33 results about "Semantic expansion" patented technology

According to embodiments, described herein are systems and methods for providing extensibility in an analytics application environment, including enabling the use of custom semantic extensions to extend a semantic layer of a semantic data model (semantic model). According to embodiments, the system enables the use of a segmented query model - when customizing a semantic model, the system is able to dynamically incorporate changes from various increments at query time at runtime to dynamically surface appropriate data based on the extended semantic model.

The present application relates to the technical field of artificial intelligence and recommendation system, in particular to a kind of dynamic interaction perception and multilayer semantic compression generative recommendation method, comprising the following steps: obtaining the original attribute information of article and carrying out semantic extension, generating enhanced article description text;The historical interaction sequence of user is obtained, and the historical interaction sequence is carried out time feature extraction, multi-factor weighting and gate memory state update, to generate the dynamic preference vector of user;The historical interaction sequence is divided into multiple blocks, and the block-level abstract of each block is generated using a large language model and recursively compressed to obtain a global preference abstract;Question and answer samples are constructed, and a large language model is trained based on the question and answer samples to obtain the trained large language model.The present application improves the modeling capability of the recommendation system for user interest evolution through dynamic interaction perception and multilayer semantic compression, reduces long sequence input redundancy, improves the inference efficiency of the large model, and enhances the accuracy and interpretability of the recommendation results.

The application discloses a dynamic intelligent agent retrieval tree-based time-sensitive news retrieval method and system, an electronic device and a storage medium, and belongs to the technical field of information retrieval and artificial intelligence. The technical scheme of the application comprises the following steps: constructing a retrieval tree with a user query as a root node, the tree being expanded in semantics through multi-agent cooperation; obtaining a current news corpus and constructing a representative evaluation agent; based on the evaluation agent, dynamically selecting an optimal sub-tree from the retrieval tree that meets a preset condition; and using the optimal sub-tree to search the full news corpus to obtain a target news document. The application decouples high-cost semantic expansion and high-concurrency online retrieval, reduces query delay and computing cost, and dynamically selects an optimal sub-tree to adapt to changes in news content in real time, thereby improving the timeliness and relevance of the retrieval result.

This invention discloses a multimodal content generation system and method that combines GEO and SEO for dual-path collaborative optimization. This addresses the technical problem in existing technologies where information content is difficult for different types of search engines to accurately crawl, resulting in low-quality query results. The system includes: crawling raw information data and generating a preprocessed document; performing entity recognition and relation extraction on the standardized text contained within, and annotating it with structured data to generate a structured document and write it into an SEO article repository; performing semantic expansion and long-tail extension based on the structured document to generate a keyword matrix and rewriting the structured document in a question-and-answer format to obtain a GEO variant set; searching and matching credible standard data related to each text paragraph in the GEO variant set to generate an additional evidence chain summary; obtaining the target content, and adding adapted multimodal materials based on the semantics of the target content to generate a multimodal content body containing the additional evidence chain summary.

The application discloses a weakly supervised video anomaly detection method based on prompt learning knowledge enhancement, and belongs to the technical field of video intelligent analysis. A video side is given an abnormal scene video, video sequence features and audio sequence features are obtained through a feature extraction network, then a relatively complete feature aggregation network which has been trained is input to perform multi-modal feature aggregation, and an abnormal score is obtained through a score prediction network. In the text representation step based on prompt learning, an abnormal video label is used to construct a prompt template through a knowledge graph, normal labels are used to perform semantic expansion through a plurality of learnable parameters, and cross-modal alignment is performed with the video side, so that the video side features are close to different normal semantics. External information is introduced to perform knowledge enhancement, the video positive anomaly boundary is learned, the detection performance is improved, and finally multi-task joint optimization is performed through different loss functions to position an abnormal video segment.

This invention discloses an AI-based trusted data query device for production systems based on DLF (Data Flow Framework). Through a configuration management interface, data sources are selected, and the device automatically scans the selected data tables. Based on metadata, semantic expansion is performed, providing instant explanations for user queries. Based on an understanding of user intent and data structure, corresponding SQL query statements are generated. Natural language descriptions are generated based on the generated SQL, and the SQL statements undergo security verification, interpretation, and data execution, generating a result set. Each module generates corresponding OFD (Organizational Data Facility) electronic vouchers simultaneously during function execution, solidifying the module's input, output, metadata, snapshot, and operational behavior information. All OFD electronic vouchers are aggregated according to the chronological order and logical relationship of the operational behavior chain to form a DLF trusted electronic voucher set. This invention creatively combines AI data querying with trusted electronic vouchers to generate corresponding OFD electronic vouchers.

