139 results about "Semantic query" patented technology

The invention discloses an information retrieval system and method based on semantic normalization, and relates to the technical field of artificial intelligence information, and the method comprises the steps: collecting a semantic query record input by a user, carrying out the preliminary semantic analysis, and generating structured data; on the basis of the structured data, entity disambiguation is carried out by utilizing a knowledge graph, abstract classes are generated through a neural network, calibration and dynamic weight adjustment are carried out, and high-confidence entity abstract classes and confidence scores are generated; entity abstract classes and confidence scores are combined with user contexts, an action-value function is calculated through a value network, and an optimal action is selected by utilizing a-greedy algorithm; executing semantic normalization mapping according to the optimal action, and obtaining an intermediate expression by using a meta-symbol dynamic generator; and performing index retrieval and multi-dimensional sorting based on the intermediate expression to generate a sorted retrieval result list. According to the method, the semantic fragmentation problem of multi-modal query is solved, and deep semantic alignment and dynamic weight calibration of heterogeneous data are realized.

The invention belongs to the technical field of artificial intelligence and information retrieval, and provides a multi-modal retrieval method and system based on a lightweight knowledge graph and an index table, and the method comprises the steps: obtaining multi-modal source data, and extracting a structured semantic tag set; constructing a lightweight knowledge graph and a metadata index table; analyzing a natural language query input by a user to obtain a semantic query vector and a keyword set, executing semantic retrieval in the knowledge graph to obtain a text candidate result, and executing keyword matching in the metadata index table to obtain a non-text candidate result; for each non-text meta record, fusing the cross-modal similarity between the non-text meta record and the text candidate result, performing index matching on an original score and a graph semantic evidence score, calculating a comprehensive correlation score, and performing reordering; and generating a natural language answer containing a non-text record link according to a reordering result. The method is suitable for efficient cross-modal knowledge retrieval in a high-security and low-resource scene.

The invention discloses a water conservancy intelligent question-answering system and method based on knowledge enhancement and data driving, and aims at water conservancy business structured and unstructured data query, through loop optimization driven by positive and negative examples, precise classification of secondary intentions of water conservancy problems is realized, knowledge questions and answers, data query and professional water conservancy subclass questions and answers can be distinguished, and the system and the method can be applied to water conservancy business. And the semantic analysis efficiency is improved. Aiming at the problem of low query accuracy of retrieval enhancement generation in the field of water conservancy, a differential water conservancy knowledge base oriented to professional books, industrial standards and laws and regulations is constructed, local and networking information is processed through a multi-source knowledge fusion and conflict resolution mechanism, and the accuracy, interpretability and traceability of question and answer content are improved. Aiming at the problems of complex operation and low semantic query accuracy in query of massive water conservancy business data and monitoring data, multi-layer constraint Text-to-SQL conversion is performed based on water conservancy business knowledge, high-precision semantic query and automatic visual output of water conservancy structured data are realized, and query efficiency is improved.

The invention discloses a large language model semantic query acceleration method based on a sparse KV Cache index. The method comprises a KV Cache semantic pruning strategy based on an attention mechanism and a set of asynchronous pipeline reasoning architecture with overlapped calculation and I / O. According to the method, the attention sparsity characteristic of the large language model in the reasoning stage and the asynchronous transmission capacity between the Host memory and the GPU video memory are fully utilized, calculation redundancy and video memory occupation in repetitive semantic query are greatly reduced, and high-throughput, low-delay and high-performance batch semantic data processing service is provided. Through a mechanism for mapping a static text into a compressed semantic index and combining a prefix cache technology in a reasoning process to realize state multiplexing, high-performance reasoning acceleration and high-efficiency storage compression are provided for a data-intensive semantic analysis task in a resource-constrained environment.

The invention relates to the technical field of mine environment risks, and discloses a mine environment risk multi-modal analysis and early warning decision-making method, which comprises the following steps of: performing time and space reference alignment on multi-source observation data, constructing a unified spatio-temporal data set, and establishing a troposphere disturbance model to realize observation disturbance coupling. And constructing a physical evolution operator and a rainfall-driven external input operator based on a seepage mechanical relationship, and generating a state evolution mechanism. Semantic direction parameters are generated by mapping semantic query information to a target attention dimension, and a lagging propagation operator is constructed by setting a rainstorm ending moment as a starting point. And determining a non-regular spectrum transition index and a peak lag moment through the non-regularity measurement and the dynamic change rate of each lag propagation operator, and setting a reference threshold value based on the spectrum transition index of a historical non-risk time period. Finally, when the early warning intensity meets the condition, early warning is output, and a query result is generated in combination with the lagging propagation operator and the semantic direction parameters.

