1005 results about "Semantic matching" patented technology

The invention discloses a power plant operation and maintenance knowledge intelligent query method based on a large language model and an RAG technology. The method comprises the following steps: constructing a power plant operation and maintenance knowledge vector library covering structured, semi-structured and unstructured data; receiving a natural language question of a user, inputting an improved instruction to align a preprocessor, and generating a question semantic vector and an intention tag; relevant knowledge fragments are retrieved and sorted through a semantic matching retriever in combination with the intention labels; constructing a large language model cue word structure based on the retrieval result and the original question, generating candidate answers and recording a reference path; and finally, performing term specification and consistency verification according to the expert rule base, and outputting a structured and traceable final answer. According to the invention, the improved RAG technology is fused to realize intelligent query of the operation and maintenance knowledge of the power plant.

The invention provides a standard document automatic generation and multi-dimensional auditing method and system based on a large model, and relates to the technical field of artificial intelligence, and the method comprises the steps: 1, building a distributed database of a multi-source document, and analyzing a heterogeneous text through natural language processing to obtain a standardized knowledge network; step 2, extracting index elements based on the standardized knowledge network, and forming a structured parameter library through verification and verification; and step 3, based on the structured parameter library, constructing a template library, analyzing user demands in combination with semantic matching, and automatically generating a standard document outline. The document generation efficiency and quality are improved, the manual auditing cost is reduced, and the auditing comprehensiveness and accuracy are enhanced.

The invention provides a multi-agent collaborative question and answer enhancement method and system based on heterogeneous data knowledge, and belongs to the technical field of information management. The planning optimization intelligent agent carries out structured processing on the input complex natural language problem; a multi-modal retrieval mechanism on the heterogeneous knowledge source is constructed based on the problem disassembly and entity recognition result, and a multi-path recall agent passes through a semantic matching model of a double-tower structure; the abstract extraction agent performs semantic fusion and information extraction on the text segments and the knowledge graph sub-graphs output by the multi-path recall module; logic verification and quality evaluation are carried out on the answers generated by the abstract extraction module by the reflection iteration agent; and by setting a threshold mechanism and combining importance weights of the sub-questions, performing scoring and reflection optimization on the generated answers by utilizing a large language model LLM (Language Language Model). The method can effectively cope with cross-domain and multi-level complex question and answer tasks, and has good expandability and intelligent level.

The invention provides an enhanced generation method based on question matching retrieval, and belongs to the field of matching generation, and the method comprises the following steps: S1, a semantic feature coding stage: carrying out real-time feature extraction and vector space mapping on a natural language query input by a user by adopting a deep neural network model, generating high-dimensional distributed representation with semantic representation capability; s2, a knowledge base intelligent retrieval stage: executing multi-dimensional semantic matching in the vectorized knowledge base based on an approximate nearest neighbor search algorithm, and screening out a candidate knowledge set highly related to query semantics through a similarity measurement function; s3, retrieval matching results are automatically associated to the structured knowledge base through the established semantic-knowledge mapping relation, the preprocessed standardized response content is directly obtained, and the response content adopts a multi-modal data organization form and comprises a structured data entity and retains a rich text expression form.

The invention discloses a civil administration service question and answer method based on a large model and knowledge graph retrieval enhancement, and the method comprises the steps: S10, inputting a user question, and carrying out the question analysis and entity recognition; s20, the problem complexity is judged, if the problem is a single-hop problem, knowledge graph single-hop retrieval is carried out, and if the problem is a multi-hop problem, knowledge graph multi-hop retrieval is carried out; s30, performing semantic matching sorting to generate sub-answers; and S40, based on the sub-answers and the user question, performing synthesis to generate a final answer. According to the method, the structured knowledge of the knowledge graph and the natural language processing capability of the large language model are fused; a question decomposition module is used to enhance the interpretability of multi-hop information retrieval and answers; and using contextual learning (ICL) and thinking chain (CoT) prompts to generate an individually processed explicit inference chain to improve authenticity; the defects of traditional civil administration service questions and answers in the aspects of knowledge accuracy, reasoning ability and interpretability are overcome.