The application relates to the technical field of computer data processing, and discloses a search formula generation method and system based on instruction recombination and local feedback optimization, the search formula generation method comprising the following steps: step S1, dynamic assembly and semantic expansion of instruction segments; step S2, controlled term filtering based on a longest match mechanism; step S3, construction of a Boolean logic tree; and step S4, closed-loop logic repair based on component ID mapping. The application significantly reduces interaction overhead and computing resource consumption, solves the semantic drift problem of composite terms in a cross-language scenario, and realizes the compatibility of the probability of a generative model and the determinacy of a database syntax.

This invention discloses a method for constructing an emotional dictionary for urban planning across multiple stakeholders, relating to the fields of natural language processing and urban planning information technology. This method addresses the differences in planning commentary discourse among three stakeholders: officials, experts, and the general public. First, it collects multi-source corpora and performs preprocessing such as text segmentation, word segmentation, and part-of-speech tagging. Then, it uses the TF-IDF-POS algorithm with incorporating part-of-speech weights to select seed words, and combines this with an improved PMI algorithm with adaptive thresholds and scaling factors to mine new domain-specific words. Subsequently, it constructs a three-layer association of "text—topic—emotional words" through formal concept analysis to achieve semantic expansion of the dictionary. Finally, it uses SO-PMI to complete preliminary polarity labeling, fine-tunes the BERT model to achieve accurate sentiment classification, and generates a subject-specific emotional dictionary. This invention solves the problems of poor domain adaptability, insufficient subject differentiation, and low sentiment recognition accuracy of general emotional dictionaries. The constructed dictionary is highly targeted and semantically rich, achieving an accuracy rate of 88.46% in sentiment analysis of expert comments, effectively supporting precise decision-making and governance in urban planning.

This invention discloses a service recommendation method and system based on knowledge graphs, belonging to the field of knowledge graph service recommendation technology. The method includes: responding to the cold start state of a target new user, acquiring basic attribute data and constructing an initial user knowledge graph; semantically expanding through intent anchor diffusion to mine potential user needs; inputting the expanded graph into a preset service recommendation model to generate a service demand subgraph, and aligning it with the service domain knowledge graph through topology-preserving joint subgraph embedding, calculating joint loss to obtain highly matched recommended services; and optimizing model parameters using an asymmetric update strategy based on real-time user feedback. This invention accurately characterizes and deeply mines the potential needs of new users through user knowledge graphs, effectively solving the problems of low recommendation accuracy and result generalization caused by data scarcity in traditional methods during cold start scenarios, significantly improving the recommendation success rate and user experience for new users.

The application discloses a text2sql accuracy improvement method and system based on a multi-path retrieval mode link, belongs to the natural language processing and database query technical field, and comprises the following steps: collecting and structuring metadata to construct a data knowledge graph, constructing an NLP dictionary and a domain knowledge base and vectorizing storage; receiving a user question and utilizing a large model to combine domain knowledge to perform semantic expansion and ambiguity elimination; recalling metadata nodes from a vector library through three parallel channels of original input, rewritten questions and NLP word segmentation; grouping the recalled nodes, searching and constructing multiple "path subgraphs" representing potential correlation in the knowledge graph; introducing a dynamic weight adjustment mechanism to calculate the correlation score of each path subgraph with the user query and select the path subgraph with the highest score, extract the correlation contained in the path subgraph, and provide the accurate mode link result to a subsequent SQL generation module, so that the accuracy and efficiency of text to SQL conversion are effectively improved.

This invention relates to the field of semantic processing technology and discloses a high-dimensional vector retrieval and semantic generalization system with cross-modal semantic alignment. The system includes: an index building unit for calculating the density weighting factor of vector clusters and storing it in index nodes; a query positioning unit for determining whether a query vector falls into a target cluster or a sparse region; an adaptive metric calibration unit for generating nonlinear correction coefficients based on the density weighting factor to dynamically adjust the distance metric weights; and a semantic expansion retrieval unit for calculating the shortest path in sparse regions based on index connectivity to aggregate retrieval results. This invention introduces a density weighting factor as metadata into the index structure to achieve adaptive nonlinear calibration of the retrieval metric standard according to local data density, thus solving the problem of retrieval accuracy imbalance caused by uneven distribution of heterogeneous modal data.