The invention discloses a meteorological intelligent question-answering system and method based on a domain knowledge graph and dynamic optimization, the system comprises a multi-source fusion training module, a meteorological knowledge graph and other modules, and the term analysis accuracy is improved through multi-source data training, knowledge graph supporting and dynamic optimization; according to the method, questions and answers are realized through multi-modal input analysis, meteorological element analysis and data feedback optimization. Compared with a traditional NLP model, the meteorological term analysis accuracy is improved from 68.2% to 92.7%, the composite query understanding bottleneck is broken through, three-layer nested semantic query is supported, the model new term adaptation period is shortened to be within 2 hours, the timeliness is improved by 10 times, meteorological element correlation analysis is achieved, and powerful support is provided for disaster reduction decision making.

The invention provides a service recommendation decision-making method based on large model retrieval enhancement generation and related equipment. The method comprises the following steps: acquiring an unstructured service demand; calling a large language model to perform semantic analysis and extension on the service demand, and generating an unstructured extended semantic text; encoding the extended semantic text to obtain a semantic query vector; querying a service description of each pre-stored service from a service database, respectively calculating a first similarity score between the semantic query vector and the service description, expanding a second similarity score between each keyword in the semantic text and the service description, and screening out a candidate service set having semantic association with the service demand; all corresponding application programming interfaces in the candidate service set are determined, and a final application programming interface set is screened out and recommended according to the use frequency of the various application programming interfaces, the semantic similarity among the labels and the semantic query vectors. According to the method and the device, the API recommendation accuracy is improved.

The invention discloses a video semantic question-answering system for a monitoring scene, and the system comprises a video collection module which is composed of a plurality of cameras and forms a monitoring network; the video preprocessing module is used for optimizing original video data through three-stage processing of key frame extraction, noise reduction and time alignment; the multi-modal feature extraction module comprises a video feature extraction channel and a text feature extraction channel; the multi-modal interactive reasoning module adopts a cross-modal neural network model; the answer generation module is used for outputting corresponding structured answer branches and natural language branches based on joint representation prediction answers; and an interaction and visualization interface module. The invention provides a video semantic question-answering system oriented to a monitoring scene, which supports a user to carry out semantic query on monitoring video contents in a natural language form, and the system automatically analyzes problems, analyzes video data and generates structured or natural language answers. And the combined understanding and response capability of open semantics, dynamic behaviors and attribute states is realized.

The invention relates to the field of intelligent evaluation, and particularly discloses an urban old city community toughness evaluation method and system, and the method comprises the steps: firstly obtaining the multi-modal data of the toughness of a to-be-evaluated old city community, extracting the multi-modal characterization characteristics of the data of the toughness of the to-be-evaluated old city community based on a deep learning technology, and meanwhile, carrying out the evaluation of the toughness of the to-be-evaluated old city community; the method comprises the following steps: firstly, extracting a preset number of multi-modal characterization characteristics of evaluated old city communities and corresponding toughness level labels from a background toughness database, so as to construct an association representation of the toughness data characteristics of the evaluated old city communities and an evaluation result; through rapid scanning semantic query of toughness data characteristics of the old city community to be evaluated and the evaluated old city community, toughness grade evaluation of the old city community to be evaluated is realized. Thus, through effective utilization of historical evaluation cases, the problem of insufficient similarity and difference analysis among different communities can be effectively solved, confusion among categories is reduced, and the consistency and accuracy of evaluation results are improved.

The invention discloses an intelligent decision data inquiry system and method based on a generative large model, and belongs to the technical field of artificial intelligence. Natural language inquiry input by a user is received, and an inquiry intention and a target field are extracted; extracting structured and unstructured cost data from the real estate project database, and constructing a candidate field set; performing semantic matching by utilizing the generative large model to generate an intention field mapping matrix; constructing a field dependency graph based on a mapping result, and extracting a minimum closed-loop field set; performing credibility evaluation on the fields, and constructing a weighted field set; generating a dynamic query template, outputting a natural language answer, and attaching a field tracing path; if the field is missing or the credibility is insufficient, automatically generating a reverse query until a credibility requirement is met; according to the method, automatic association and accurate question and answer between complex semantic query and heterogeneous cost data are realized, and the method has good interpretability, traceability and intelligent interaction capability.