The invention discloses an intelligent contract risk recognition method and system, and relates to the technical field of text recognition, and the method comprises the steps: extracting a semantic vector of a to-be-recognized file; obtaining a first risk identification result based on the semantic vector and the review list; performing semantic matching on the semantic vector and the review knowledge graph to obtain a potential risk; obtaining a first risk category based on the potential risk and the review knowledge spectrogram, obtaining a preset risk judgment rule based on the first risk category, and obtaining a second risk category based on the preset risk judgment rule; obtaining a second risk identification result based on the contract category and the second risk category; constructing a clause rule base, and obtaining a compliance result based on the semantic vector and the clause rule base; obtaining a complete result based on the semantic vector and the standardized contract template library; and obtaining a total risk identification result based on the above identification result, thereby solving the problems of low risk identification efficiency and low accuracy caused by the fact that an existing contract term risk identification method depends on the determination of the license experience of professionals.

The invention discloses a search method and system based on computer natural language processing, and the method comprises the steps: extracting a synonym set and a context association relationship of keywords in a query text through a preset semantic knowledge graph, and generating a semantic vector representing a semantic dimension in combination with a deep learning model; extracting a historical behavior feature sequence from the query log based on the semantic vector, and analyzing a user search intention by adopting an attention mechanism model; performing similarity matching on the pre-constructed database by utilizing a semantic matching algorithm, and screening an information matching set meeting a threshold value; performing distributed processing on the matching set through a context-aware dynamic fragmentation algorithm, and constructing an index fragmentation cluster; semantic aggregation is realized by adopting a cross-fragment graph attention network, and an optimized search result set is generated through dynamic semantic projection. According to the method, through multi-modal fusion of the semantic knowledge graph and deep learning and in combination with a dynamic distributed processing architecture, the accuracy of search intention recognition and the efficiency of large-scale semantic matching are improved.

The invention relates to the technical field of intelligent retrieval, in particular to an enterprise RAG intelligent question answering system based on vector database reordering. The system specifically comprises: a document recall module, which retrieves a vector database to obtain candidate document blocks containing business metadata; the comprehensive scoring module is used for calculating a semantic correlation score by adopting a later-stage interaction architecture based on bidirectional token importance weighting, performing path semantic matching and context sensing rule evaluation according to a preset metadata ontology graph to obtain a service attribute score, and analyzing the evidence sub-graph to obtain a fact path score; fusing the semantic correlation score, the service attribute score and the fact path score to generate a comprehensive correlation score; and the sorting output module performs optimization resorting based on a preset punishment mechanism and the comprehensive correlation score to generate an optimized context set, and calls a generation model to output answers based on the optimized context set. According to the method, semantic accuracy, business compliance and fact reliability can be considered, and more trustworthy high-quality enterprise-level answers can be generated.

The invention relates to the technical field of artificial intelligence, and discloses an English teaching training system and method fusing semantic matching and cognitive assessment, and the method comprises the steps: synchronously capturing a text response, a voice intonation, an eye movement track, a facial micro-expression and a touch rhythm generated in a learning process; semantic deviation deconstruction and cognitive intention quantization processing are carried out on the learning interaction original sequence, and a bidirectional deep semantic matching network is adopted to carry out context alignment on student answers and target corpora; based on the word meaning divergence point set and the cognitive load multi-scale vector, extracting a nonlinear diffusion trajectory of a learning state by using a time gating multi-layer recursive trajectory evolution algorithm; the knowledge point nodes, the deviation type nodes and the emotion triggering nodes associated with the emotion instability candidate segments are fused to construct a local learning map; and forming an emotion cognition feedback result driven by learning interest based on the local learning map and the self-adaptive error correction intervention sequence. The method has the advantage of improving the learning interest of students.

The invention discloses a data import and field intelligent matching method for a low-code platform, and particularly relates to the technical field of low-code platform data processing. Constructing a field initial feature vector, extracting a standard field and historical matching features thereof from the target data model, and calculating a semantic matching score matrix by using a multi-source feature fusion model; screening the initial mapping table based on a dynamic threshold value, performing statistical deviation analysis in combination with a field value distribution difference measurement feature delta, and when delta exceeds a preset threshold value, calling a value mapping correction model to complete rematching to obtain an optimized mapping table; if it is detected that user history adjustment exists in the same source file type, re-correction is carried out based on a user behavior log, and a final field mapping table MP3 is output; finally, the MP3 is applied to data import, and automatic matching and binding are achieved. The method improves the accuracy and automation level of field matching, and is suitable for a multi-source heterogeneous data integration scene.