This invention provides a class-based incremental learning method based on prompt guidance and multimodal fusion, relating to the fields of artificial intelligence and computer vision. First, the category labels are semantically expanded by constructing a semantically enhanced text representation using a text encoder. Then, a pre-trained visual encoder is used to perform block embedding and hierarchical feature extraction on the input image, constructing a cross-modal unified embedding space. A bimodal prompt gating fusion module is introduced within this unified embedding space, adaptively weighting text and image prompts according to gating weights to generate fused prompts. A bimodal prompt collaborative filtering module selects the set of prompts most relevant to the current task based on the similarity of image and text semantic features. In the incremental stage, the pre-trained backbone network is frozen, and only the prompt parameters and fusion layer weights are optimized. A joint loss function is used for parameter updates. Finally, in the inference stage, image and text data are input, cross-modal similarity is calculated, and classification prediction results are output.

The keyword matching method, device and electronic equipment across language environments provided by the embodiments of the present application relate to the technical field of information retrieval. First, a source language keyword used for matching a target language text is obtained; then, the source language keyword is processed by word segmentation, and the source language keyword is classified into a short keyword string or a long keyword string; then, when the source language keyword is a short keyword string, keyword cross-language matching is performed through semantic expansion to optimize the missing report problem of the source language keyword exact matching; when the language keyword is a long keyword string, the source language keyword cross-language matching is performed based on a semantic-level fuzzy matching technology, the overall matching degree of the keyword is calculated in combination with the semantic matching value and the overall relevance of the matching segment to the target text, so as to comprehensively consider the overall matching degree and the local matching degree of the source language keyword. The above scheme can adopt different matching strategies based on the classification of the source language keyword, and ensure the accuracy of the matching result.

The application belongs to the technical field of animation production and relates to a 3D dance generation method and system based on multi-modal large model audio semantic guidance, which sequentially converts original audio data into long text description, structured audio semantic information and audio semantic embedding vectors, realizes a two-stage audio semantic mining path, can capture deep semantic information of the audio data, and simultaneously, through semantic expansion of original style labels and conversion into style expansion text embedding vectors, combines with a frequency domain modulation mode to enhance original audio features, solves the problem that the prior art only uses a short string as a style label and does not sufficiently mine style fine-grained semantic information, realizes directional modulation and strengthening of style semantic information on the audio features in the frequency domain level, narrows the cross-modal gap between the audio signal and the semantic information, and improves the matching degree of the generated dance and music in deep connotation and style performance.

This application discloses an automated parsing method, system, device, medium, and product for ophthalmic biometric reports, relating to the field of ophthalmic artificial intelligence research. The method includes: converting the ophthalmic biometric report to be processed into a report image; partitioning the report image into a table grid and locating cells to obtain a set of cells; sorting all cells by rows and columns in a grid, and cropping each cell to generate a cell image; recognizing text using OCR and filling it into the corresponding positions to obtain cell text; subsequently performing structured preprocessing and semantic expansion on the cell text, completing field parsing and standardized mapping based on the preprocessed two-dimensional grid structure to generate an initial structured record; extracting values ​​from the intraocular lens calculation parameter area and writing them into the initial record to obtain the final structured record, and then writing it into an output file according to the target database template. This application can effectively improve data entry efficiency and reduce field mismatch and data cleaning costs.

The application relates to a knowledge graph enhanced multi-modal archive retrieval method and belongs to the field of artificial intelligence and multi-modal information retrieval. In order to solve the problems of multi-modal information semantic splitting, weak semantic reasoning capability and low semantic matching precision in the existing multi-modal archive resource retrieval process, through four stages of archive multi-modal data preprocessing and feature extraction, knowledge graph construction and enhancement, semantic retrieval request analysis and intention modeling, multi-modal semantic matching and sorting, the semantic relationship is enhanced by using the knowledge graph, the semantic correlation and context consistency of the retrieval result are significantly improved, the user can input the query in the form of text, image, voice and the like, and semantic extension retrieval is supported. The application can realize more efficient and accurate archive resource retrieval, and improve the user retrieval experience and archive knowledge utilization.

The application discloses a kind of education research topic and research gap mining method and device based on dynamic knowledge graph, its method includes: constructing and maintaining the dynamic knowledge graph of education research containing time stamp, dynamic weight and evolution record;According to the request parameter of user topic request formed by intelligent agent configuration file analysis and generated output structure constraint;Anchoring and semantic extension are carried out, and the local subgraph of controlled scale is extracted;Topological features such as density and connectivity are calculated to form topological evidence;Sparse area type gap and / or structure hole type gap are identified to generate candidate set and cause element;Potential is evaluated using time series heat prediction and / or relationship evolution prediction, and uncertainty is prompted when signal is insufficient;Comprehensive matching, gap, trend and feasibility score ranking are generated to generate candidate research direction priority list;The structured evidence chain corresponding to the sorting result is generated and output, and the feedback is updated graph weight / parameter to form a closed loop.