The invention discloses a recruitment full-process automatic cooperation method and system based on a large language model, and relates to the technical field of artificial intelligence and human resource management, and the method specifically comprises the following steps: a recruitment platform receives recruitment position information issued by an enterprise recruiter, and generates a standardized recruitment position information description; performing text semantic analysis on the standardized recruitment post information description, and extracting a job key field of a recruitment post; extracting job seeker resume key fields from the job seeker resume; converting the job key field of the recruitment post into a high-dimensional semantic query vector, constructing a resume semantic vector library based on the resume key field of the job seeker, and performing query matching to generate a candidate resume list set; generating an interview comprehensive evaluation report based on the interview data; returning the interview comprehensive evaluation report to a training library of the large language model, and updating parameters of the large language model; and the enterprise recruiter determines an intention job seeker according to the interview comprehensive evaluation report, and generates an in-job notification and an in-job guide process.

The invention is suitable for the technical field of data processing, and provides a database query method, system and device based on a natural language and a medium. Based on the natural language text and the field library, calling a root matching strategy, a semantic matching strategy and a fuzzy matching strategy through a hierarchical collaboration mechanism to perform field mapping, and determining a target field; splicing the target fields to obtain an initial semantic query statement; loading and executing the plurality of modifiers in sequence through a service discovery mechanism to obtain a target semantic query statement; converting the target semantic query statement to obtain a target physical query statement; and executing the target physical query statement to obtain a target query result and returning the target query result to the query party. Through hierarchical cooperation of the root matching strategy, the semantic matching strategy and the fuzzy matching strategy, the expandability and adaptability of the matching strategy are improved, the field recognition accuracy is improved, the field matching range is expanded, and the matching efficiency is optimized.

The embodiment of the invention provides an index system creation method and device, equipment, a medium and a program product. The method comprises the steps that if a query request requesting for creating a target index system is detected, a large language model is controlled to convert the query request into a semantic query request for a knowledge graph, and the target index system is an index system used for evaluating a target object; the control chart retrieval enhancement model queries a target community hit by the query request in a target knowledge graph; the target knowledge graph is a knowledge graph created according to a triple extracted from target knowledge data, the triple comprises entities, attributes of the entities and relationships among the entities, the target knowledge data is knowledge data containing a target index, and the target index is an index used for evaluating a target object; and controlling the large language model to generate a target index system according to the searched triple in the target community and the searched abstract of the target community.

The invention belongs to the technical field of artificial intelligence, and discloses a semantic enhancement processing system based on a large model agent RAG database. The objective of the invention is to solve the problem of balance between creative thinking and factual basis retrieval in anti-factual reasoning. An anti-fact hypothesis condition of a user is obtained through a semantic query analysis module, semantic conflict intensity is accurately quantified through a knowledge retrieval and conflict recognition module, and a balance proportion of dynamic trade-off factor control creativity and factuality is constructed. A multilayer semantic alignment network is introduced to perform semantic reconstruction on fact knowledge, an enhanced knowledge fragment set compatible with anti-fact hypothesis is generated, and a reliable anti-fact reasoning path is constructed. And meanwhile, a feedback optimization mechanism is also provided, so that continuous learning and improvement can be realized from user interaction, and adaptive adjustment of parameters is realized. According to the method, the reasoning quality of a large model is improved, and high-quality contents which conform to the anti-fact premise and keep basic reasonability are generated.

The invention discloses a fine-grained unsupervised cross-modal pedestrian re-identification method based on large model semantic driving, and relates to a computer vision and mode identification technology. The method comprises the following steps: inputting an unlabeled visible light-infrared pedestrian data set, generating an image text description by using a vision-language model, and analyzing the image text description into a structured semantic attribute vector through a large language model; extracting visual features of the image and semantic query embedding corresponding to attributes, and generating fine-grained features of semantic alignment through a multi-head cross attention mechanism; fusing the visual similarity and the attribute similarity to generate a cross-modal pseudo tag; combining attribute-visual alignment loss and inter-attribute decoupling loss to optimize features; in the test stage, cross-modal matching retrieval is completed based on the optimized features. Through semantic analysis and enhancement, the problem that cross-modal feature alignment is insufficient in an unsupervised scene is solved, and experiments show that compared with a mainstream method, the model performance is improved, and the method can be applied to the fields of intelligent monitoring, cross-modal retrieval and the like.