The invention discloses a multi-modal fused intelligent question answering and knowledge retrieval method and system, and the method comprises the steps: building a multi-modal data index model oriented to a heterogeneous knowledge source, carrying out the feature mapping of text, image, table, chart, audio and video contents through a unified semantic embedding space, and generating a cross-modal index set; after a query request is received, performing semantic matching and structure matching on the cross-modal index set by using a multi-channel retriever to obtain candidate evidence fragments; and based on an evidence granularity decomposition strategy, performing minimum evidence unit division on text statements, table units, chart data points and multimedia frame contents in the candidate evidence fragments, and establishing a semantic consistency graph among the units. According to the method, high-credibility traceable generation of question and answer results is realized through multi-modal fusion and space-time consistency constraint, and the retrieval precision and interpretation transparency in a complex knowledge scene are remarkably improved.

The invention discloses a retrieval generation method and system based on multi-agent collaboration, a terminal and a medium, and relates to the field of artificial intelligence. Performing semantic analysis on the input word embedding converted by the natural language query instruction through a query analysis agent, and determining a semantic intention vector; performing reinforcement learning and meta learning on the semantic intention vector through a strategy construction agent, and determining a retrieval strategy; performing semantic enhancement on the semantic intention vector according to knowledge graph node embedding to obtain a semantic enhancement vector; determining a data channel according to the semantic enhancement vector, a retrieval strategy and a real-time system load, and calling the data channel for retrieval to obtain candidate documents; and generating a target answer according to each candidate document based on an adaptive reflection feedback mechanism in combination with an auto-encoder and a generative adversarial network. The problems that the prior art depends on a fixed retrieval strategy, has limitation when facing complex query, multi-round interaction and cross-modal data fusion, is easily interfered by noise and is not accurate enough in semantic matching are effectively solved.

The invention discloses an automatic label labeling and classifying method and system oriented to unstructured system documents, and relates to the technical field of artificial intelligence. The method comprises the steps that semantic structure pre-analysis is conducted on an original system text, and a system semantic structure tree is constructed; establishing a system semantic enhancement vector space based on the semantic units and the logic relationship thereof; performing semantic deconstruction on the preset tag and extracting a feature vector; realizing cross-space semantic matching of the document and the tag through a system semantic attention mechanism; a confidence evaluation module is introduced to screen high-confidence labels from the three dimensions of structural integrity, coverage and logic consistency; and outputting a final label and a score through semantic conflict detection and resolution. According to the method, the problems that in the prior art, unstructured system text labeling accuracy is low and large-scale labeling samples are dependent on polysemy ambiguity, high context dependency, complex semantic structure and the like are solved, and labeling accuracy and robustness are remarkably improved.

The invention provides a patient information collection and medical record construction system and method based on multiple rounds of dialogues, and the system comprises an intelligent guide interaction module which receives the natural language input of a patient; the context-aware question and answer engine adopts a dialogue state representation method based on a graph structure to construct entities, relationships and attributes of each round of dialogue into knowledge sub-graphs; the medical record information dynamic builder monitors an updating event of the dialogue state diagram in real time; the abnormal information detection module adopts a mixed conflict detection method combining rules and learning; and the medical knowledge graph support system maps the oral expression of the patient to a standard medical term system in real time through a multi-level semantic matching strategy. According to the invention, the doctor does not need to distract the record in the inquiry process, and can pay more attention to patient observation and clinical thinking. The quality of the medical record first draft automatically generated by the system is high, a doctor only needs to perform a small amount of auditing and supplementing, and the medical record writing time is greatly shortened.

The invention relates to the technical field of big language model reasoning, and discloses a big language model key value cache multiplexing method and system based on fuzzy semantic matching, and the method comprises the steps: generating a key value cache according to lexical elements in a historical reasoning request of a user, gathering a plurality of lexical elements into lexical element blocks, generating embedded vectors of the lexical element blocks, and building a vector database; calculating a cosine similarity between an embedded vector of a lexical block of a new reasoning request and a historical embedded vector in a vector database, and if a historical lexical block of which the cosine similarity exceeds a set threshold exists, obtaining a corresponding key value cache through a Hash index and multiplexing the key value cache; calculating an attention score, and dividing the plurality of lexical elements in the current lexical element block into keyword elements and non-keyword elements based on the attention score; key value caches of the keyword elements are recalculated; and the re-calculated key value cache of the keyword elements and the reuse key value cache of the non-key sub-elements form a mixed key value cache. According to the method, on the premise that the model precision is almost not reduced, the key value cache multiplexing technology is expanded to fuzzy semantic matching from accurate matching, unnecessary calculation overhead is effectively reduced, and then the reasoning efficiency is improved.