This invention discloses a multimodal retrieval method and system based on knowledge graphs. The method includes: identifying entity information in multimodal archive data to calculate cross-modal consistency scores; constructing semantic anchors based on the cross-modal consistency scores to generate knowledge events; acquiring several cross-modal verification information of the knowledge events to structurally optimize a preset knowledge graph, obtaining a target knowledge graph; determining the query intent based on query information and user historical query habits, and extracting semantic extension information from the target knowledge graph to disambiguate the query intent; determining several retrieval results based on the disambiguated query intent using the target knowledge graph, and analyzing the semantic path between the disambiguated query intent and each retrieval result; and calculating the relevance score of each retrieval result based on the semantic path for ranking. This invention can provide more in-depth retrieval results, improving the relevance and coverage of the retrieval results.

The application belongs to the technical field of knowledge product construction, and particularly relates to a knowledge service platform monograph construction method, which comprises the following steps: step one, document demand change sensing: analyzing user search logs of the knowledge service platform to sense changes in document demand of the user; step two, document demand content analysis: after judging that the user's document demand has changed, analyzing documents that have interacted with the knowledge service platform and the user within a demand change interval, mining core content of the user's document demand within the demand change interval, and providing topics and initial query words of the topics for monograph construction; step three, document query word expansion: performing semantic expansion on the initial query words of the topics to obtain an expanded query word sequence; and step four, monograph construction: performing semantic search on the knowledge service platform by using the multiple groups of expanded query word sequences of the topics to construct a monograph from obtained documents.

The present application relates to the technical field of information retrieval, in particular to a self-adaptive optimization intelligent material library vector matching method and system, which comprises receiving query information and generating query vector through a dynamic feature coding model; accessing an updatable semantic relation graph, retrieving and generating an extended query vector; performing joint vector retrieval on the query vector and the extended query vector to obtain a candidate material set; reordering the candidate set through a personalized dynamic reordering model and returning the result; and finally, based on user serialized behavior feedback data, synchronously optimizing the feature coding model, the semantic relation graph and the reordering model. The present application realizes deep semantic extension through the semantic relation graph, adopts a combination of joint retrieval and dynamic reordering, and realizes full-link self-adaptive optimization based on user feedback, thereby significantly improving the accuracy of material retrieval and user experience.

This invention relates to the field of scientific and technological achievement retrieval, and particularly to a method and system for intelligent retrieval and recommendation of architectural scientific and technological achievements based on a knowledge graph. The method includes the following steps: identifying multi-source achievement projects on an architectural scientific and technological achievement platform; constructing a hierarchical index for the multi-source achievement projects to build an index knowledge graph; performing value propagation prediction and scenario benefit comprehensive scoring on the multi-source achievement projects to obtain a total achievement benefit value; adjusting the priority storage of the index knowledge graph based on the total achievement benefit value to output a stored knowledge graph; identifying the original search command based on the architectural scientific and technological achievement platform, performing semantic expansion parsing and intent inference to obtain user search requirements; performing semantically guided retrieval on the stored knowledge graph based on user search requirements to obtain a candidate set of achievements; and adaptively recommending and ranking the candidate set of achievements to output a recommended achievement list. This invention improves the efficiency of architectural scientific and technological achievement retrieval and accurately matches achievement projects that meet user needs.

The application discloses an AI trusted data inquiry device based on a DLF production system, through a configuration management interface, data sources are selected, then the selected data table is automatically scanned, semantic expansion is carried out according to metadata, instant explanation is provided when a user inquires, corresponding SQL query statements are generated according to the understanding of the user intention and the data structure, natural language description is generated based on the generated SQL, the SQL statements are subjected to safety verification and explanation and data execution, and a result set is generated, when each module executes a function, corresponding OFD electronic vouchers are synchronously generated, input, output, metadata, snapshots and operation behavior information of the module are solidified, all OFD electronic vouchers are aggregated according to the order and logical relationship of the operation behavior chain to form a DLF trusted electronic voucher set, the application organically combines AI data inquiry and trusted electronic vouchers to generate corresponding OFD electronic vouchers.