The invention discloses a multi-modal assembly action recognition method for comparative semantic query, and relates to the technical field of man-machine cooperation assemblation.The method comprises the steps that a visual sensor is arranged on an assembly workbench to obtain an operator action video, a sampling frame sequence is obtained through random frame sampling, a skeleton sequence is obtained through human body posture estimation, and a skeleton sequence is obtained through human body posture estimation; and inputting an assembly action recognition model to complete recognition. The model comprises an image coding module, a skeleton coding module, a feature fusion module, a text coding module and a semantic comparison module which are used for extracting image and skeleton features, fusing features, coding preset category text description, comparing action features with category text features and outputting a result with the highest similarity, and a comparison loss function is adopted during training. According to the method, multi-modal information is fused, the problems of single-modal limitation and multi-modal semantic segmentation are solved, category text semantics are fully utilized, the fine-grained action recognition precision is improved, the over-fitting risk is reduced, and the generalization and task migration ability of the model in a dynamic industrial scene is enhanced.

A computer-implemented method for managing the lifecycle of a semantic index within a cloud-based environment is disclosed. The method involves detecting a signal indicating a tenant's eligibility for semantic indexing and, in response, identifying tenant-specific content for vectorization based on predefined criteria. Semantic vectors are generated from the identified content and stored in a primary index storage. These vectors are then propagated to a secondary index storage, where a semantic index is built from the propagated vectors. The method further includes enabling semantic queries based on the semantic index within the secondary index storage.

The invention provides a data management method and device based on a space-time hybrid index and a program product. The method comprises the steps of obtaining to-be-queried data; querying the double-layer index structure according to the to-be-queried data to obtain a query result; wherein the double-layer index structure comprises a first-layer index and a second-layer index, and the first-layer index is obtained by dividing a minimum containing area corresponding to each physics teaching space on the basis of a teaching behavior data set and constructing a concept hierarchy tree on the basis of each minimum containing area; the minimum containing area represents a minimum area which completely covers all activity ranges of the corresponding physics teaching space geometrically, the second-layer index is obtained on the basis of constructing a balance tree in each teaching unit, and the teaching units are used for representing the minimum containing area after the concept hierarchy tree is constructed in the first-layer index; and updating the double-layer index structure according to the teaching behavior data when determining that the to-be-queried data is the teaching behavior data based on the query result that the to-be-queried data is the non-queried data. The method can support complex space-time semantic query.

The invention provides a large model illusion detection method based on deep learning and semantic structure information, and aims to solve the problems that model imaginary information is difficult to judge and locate in a verifiable knowledge range and evidences are not traceable in an existing method. The method comprises the following steps: extracting standardized recognition features, retrieving candidate objects and forming a distinguishing condition set by minimum condition combination, generating a semantic query plan and recording ambiguity exclusion, retrieving a knowledge base and a document library and forming an evidence list, affiliating evidence according to distinguishing conditions, generating claim-level support and conflict and unknown judgment, and calculating an external consistency value on a target object. The illusion detection method has the advantages that the illusion detection accuracy and interpretability can be improved, and the illusion detection method is suitable for scenes such as model output auditing and fact verification.

The invention discloses an import and export commodity intelligent classification method based on knowledge graph metadata topology, and relates to the technical field of reinforcement learning, and the method comprises the steps: inputting an initial data packet into a dynamic interaction model, carrying out explicit association mining through a semantic enhancement layer, optimizing a rule matching path through a rule evolution layer, and constructing a dynamic commodity knowledge graph; performing topological structure derivation on the dynamic commodity knowledge graph to generate a graph topological analysis report and a metadata list, and performing knowledge reasoning integration on the graph topological analysis report and the metadata list to generate an intelligent navigation engine; calling an intelligent navigation engine to execute multi-path semantic query and rule verification on the dynamic knowledge graph to generate a candidate classification scheme set; and performing multi-target collaborative optimization on the candidate classification scheme set to generate a sorting scheme sequence, performing traceability packaging on the sorting scheme sequence, and outputting an intelligent classification scheme. According to the invention, through the dynamic interaction model and multi-target collaborative optimization, the rule adaptation efficiency in a complex scene is improved.