The invention belongs to the field of artificial intelligence, and discloses a candidate question recommendation method for an intelligent dialogue system and a related device.Firstly, a deep learning model is adopted for conducting semantic coding and intention classification on an original question of a user, a semantic vector and an intention label are generated, and the semantic limitation of traditional keyword matching is broken through; screening the candidate question database by using the intention label to form a primary screening set, and narrowing the retrieval range; semantic matching of problem levels is achieved through semantic vector similarity calculation; and finally, performing dynamic weighted sorting by integrating multi-dimensional features such as semantic similarity, user portrait matching degree, question popularity and type adjustment factors to form a personalized recommendation list. By adopting the method, the accuracy of question recommendation and the scene generalization ability are effectively improved, so that the recommendation result not only meets the real-time semantic demand of the user, but also gives consideration to personalized preference and business scene characteristics.

The invention discloses an image diagnosis report generation method and device and a storage medium. The method comprises the following steps: determining the matching degree on three different levels, namely, the spatial matching degree between a diagnosis text and a candidate template on an anatomical part level, the semantic matching degree between the diagnosis text and the candidate template and related to pathological information, and the clinical association degree between the diagnosis text and the candidate template; according to the method, the target template matched with the diagnosis text is searched based on the diagnosis text input by the user, and the search is further optimized through the time decay factor, the clinical priority and the conflict degree among the regions of interest, so that the accuracy of searching the required image diagnosis template for the user is improved, and the user experience is improved. Therefore, the technical problems that in the prior art, a traditional keyword matching algorithm is poor in keyword matching effect in a medical scene, and the accuracy of searching out a correct template is reduced are solved.

The invention relates to the technical field of data analysis, provides an intelligent factory automatic monitoring method and system based on knowledge base enhancement, and realizes more accurate anomaly analysis and more effective process adjustment of an intelligent factory. The method comprises the steps of performing knowledge enhancement fusion processing on an obtained real-time monitoring data set of an intelligent factory through a pre-constructed process knowledge base and a pre-constructed monitoring rule base, and generating a process knowledge graph; performing abnormal mode recognition processing on the process knowledge graph based on a semantic matching strategy, extracting feature description of an abnormal event and a semantic association path with a historical monitoring text, and generating an abnormal mode analysis result containing abnormal root cause inference; according to the abnormal mode analysis result and the dynamic incidence relation in the process knowledge graph, an automatic monitoring report containing root cause priority ranking and optimization operation guidance is generated, and the automatic monitoring report is fed back to the intelligent factory control terminal to trigger process adjustment operation.

The invention relates to the technical field of data mining, in particular to an intelligent computer data analysis system based on artificial intelligence, which comprises a data deviation identification module, an attribution correction module, a tension correction module, a label frequency analysis module and a semantic deviation adjustment module. According to the method, by constructing the local window set and analyzing the difference fluctuation between the dimensions, the key dimension of the continuous deviation feature is accurately recognized, the situation that local anomaly disturbs the re-weighting of the high deviation dimension in classification judgment and affiliation evaluation is avoided, the stability of classification under label missing or affiliation fuzziness is improved, and the classification accuracy is improved. The dynamic modeling of the data track enhances the recognition and compensation capability of disturbance points and improves the classification continuity, the time sequence monitoring and fluctuation adjustment mechanism of the tag frequency enhances the consistency of tag expression, the semantic track offset response optimizes the semantic matching accuracy of category attribution, and the classification accuracy is improved. And constructing closed-loop linkage among data behaviors, classification stability and semantic adaptation.

The invention discloses a structured data retrieval system and method based on semantic matching and hierarchical indexing, and the related retrieval system comprises a first construction module which is used for extracting slice data in a preset vector library and meta-information corresponding to the slice data, and constructing a text node object containing an id; the second construction module is used for traversing a text node object to obtain meta-information subjected to hierarchical structure processing, and constructing a nested index tree; the directory decomposition module is used for receiving an input text, performing decomposition based on a hierarchical structure and generating a corresponding query vector; the retrieval module is used for performing semantic retrieval and hierarchical retrieval on the text in sequence to obtain a retrieval result; the grouping and sorting module is used for grouping the retrieval results according to the hit hierarchy, sorting the retrieval results in each group according to a descending order, and combining all groups to obtain a final retrieval result list; and the data backtracking module is used for acquiring an original text field from the vector database according to the id corresponding to the retrieval result.

