves the retrieval accuracy, relevance and interpretability.

This invention discloses a method and system for correcting advertising text collocation errors in advertising design. The method includes the following steps: constructing a time-series corpus and generating high-dimensional semantic vectors for the text; calculating the dynamic semantic correlation between the advertising text to be examined and the corpus to assess potential risks; classifying risk levels based on the risk assessment results and dynamically generating corresponding elastic error correction rule sets; and finally performing hierarchical error correction processing on the advertising text based on the elastic error correction rule sets. This invention, by constructing a time-series corpus, calculating dynamic semantic correlations, and generating elastic error correction rule sets, achieves dynamic perception and accurate error correction of advertising text collocation risks, solving the problem of low accuracy in identifying advertising text collocation errors in existing technologies.

The application discloses a GPS trajectory travel mode recognition method based on semi-supervised double-map learning, first acquires GPS trajectory data and segments according to time sequence, extracts point features to obtain a plurality of trajectory segments; then constructs a double-map topological structure for each segment, including a sequence map representing local time continuity, and a learnable dependency graph that adaptively captures non-local semantic correlation; then, a self-supervised mask graph autoencoder pre-training is performed on the unlabeled segments to learn a dependency-aware spatiotemporal representation; finally, the pre-trained model is supervised fine-tuned with labeled segments, and the travel mode prediction result is output; the application can fully mine the structured semantic information of short trajectory segments without long sequence trajectories and massive labeled data, effectively overcoming the defects of the prior art that the recognition performance seriously decreases in the short sequence and sparse label scene, and significantly improving the robustness and generalization ability of the model.

This invention relates to the field of natural language processing technology, specifically to a method and system for intelligent identification of patient safety events based on transfer learning. The method includes: reconstructing multi-source heterogeneous medical records into a medical semantic flow graph with temporal continuity and semantic relevance; introducing risk semantic prototypes formed by real safety events in the source domain, correcting distribution offsets by joint maximum mean difference, and mapping target domain comments to the risk semantic space; identifying node conflicts through multi-dimensional semantic consistency analysis, constructing conflict evolution chains, and elevating safety event identification to structured path analysis; and introducing multi-source feedback information to back-map identification errors to the semantic structure and risk prototypes, achieving collaborative adaptive closed-loop optimization of each module of the system. This invention enables cross-domain transfer of risk knowledge, achieving interpretable intelligent identification of patient safety events.

Disclosed is an artificial intelligence-based matchmaking system (100) for suggesting compatible user profiles on a graphical user interface (GUI). The system (100) includes a user device (102) comprising an input unit adapted to receive one or more input data from a user and an output unit communicatively coupled with a server (104) through a communication network (106). The server (104) includes a processing unit configured to receive and store profile inputs comprising textual, categorical, or behavioral data such as occupation, interests, and intent parameters, process the profile inputs to generate a prioritized list of candidate profiles based on semantic correlation, contextual similarity, and inferred intent, display the candidate profiles sequentially on the GUI, receive gesture inputs indicating interest or skip actions and update a dynamic parameter-weighted preference model to adaptively reorder subsequent profiles in real-time. The present disclosure also relates to a method (200) for artificial intelligence-based matchmaking system (100).

This invention relates to the field of data processing technology, and more particularly to a standard content summarization method that integrates embedded vectors and semantic supervision. The method involves generating word vectors from each word and updating hidden layer variables, then calculating and generating word feature representation vectors. Next, word vectors are used to generate several feature map vectors, and the vectors with the largest response values ​​are concatenated to generate character feature representation vectors. The word feature representation vectors and character feature representation vectors are then concatenated to generate a concatenated feature vector, and a probability distribution is calculated. Based on the probability distribution, keywords are pre-classified and combined to generate combined phrases. The semantic similarity between the combined semantic vectors and the natural semantic vectors is calculated for text recombination. Based on semantic relevance, it is determined whether the pre-classified keywords are indeed keywords. Non-keywords in the text are further segmented into words, and the above steps are repeated. This invention improves the accuracy of the standard content summarization method that integrates embedded vectors and semantic supervision.

This invention discloses a multi-architecture hybrid method for recognizing fish feeding behavior, belonging to the field of aquaculture image processing technology. The method includes: acquiring video data of fish feeding processes; preprocessing the video data to obtain a spatiotemporal input tensor; using a multi-architecture hybrid backbone network to extract spatiotemporal features from the spatiotemporal input tensor to obtain an augmented tensor; mapping the augmented tensor to a classifier, outputting the probability distribution of each feeding behavior category, and using the category index corresponding to the highest probability as the recognition result. This invention achieves a unified expression of local burst features and long-range temporal dependencies in fish feeding behavior, reducing computational load while improving the ability to eliminate channel redundancy and model semantic relevance; it also balances computational overhead, enabling high-efficiency and high-precision recognition of fish feeding behavior.

The application provides a visual language large model attribution method based on semantic perception optimization, comprising the following steps: obtaining an input image, and obtaining a visual token set after model tokenization processing; calculating an attribution score of each visual token to a target text token, arranging the spatial position in the original image to form a single-scale attribution map; generating different resolution versions by performing multi-scale scaling on the input image, generating an aggregated attribution map after generating a single-scale attribution map for each version; calculating the output probability distribution and semantic correlation score of the pre-token and the target token, and generating an interference aggregated map by weighted aggregation of the aggregated attribution map of the pre-token; determining an adaptive suppression intensity by least squares optimization, subtracting the corresponding multiple of the interference aggregated map from the aggregated attribution map to obtain a final attribution map. The application solves the problems of insufficient spatial context capture and ineffective suppression of pre-token interference in the visual attribution method by generating an attribution map through decoding the visual token hidden state.

The application provides a teaching resource recommendation method and system based on natural language processing, relates to the technical field of natural language processing, and performs structural segmentation on text features of teaching text data to obtain a plurality of teaching segments; determines a semantic dependency tree of each teaching segment, determines an explicit dependency degree of teaching logic between the teaching segments according to all semantic dependency trees and semantic correlations between different teaching segments; performs implicit correlation analysis on teaching semantics between the teaching segments to obtain an implicit correlation degree, and determines implicit constraint conditions of the teaching logic between the teaching segments from all implicit correlation degrees; constructs a collaborative relationship graph between the teaching segments in a teaching resource allocation process based on the explicit dependency degree and the implicit constraint conditions of the teaching logic between the teaching segments, and adaptively allocates teaching resources according to the collaborative relationship graph. The above scheme can realize adaptive allocation of teaching resources based on the explicit dependency and the implicit constraint of the teaching logic between the teaching segments.