The invention discloses a refined target detection method based on superpixel guidance and double-flow feature fusion, and belongs to the field of computer vision. According to the method, a global semantic flow and super-pixel contour flow parallel architecture is constructed, high-level context features are extracted from the global semantic flow, super-pixel boundary significance of the super-pixel contour flow is modeled through a boundary feature attention graph convolutional network (BA-GCN), and fine contour features are obtained. A semantic query gating fusion (SQ-GF) module is innovatively designed to realize deep interaction of semantics and contour features, and a refined bounding box fitting a real contour of a target is output in combination with a dual-supervision detection head. The method solves the problems that a traditional rectangular frame is limited, fine-grained information is lost and feature fusion is insufficient, improves irregular target boundary positioning precision, and is suitable for high-precision scenes such as medical images and industrial defect detection.

The invention relates to the technical field of intelligent recommendation of design schemes, and particularly discloses an intelligent recommendation method and system for customized gift design schemes based on enterprise culture, and the method comprises the steps: carrying out the semantic analysis of gift customization demands of enterprises through employing a natural language processing technology based on deep learning; the method comprises the following steps of: extracting semantic feature representation of enterprise gift customization requirements, acquiring supplementary description information of enterprise culture from official websites and social media of enterprises, and performing rapid semantic query interaction fusion on the enterprise gift customization requirements and the supplementary description information of the enterprise culture; the semantic feature expression ability of enterprise gift customization requirements is enhanced, and intelligent screening and recommendation of alternative gifts are carried out on the basis of the semantic feature expression ability. According to the method and the device, accurate recommendation of customized gifts can be realized, the gifts are ensured to meet actual requirements of enterprises, and culture and value views of the enterprises can be effectively transmitted.

The invention provides an unstructured resource data searching method and a related device, and the method comprises the steps: based on a file type of unstructured resource data, adopting a workflow mode to extract basic metadata; based on the basic metadata, adopting a multi-modal large model to obtain corresponding extended metadata; on the basis of the extended metadata, adopting Neo4j to construct a knowledge graph; receiving an unstructured data retrieval request of a user, and performing semantic extraction by using the semantic large model to generate a semantic query vector; and on the basis of the semantic query vector, utilizing Elasticsearch to retrieve semantic similar contents in the knowledge graph so as to generate a recommendation result. According to the method, through cooperation of the multi-modal large model and the knowledge graph, different modal data are mapped to the unified semantic space, deep association of cross-modal data is realized, and the problem of cross-modal association fracture is solved.

The invention relates to a zero sample semantic segmentation method and device based on adaptive prompt and dynamic query. The method comprises the steps of obtaining image data and corresponding category semantic information, and performing preprocessing to obtain an input image and an input category semantic text; constructing a zero-sample semantic segmentation model based on adaptive prompt and dynamic query, wherein the model comprises a visual encoder, a text encoder, an adaptive visual prompt module, a dynamic semantic query module and an image segmentation prediction module; inputting the input image and the input category semantic text into the zero sample semantic segmentation model, designing a loss function combining segmentation supervision and different modal alignment constraints, and performing model training and optimization; and under conduction type or induction type zero sample setting, testing the zero sample semantic segmentation model by using an image to be processed, and outputting a corresponding segmentation result. And the zero sample semantic segmentation accuracy and robustness in a complex scene are remarkably improved.

The invention discloses an SAM2 small sample segmentation method based on semantic-visual dual-memory fusion, and the method comprises the steps: constructing semantic query memory, visual query memory and query-related support visual memory, and fusing the semantic query memory and the visual query memory through a memory refinement module guided by the query-related support visual memory. SAM2 dense matching and decoding module end-to-end training are combined. The problems of single memory and foreground-background confusion of an existing method are solved, target semantic consistency and fine-grained modeling ability are enhanced, segmentation precision and generalization ability are remarkably improved in a complex scene, and the method is suitable for the fields of medical image analysis, automatic driving and the like.