The invention relates to the technical field of natural language processing, in particular to a context-based Prompt generative model optimization system, which comprises a context analysis module, a cue word generation module, a context optimization module, a semantic check module and a structure reconstruction module. According to the method, the context path and the semantic hierarchy information of the semantic unit are introduced, the fine degree of semantic matching degree recognition is improved, semantic guide deviation caused by statement template solidification is avoided, the cue words are recombined in combination with the semantic coherence weight and the logic dependency relationship, and the recognition accuracy is improved. The consistency and expression accuracy of the prompt content in the context are enhanced, the prompt word insertion sequence and connection mode are dynamically adjusted through a semantic conflict detection and structure rechecking mechanism, coherence and stability of a semantic structure and controllable generation of the prompt content are kept, semantic conflicts and expression chaos caused by static matching are effectively avoided in the generation process, and the generation efficiency is improved. And dynamic adaptation of prompt configuration and smooth optimization of language output are integrally realized.

The invention relates to the technical field of intelligent medical treatment and personalized nursing assistance, and particularly discloses a method and a system for generating a nursing assistance scheme for orthopedic patients. The method comprises the following steps: acquiring multi-modal data such as an electronic medical record, a medical image and rehabilitation monitoring, and constructing a semantic structure body; extracting a key rehabilitation state by using a graph neural network and time sequence modeling and generating a label graph; semantic matching and limiting condition extraction are carried out in combination with a nursing target library, and a target-constraint pair set is constructed; reasoning a nursing path based on a path generation model, and performing scoring modeling and weight updating in combination with execution feedback; and finally, cross-patient optimization and closed-loop semantic updating of the path model are realized. According to the method, the individuation and dynamic adaptive capacity of nursing path generation can be improved, and the intelligent level of nursing aid decision making is improved.

The invention provides an intelligent self-adaptive video editing method and system based on deep learning, and relates to the technical field of video processing.The method comprises the steps that firstly, a semantic mapping relation between a to-be-edited video material and a preset editing requirement is established, and an editing requirement mapping result is generated, the preset editing demand comprises a content style and a rhythm control demand, and then semantic feature association processing is carried out based on the mapping result to obtain a semantic association feature set comprising lens unit content semantic features and rhythm association features; then calling a pre-trained editing decision model (including a semantic matching module and a rhythm adjusting module) to carry out editing strategy matching on the set, generating a preliminary editing strategy set, generating an initial video editing scheme according to the preliminary editing strategy set, carrying out parameter adjustment on the initial scheme according to a strategy optimization suggestion output by the model, and carrying out video editing on the initial scheme; and a final video editing scheme is obtained, and intelligent self-adaptive editing of the video is realized.

The invention discloses an intelligent Bug management method and platform based on an AI large model. The method comprises the steps that structured error codes, unstructured log texts and development communication records are collected; standardized Bug features are obtained after preprocessing; semantic matching is performed based on a CodeBERT large model to generate a historical similar Bug case set, and a problem code segment is analyzed and positioned in combination with an AST abstract syntax tree; according to the project technology stack features, the historical similar Bug case set and the problem code segment, generating a repair scheme including code modification suggestions, automatically generated repair code snippets and implementation risk assessment; constructing and optimizing a triple knowledge graph; and finally, associating the standardized Bug features, the problem code segment, the repair scheme and the knowledge graph to obtain a Bug diagnosis result, and outputting the Bug diagnosis result. According to the method, intelligent positioning and accurate repairing of the Bug are achieved, the average Bug solving time is remarkably shortened, and meanwhile automatic precipitation of development experience is achieved through the continuously optimized knowledge graph.

The invention discloses a data processing method and system based on intelligent correction and electronic equipment, and relates to the technical field of educational informationization, and the method comprises the steps: analyzing a standard answer through natural language processing, extracting a standard knowledge point node and a connection edge, distributing a logic priority, calculating a node weight, and constructing a standard cognitive path map; analyzing student answers, mapping student knowledge point nodes through semantic matching, and constructing a student cognitive path map; comparing the two maps, and generating a score mark through a node matching state and sequence offset; detecting and complementing missing nodes and fracture paths, and constructing an atlas residual error scoring structure atlas; and calculating a structured score, and generating a report containing explanatory feedback. The system comprises five corresponding modules, and the equipment comprises a memory and a processor. The correction accuracy and interpretability are improved, the output report adapts to student feedback and teacher rechecking, and the method is suitable for education informatization automatic correction.