This invention discloses an adaptive retrieval strategy optimization method and system based on artificial intelligence. The method includes acquiring user query statements, processing and filtering the corpus, matching with resource libraries to obtain query resource libraries, performing multi-perspective semantic expansion and hierarchical fusion to obtain three-level expanded queries, calculating the query-resource coupling degree of each query resource library and determining an initial retrieval strategy, executing the initial retrieval strategy to calculate the retrieval utility and cumulative regret value of each query resource library, determining an adaptive filtering threshold to divide positive feedback and negative feedback query resource libraries, performing characterization analysis to obtain feedback content and update the three-level expanded queries, iteratively updating the retrieval strategy for the negative feedback query resource library, and outputting an optimized retrieval strategy when no negative feedback query resource library exists. This method not only improves data retrieval efficiency and accuracy but also has good interpretability and can be directly applied to adaptive retrieval strategy optimization systems.

This invention belongs to the field of blockchain collaborative process modeling and automatic smart contract generation technology. It discloses a method and system for asset semantic extension modeling, smart contract generation and execution for cross-organizational process collaboration. The method introduces asset elements (Asset), asset tasks (Asset Task), and related relationships into the process model, explicitly expressing asset attributes, operation types, and invocation relationships. Assets are divided into four categories: homogeneous / non-homogeneous, distributed / value-added, and allowable operation and state rules are defined. The model is parsed into an intermediate model consisting of an element relationship diagram, a description set, and semantic objects. The method automatically generates a process contract (BPMN SC) (responsible for process advancement and task scheduling) and an asset contract (ERC SC) (responsible for state maintenance and rule execution). Identity and contract binding are completed during instance creation, and asset governance is executed at runtime through cross-contract calls to the ERC SC via the BPMN SC.

The application relates to the field of natural language processing, and provides a literature knowledge base self-evolution construction method and system based on a large model. The method comprises the following steps: performing semantic extension on a retrieval request input by a user to obtain a standardized query word set; performing retrieval and quality evaluation on a literature database through a double-channel retrieval mechanism according to the standardized query word set to obtain a candidate literature list; performing paragraph structure analysis and template-driven extraction on a plurality of literatures in the candidate literature list to obtain an original knowledge fragment set; performing term standardization and semantic fusion processing on the original knowledge fragment set to obtain unified knowledge representation; and performing incremental comparison between the unified knowledge representation and an existing knowledge graph to trigger conflict detection and version evolution management, and obtaining an updated knowledge graph. The application improves the cross-literature integration capability and improves the reliability of literature retrieval output.

This invention proposes an integrated method and device for the design and construction of deep foundation pit engineering. The invention includes: semantic expansion and multi-dimensional data improvement of the building information model (BIM) for deep foundation pit engineering, constructing an integrated model containing information related to engineering geometry, construction technology, and carbon emissions; decomposing the construction process into discrete work units and determining their time characteristic parameters; establishing a discrete event simulation logic model for construction resources, simulating multi-state dynamic switching and calculating dynamic energy consumption; integrating static and dynamic carbon emission data to complete the entire process of carbon emission accounting; and building an integrated interactive visualization platform to achieve real-time display of construction and carbon emission data, parameter adjustment, and closed-loop control of simulation feedback. This invention achieves refined and automated carbon footprint accounting for deep foundation pit engineering, improves design-construction collaboration and the scientific nature of low-carbon decision-making, and adapts to the green and low-carbon construction management needs of complex deep foundation pit engineering.

This invention discloses an AI-native semantic cognition method and system, belonging to the interdisciplinary field of artificial intelligence and system software. The method includes: a semantic extension step, which adds extended fields such as semantic target identifiers, data flow roles, and tensor network anchor point identifiers to the native data structures in the program runtime environment, forming a semantically enhanced extended structure; a holographic semantic graph construction step, which performs unified graph representation on all extended structures, constructing a holographic semantic graph that integrates capability dependencies, semantic relationships, and tensor network anchor point information into a single data structure for each node, enabling the complete semantic context to be directly obtained when querying any node without cross-graph indexing; and a deterministic rhythmic scheduling step, which, based on a preset number of attention domains 'a' and granularity levels 'b', deterministically calculates the cognitive scheduling index D(n) = ((n-1) mod a) + 1 and G(n) = ((n-1) mod b) + 1 corresponding to the current clock cycle 'n', to reproducibly allocate AI cognitive resources to the corresponding attention domains and granularity levels. The system includes a semantic extension structure module, a holographic semantic graph module, an AI common core module, a semantic call bridge module, and a ghost cache module. This invention enables AI to accurately perceive program semantics, auditable cognitive scheduling based on deterministic rhythms, and secure online evolution with cognitive self-protection capabilities. It solves the problems of semantic gaps between existing AI systems and operating environments, lack of evolutionary security boundaries, blind spots in attention allocation, and fragmented semantic graph representations.