An exemplary system comprises a memory that stores and a processor that executes computer executable components stored in the memory, wherein the computer executable components comprise an obtaining component that intercepts a semantic source from being submitted to a retrieval augmented generation (RAG) architecture, and a transforming component that transforms the semantic source into a transformed source by identifying and converting prompt-misleading text of the semantic source into prompt-non-misleading text. In one or more embodiments, the semantic source is a semantic query having been submitted to the RAG architecture and / or a retrieved source having been retrieved by the RAG architecture in a process of providing a prompt. In one or more embodiments, the prompt-misleading text originated in connection with an origination of the semantic source and / or was caused by an adversarial attack corresponding to the semantic source.

The invention discloses a standard text semantic analysis and key clause extraction method and system, and the method comprises the steps: carrying out the multi-level semantic analysis of a standard text, recognizing a to-be-evaluated clause and a corresponding type in the standard text, extracting a key parameter entity of the to-be-evaluated clause, and constructing a clause-level semantic association structure; based on the semantic association structure, a standard term parameter knowledge graph is constructed, and the knowledge graph supports semantic query of technical parameters; and performing weight calculation and screening on the to-be-evaluated terms based on the standard term parameter mapping knowledge domain, and generating a structured standard requirement abstract list based on the screened key terms and associated parameter information in the corresponding mapping knowledge domain. In combination with multi-level semantic analysis and parameterized knowledge graph construction, automatic and structured extraction of standard text core requirements is realized, and the technical problems of low efficiency, incomplete key information coverage and lack of semantic association caused by dependence on manpower or shallow text processing are solved.

The invention relates to the technical field of computer vision and artificial intelligence, in particular to a cross-domain three-dimensional perception method combining geometric and semantic dual paths, which comprises the following steps of: 1, acquiring a multi-view image, and extracting a multi-scale two-dimensional feature map by using a two-dimensional feature encoder; 2, inputting the extracted two-dimensional feature map into a geometric generalization path, and carrying out the operation of self-supervised depth enhancement and explicit fusion to obtain geometric enhancement features; 3, inputting the extracted two-dimensional feature map into a semantic generalization path, and carrying out semantic query and domain confrontation fusion operation to obtain semantic enhancement features; and 4, finally fusing and decoding the geometric enhanced features and the semantic enhanced features to obtain a three-dimensional occupancy graph. The generalization performance of the method is improved in a breakthrough mode, and the method has the advantages that the practicability of the model is greatly enhanced, geometric perception is more accurate, three-dimensional reconstruction is more fidelity, semantic understanding is more consistent, and cross-domain recognition is more reliable.

This invention presents a face forgery detection method based on a query-driven forensic adapter, belonging to the fields of artificial intelligence and machine learning. It aims to address issues such as insufficient modeling of local forgery regions, limited interaction between semantic and visual features, and poor cross-distribution generalization performance. First, this invention proposes a dynamic semantic query module. This semantic query is injected as a dynamic attention signal into the multi-head attention mechanism of the CLIP encoder, effectively guiding CLIP to focus on potential forgery regions, improving its response to local anomalies such as boundary discontinuities and texture inconsistencies, and compensating for its insufficient local modeling capabilities without altering the CLIP's core structure. Second, this invention proposes a multi-scale feature aggregation mechanism. It proposes a semantically guided attention fusion module and a feature enhancement strategy based on linear interpolation. This method can effectively improve sample diversity and enhance the model's robustness and generalization ability across forgery methods and datasets.

The application is suitable for the technical field of data processing, and provides a database query method, system, device and medium based on natural language. The method comprises the following steps: acquiring a natural language text; based on the natural language text and a field library, calling a root matching strategy, a semantic matching strategy and a fuzzy matching strategy through a hierarchical cooperation mechanism to perform field mapping and determine a target field; splicing the target field to obtain an initial semantic query statement; loading and executing multiple correctors in sequence through a service discovery mechanism to obtain a target semantic query statement; converting the target semantic query statement to obtain a target physical query statement; executing the target physical query statement to obtain a target query result and return the query party. Through hierarchical cooperation of the root matching strategy, the semantic matching strategy and the fuzzy matching strategy, the application improves the scalability and adaptability of the matching strategy, improves the field recognition accuracy, expands the field matching range, and optimizes the matching efficiency.