The invention provides a multi-modal large model incremental training data screening method, and relates to the technical field of data processing, and the method comprises the steps: executing modal structure analysis on newly added multi-modal data, extracting each modal vector, calculating a semantic matching degree, and removing samples lower than a preset first threshold value; calculating a multi-level semantic distance between a sample embedding vector and a historical clustering center in a unified semantic space, and dividing a core semantic region sample, a boundary semantic region sample and a discrete semantic region sample according to the change rate of the multi-level semantic distance; performing semantic fine-grained alignment on the boundary semantic region samples, when multimodal unstable distribution is detected, executing local context reconstruction to repair semantic deviation, and if the multimodal unstable distribution is still unstable, removing the semantic deviation; performing multiple rounds of small-batch reasoning, calculating a semantic stability coefficient based on a semantic prediction result, and when the semantic stability coefficient is lower than a preset second threshold value, determining that the sample is a potential drift sample and removing the potential drift sample; constructing an incremental training data set; according to the method, the autonomy and accuracy of incremental training data screening are improved.

The invention provides an interactive automatic explanation method for converting a traditional video into an artificial intelligence digital human, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining original video data and an audio track, and carrying out the semantic analysis of the audio track, and obtaining multi-mode deconstruction data; generating an explanation script for each time period of the video based on the explanation text, and performing timestamp labeling on the visual elements to form a time sequence synchronization data structure; in the playing process, a virtual image generator is driven to synthesize digital human dynamic expression output in real time according to the current playing time point; after a user interruption request is received, semantic matching is carried out on a query intention in the explanation script, a target explanation fragment and visual elements are positioned, and complementary explanation content is generated; and driving the virtual image generator to synthesize dynamic output synchronized with the supplementary explanation, and after interaction is completed, recovering playing or skipping to a specified time point according to a user instruction. According to the invention, the conversion from the traditional video to the interactive intelligent explanation video is realized, and the watching experience and learning efficiency of the user are improved.

The invention provides an mbse-based system full life cycle management method and system, and relates to the technical field of data synchronization, and the method comprises the steps: obtaining to-be-synchronized data of a source system, constructing a semantic graph structure, generating a mapping strategy based on implicit association between semantic matching learning systems, executing transactional data transmission, and constructing a time sequence traceability tree to track change propagation. And adaptive reconstruction is carried out when a constraint conflict is detected. According to the method, the problem of data synchronization among heterogeneous systems is solved, and the model consistency maintenance capability and the adaptive conflict processing efficiency are improved.

The invention relates to an information retrieval method and device for hierarchical planning reinforcement learning based on retrieval enhancement. The method comprises the following steps: acquiring a natural language query request of a user; calling a large language model of hierarchical planning reinforcement learning training based on retrieval enhancement to generate semantic keywords; the high-level strategy splits the natural language query request into a series of sub-query requests; the low-level strategy generates a retrieval query according to the context of the current sub-query request, and obtains a result verification condition from an external knowledge base through an RAG model; calling a search engine to execute multiple rounds of fine-grained information retrieval to obtain a plurality of candidate retrieval results; and performing deep semantic matching analysis on the candidate retrieval results and the result verification conditions by utilizing the post-trained large language model to calculate a matching degree score, and screening the candidate results according to the matching degree to obtain a final information retrieval result. Compared with the prior art, the method has the advantages that high-quality content retrieval and result display of complex query problems of users in various scenes can be realized.

The invention discloses a retrieval enhancement generation method and system based on hybrid retrieval and adaptive sorting, and relates to the technical field of artificial intelligence and natural language processing. Comprising the following steps: 1, analyzing a query language and providing multi-path retrieval: receiving a natural language query input by a user, and performing semantic analysis and structured processing; starting dense vector retrieval and sparse semantic retrieval in parallel, and respectively obtaining candidate document sets from the knowledge base; 2, candidate mixed result fusion is carried out, wherein duplicate removal and preliminary fusion are carried out on candidate documents obtained through dense retrieval and sparse retrieval, and a unified candidate document pool is formed; all the candidate documents are evaluated according to the query semantic matching degree, document authority and quality, context coherence and generation task type factors, sorting weights are dynamically generated based on all the factors, and a candidate document pool is resorted; and constructing a structured context prompt prompt, inputting the structured context prompt prompt into a pre-trained large model, and generating final response content.